GIT update of https://sourceware.org/git/glibc.git/release/2.28/master from glibc-2.28
GIT update of https://sourceware.org/git/glibc.git/release/2.28/master from glibc-2.28
Gbp-Pq: Name git-updates.diff
+2019-09-13 Wilco Dijkstra <wdijkstr@arm.com>
+
+ * string/memmem.c (__memmem): Rewrite to improve performance.
+
+2019-06-12 Wilco Dijkstra <wdijkstr@arm.com>
+
+ * string/str-two-way.h (two_way_short_needle): Add inline to avoid
+ warning.
+ (two_way_long_needle): Block inlining.
+ * string/strstr.c (strstr2): Add new function.
+ (strstr3): Likewise.
+ (STRSTR): Completely rewrite strstr to improve performance.
+
+2019-09-13 Rajalakshmi Srinivasaraghavan <raji@linux.vnet.ibm.com>
+
+ * string/memmem.c: Use memcmp for first match.
+
+2019-09-13 Wilco Dijkstra <wdijkstr@arm.com>
+
+ * string/strcasestr.c (STRCASESTR): Simplify and speedup first match.
+ * string/strstr.c (AVAILABLE): Likewise.
+
+2019-09-06 Wilco Dijkstra <wdijkstr@arm.com>
+
+ * manual/tunables.texi (glibc.cpu.name): Add ares tunable.
+ * sysdeps/aarch64/multiarch/memcpy.c (__libc_memcpy): Use
+ __memcpy_falkor for ares.
+ * sysdeps/unix/sysv/linux/aarch64/cpu-features.h (IS_ARES):
+ Add new define.
+ * sysdeps/unix/sysv/linux/aarch64/cpu-features.c (cpu_list):
+ Add ares cpu.
+
+2019-07-12 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+
+ [BZ #24699]
+ * posix/tst-mmap-offset.c: Mention BZ #24699.
+ (do_test_bz21270): Rename to do_test_large_offset and use
+ mmap64_maximum_offset to check for maximum expected offset value.
+ * sysdeps/generic/mmap_info.h: New file.
+ * sysdeps/unix/sysv/linux/mips/mmap_info.h: Likewise.
+ * sysdeps/unix/sysv/linux/mmap64.c (MMAP_OFF_HIGH_MASK): Define iff
+ __NR_mmap2 is used.
+
+2019-07-12 Szabolcs Nagy <szabolcs.nagy@arm.com>
+
+ * sysdeps/aarch64/dl-machine.h (elf_machine_lazy_rel): Check
+ STO_AARCH64_VARIANT_PCS and bind such symbols at load time.
+
+2019-06-13 Szabolcs Nagy <szabolcs.nagy@arm.com>
+
+ * elf/elf.h (STO_AARCH64_VARIANT_PCS): Define.
+ (DT_AARCH64_VARIANT_PCS): Define.
+
+2019-06-28 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #24744]
+ io: Remove the copy_file_range emulation.
+ * sysdeps/unix/sysv/linux/copy_file_range.c (copy_file_range): Do
+ not define and call copy_file_range_compat.
+ * io/Makefile (tests-static, tests-internal): Do not add
+ tst-copy_file_range-compat.
+ * io/copy_file_range-compat.c: Remove file.
+ * io/copy_file_range.c (copy_file_range): Define as stub.
+ * io/tst-copy_file_range-compat.c: Remove file.
+ * io/tst-copy_file_range.c (xdevfile): Remove variable.
+ (typical_sizes): Update comment. Remove 16K sizes.
+ (maximum_offset, maximum_offset_errno, maximum_offset_hard_limit):
+ Remove variables.
+ (find_maximum_offset, pipe_as_source, pipe_as_destination)
+ (delayed_write_failure_beginning, delayed_write_failure_end)
+ (cross_device_failure, enospc_failure_1, enospc_failure)
+ (oappend_failure): Remove functions.
+ (tests): Adjust test case list.
+ (do_test): Remove file system search code. Check for ENOSYS from
+ copy_file_range. Do not free xdevfile.
+ * manual/llio.texi (Copying File Data): Document ENOSYS error from
+ copy_file_range. Do not document the EXDEV error, which future
+ kernels may not report. Update the wording to reflect that
+ further errors are possible.
+ * sysdeps/unix/sysv/linux/alpha/kernel-features.h
+ [__LINUX_KERNEL_VERSION < 0x040D00] (__ASSUME_COPY_FILE_RANGE): Do
+ not undefine.
+ * sysdeps/unix/sysv/linux/kernel-features.h
+ [__LINUX_KERNEL_VERSION >= 0x040500] (__ASSUME_COPY_FILE_RANGE):
+ Remove definition.
+ * sysdeps/unix/sysv/linux/microblaze/kernel-features.h
+ [__LINUX_KERNEL_VERSION < 0x040A00] (__ASSUME_COPY_FILE_RANGE): Do
+ not undefine.
+
+2019-06-20 Dmitry V. Levin <ldv@altlinux.org>
+ Florian Weimer <fweimer@redhat.com>
+
+ [BZ #24228]
+ * libio/genops.c (_IO_unbuffer_all)
+ [SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_1)]: Do not attempt to free wide
+ buffers and access _IO_FILE_complete members of legacy libio streams.
+ * libio/tst-bz24228.c: New file.
+ * libio/tst-bz24228.map: Likewise.
+ * libio/Makefile [build-shared] (tests): Add tst-bz24228.
+ [build-shared] (generated): Add tst-bz24228.mtrace and
+ tst-bz24228.check.
+ [run-built-tests && build-shared] (tests-special): Add
+ $(objpfx)tst-bz24228-mem.out.
+ (LDFLAGS-tst-bz24228, tst-bz24228-ENV): New variables.
+ ($(objpfx)tst-bz24228-mem.out): New rule.
+
+2019-05-22 Wilco Dijkstra <wdijkstr@arm.com>
+
+ [BZ #24531]
+ * malloc/malloc.c (MAX_TCACHE_COUNT): New define.
+ (do_set_tcache_count): Only update if count is small enough.
+ * manual/tunables.texi (glibc.malloc.tcache_count): Document max value.
+
+2019-05-15 Mark Wielaard <mark@klomp.org>
+
+ [BZ#24476]
+ * dlfcn/dlerror.c (__dlerror_main_freeres): Guard using
+ __libc_once_get (once) and static_buf == NULL.
+ (__dlerror): Check we have a valid key, set result to static_buf
+ otherwise.
+
+2019-05-15 Andreas Schwab <schwab@suse.de>
+
+ [BZ #20568]
+ * libio/wfileops.c (_IO_wfile_sync): Correct last argument to
+ __codecvt_do_length.
+ * libio/Makefile (tests): Add tst-wfile-sync.
+ ($(objpfx)tst-wfile-sync.out): Depend on $(gen-locales).
+ * libio/tst-wfile-sync.c: New file.
+ * libio/tst-wfile-sync.input: New file.
+
+2018-12-21 Istvan Kurucsai <pistukem@gmail.com>
+
+ * malloc/malloc.c (munmap_chunk): Verify chunk alignment.
+
+2018-12-20 Istvan Kurucsai <pistukem@gmail.com>
+
+ * malloc/malloc.c (mremap_chunk): Additional checks.
+
+2018-08-17 Istvan Kurucsai <pistukem@gmail.com>
+
+ * malloc/malloc.c (_int_malloc): Additional binning code checks.
+
+2019-04-23 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+
+ [BZ #18035]
+ * elf/pldd-xx.c: Use _Static_assert in of pldd_assert.
+ (E(find_maps)): Avoid use alloca, use default read file operations
+ instead of explicit LFS names, and fix infinite loop.
+ * elf/pldd.c: Explicit set _FILE_OFFSET_BITS, cleanup headers.
+ (get_process_info): Use _Static_assert instead of assert, use default
+ directory operations instead of explicit LFS names, and free some
+ leadek pointers.
+
+2019-04-03 TAMUKI Shoichi <tamuki@linet.gr.jp>
+
+ [BZ #22964]
+ * localedata/locales/ja_JP (LC_TIME): Add entry for the new Japanese
+ era.
+
+2019-03-21 Stefan Liebler <stli@linux.ibm.com>
+
+ * sysdeps/s390/dl-procinfo.h (HWCAP_IMPORTANT):
+ Add HWCAP_S390_VX and HWCAP_S390_VXE.
+
+2019-01-31 Paul Eggert <eggert@cs.ucla.edu>
+
+ CVE-2019-9169
+ regex: fix read overrun [BZ #24114]
+ Problem found by AddressSanitizer, reported by Hongxu Chen in:
+ https://debbugs.gnu.org/34140
+ * posix/regexec.c (proceed_next_node):
+ Do not read past end of input buffer.
+
+2018-11-07 Andreas Schwab <schwab@suse.de>
+
+ [BZ #23864]
+ * sysdeps/unix/sysv/linux/riscv/kernel-features.h
+ (__ASSUME_SET_ROBUST_LIST) [__LINUX_KERNEL_VERSION < 0x041400]:
+ Undef.
+
+2018-09-21 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+
+ * NEWS: Add note about new TLE support on powerpc64le.
+ * sysdeps/powerpc/nptl/tcb-offsets.sym (TM_CAPABLE): Remove.
+ * sysdeps/powerpc/nptl/tls.h (tcbhead_t): Rename tm_capable to
+ __ununsed1.
+ (TLS_INIT_TP, TLS_DEFINE_INIT_TP): Remove tm_capable setup.
+ (THREAD_GET_TM_CAPABLE, THREAD_SET_TM_CAPABLE): Remove macros.
+ * sysdeps/powerpc/powerpc32/sysdep.h,
+ sysdeps/powerpc/powerpc64/sysdep.h (ABORT_TRANSACTION_IMPL,
+ ABORT_TRANSACTION): Remove macros.
+ * sysdeps/powerpc/sysdep.h (ABORT_TRANSACTION): Likewise.
+ * sysdeps/unix/sysv/linux/powerpc/elision-conf.c (elision_init): Set
+ __pthread_force_elision iff PPC_FEATURE2_HTM_NOSC is set.
+ * sysdeps/unix/sysv/linux/powerpc/powerpc32/sysdep.h,
+ sysdeps/unix/sysv/linux/powerpc/powerpc64/sysdep.h
+ sysdeps/unix/sysv/linux/powerpc/syscall.S (ABORT_TRANSACTION): Remove
+ usage.
+ * sysdeps/unix/sysv/linux/powerpc/not-errno.h: Remove file.
+
+2019-01-13 Jim Wilson <jimw@sifive.com>
+
+ [BZ #24040]
+ * elf/Makefile (CFLAGS-tst-unwind-main.c): Add -DUSE_PTHREADS=0.
+ * elf/tst-unwind-main.c: If USE_PTHEADS, include pthread.h and error.h
+ (func): New.
+ (main): If USE_PTHREADS, call pthread_create to run func. Otherwise
+ call func directly.
+ * nptl/Makefile (tests): Add tst-unwind-thread.
+ (CFLAGS-tst-unwind-thread.c): Define.
+ * nptl/tst-unwind-thread.c: New file.
+ * sysdeps/unix/sysv/linux/riscv/clone.S (__thread_start): Mark ra
+ as undefined.
+
+2019-01-31 Carlos O'Donell <carlos@redhat.com>
+ Torvald Riegel <triegel@redhat.com>
+ Rik Prohaska <prohaska7@gmail.com>
+
+ [BZ# 23844]
+ * nptl/Makefile (tests): Add tst-rwlock-tryrdlock-stall, and
+ tst-rwlock-trywrlock-stall.
+ * nptl/pthread_rwlock_tryrdlock.c (__pthread_rwlock_tryrdlock):
+ Wake waiters if PTHREAD_RWLOCK_FUTEX_USED is set.
+ * nptl/pthread_rwlock_trywrlock.c (__pthread_rwlock_trywrlock):
+ Set __wrphase_fute to 1 only if we started the write phase.
+ * nptl/tst-rwlock-tryrdlock-stall.c: New file.
+ * nptl/tst-rwlock-trywrlock-stall.c: New file.
+ * support/Makefile (libsupport-routines): Add xpthread_rwlock_destroy.
+ * support/xpthread_rwlock_destroy.c: New file.
+ * support/xthread.h: Declare xpthread_rwlock_destroy.
+
+2019-02-08 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #24161]
+ * sysdeps/nptl/fork.h (__run_fork_handlers): Add multiple_threads
+ argument.
+ * nptl/register-atfork.c (__run_fork_handlers): Only perform
+ locking if the new do_locking argument is true.
+ * sysdeps/nptl/fork.c (__libc_fork): Pass multiple_threads to
+ __run_fork_handlers.
+
+2019-02-07 Stefan Liebler <stli@linux.ibm.com>
+
+ [BZ #24180]
+ * nptl/pthread_mutex_trylock.c (__pthread_mutex_trylock):
+ Add compiler barriers and comments.
+
+2019-02-04 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #20018]
+ nscd: Do not rely on new GLIBC_PRIVATE ABI after CVE-2016-10739 fix.
+ * nscd/nscd-inet_addr.c: New file. Build resolv/inet_addr.c for
+ nscd, without public symbols.
+ * nscd/Makefile (nscd-modules): Add it.
+ * nscd/gai.c: Include <arpa/inet.h> and change visibility of
+ __inet_aton_exact.
+
+2019-01-21 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #20018]
+ CVE-2016-10739
+ resolv: Reject trailing characters in host names
+ * include/arpa/inet.h (__inet_aton_exact): Declare.
+ (inet_aton): Remove hidden prototype. No longer used internally.
+ * nscd/gai.c (__inet_aton): Do not define.
+ * nscd/gethstbynm3_r.c (__inet_aton): Likewise.
+ * nss/digits_dots.c (__inet_aton): Likewise.
+ (__nss_hostname_digits_dots_context): Call __inet_aton_exact.
+ * resolv/Makefile (tests-internal): Add tst-inet_aton_exact.
+ (tests): Add tst-resolv-nondecimal, tst-resolv-trailing.
+ (tst-resolv-nondecimal): Link with libresolv.so and libpthread.
+ (tst-resolv-trailing): Likewise.
+ * resolv/Versions (GLIBC_PRIVATE): Export __inet_aton_exact from
+ libc.
+ * resolv/inet_addr.c (inet_aton_end): Remame from __inet_aton.
+ Make static. Add endp parameter.
+ (__inet_aton_exact): New function.
+ (__inet_aton_ignore_trailing): New function, aliased to inet_aton.
+ (__inet_addr): Call inet_aton_end.
+ * resolv/res_init.c (res_vinit_1): Truncate nameserver for IPv4,
+ not just IPv6. Call __inet_aton_exact.
+ * resolv/tst-aton.c: Switch to <support/test-driver.c>.
+ (tests): Make const. Add additional test cases with trailing
+ characters.
+ (do_test): Use array_length.
+ * resolv/tst-inet_aton_exact.c: New file.
+ * resolv/tst-resolv-trailing.c: Likewise.
+ * resolv/tst-resolv-nondecimal.c: Likewise.
+ * sysdeps/posix/getaddrinfo.c (gaih_inet): Call __inet_aton_exact.
+
+2019-01-18 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #24112]
+ resolv: Do not send queries for non-host-names in nss_dns.
+ * resolv/nss_dns/dns-host.c (check_name): New function.
+ (_nss_dns_gethostbyname2_r): Use it.
+ (_nss_dns_gethostbyname_r): Likewise.
+ (_nss_dns_gethostbyname4_r): Likewise.
+
+2019-01-21 Florian Weimer <fweimer@redhat.com>
+
+ * resolv/inet_addr.c: Reformat to GNU style.
+ (__inet_addr, __inet_aton): Update comment.
+
+2019-02-04 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24155]
+ CVE-2019-7309
+ * NEWS: Updated for CVE-2019-7309.
+ * sysdeps/x86_64/memcmp.S: Use RDX_LP for size. Clear the
+ upper 32 bits of RDX register for x32. Use unsigned Jcc
+ instructions, instead of signed.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memcmp-2.
+ * sysdeps/x86_64/x32/tst-size_t-memcmp-2.c: New test.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/strlen-avx2.S: Use RSI_LP for length.
+ Clear the upper 32 bits of RSI register.
+ * sysdeps/x86_64/strlen.S: Use RSI_LP for length.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-strnlen
+ and tst-size_t-wcsnlen.
+ * sysdeps/x86_64/x32/tst-size_t-strnlen.c: New file.
+ * sysdeps/x86_64/x32/tst-size_t-wcsnlen.c: Likewise.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/strcpy-sse2-unaligned.S: Use RDX_LP
+ for length.
+ * sysdeps/x86_64/multiarch/strcpy-ssse3.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-strncpy.
+ * sysdeps/x86_64/x32/tst-size_t-strncpy.c: New file.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/strcmp-avx2.S: Use RDX_LP for length.
+ * sysdeps/x86_64/multiarch/strcmp-sse42.S: Likewise.
+ * sysdeps/x86_64/strcmp.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-strncasecmp,
+ tst-size_t-strncmp and tst-size_t-wcsncmp.
+ * sysdeps/x86_64/x32/tst-size_t-strncasecmp.c: New file.
+ * sysdeps/x86_64/x32/tst-size_t-strncmp.c: Likewise.
+ * sysdeps/x86_64/x32/tst-size_t-wcsncmp.c: Likewise.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/memset-avx512-no-vzeroupper.S: Use
+ RDX_LP for length. Clear the upper 32 bits of RDX register.
+ * sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-wmemset.
+ * sysdeps/x86_64/x32/tst-size_t-memset.c: New file.
+ * sysdeps/x86_64/x32/tst-size_t-wmemset.c: Likewise.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/memrchr.S: Use RDX_LP for length.
+ * sysdeps/x86_64/multiarch/memrchr-avx2.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memrchr.
+ * sysdeps/x86_64/x32/tst-size_t-memrchr.c: New file.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/memcpy-ssse3-back.S: Use RDX_LP for
+ length. Clear the upper 32 bits of RDX register.
+ * sysdeps/x86_64/multiarch/memcpy-ssse3.S: Likewise.
+ * sysdeps/x86_64/multiarch/memmove-avx512-no-vzeroupper.S:
+ Likewise.
+ * sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:
+ Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memcpy.
+ tst-size_t-wmemchr.
+ * sysdeps/x86_64/x32/tst-size_t-memcpy.c: New file.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/multiarch/memcmp-avx2-movbe.S: Use RDX_LP for
+ length. Clear the upper 32 bits of RDX register.
+ * sysdeps/x86_64/multiarch/memcmp-sse4.S: Likewise.
+ * sysdeps/x86_64/multiarch/memcmp-ssse3.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memcmp and
+ tst-size_t-wmemcmp.
+ * sysdeps/x86_64/x32/tst-size_t-memcmp.c: New file.
+ * sysdeps/x86_64/x32/tst-size_t-wmemcmp.c: Likewise.
+
+2019-02-01 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24097]
+ CVE-2019-6488
+ * sysdeps/x86_64/memchr.S: Use RDX_LP for length. Clear the
+ upper 32 bits of RDX register.
+ * sysdeps/x86_64/multiarch/memchr-avx2.S: Likewise.
+ * sysdeps/x86_64/x32/Makefile (tests): Add tst-size_t-memchr and
+ tst-size_t-wmemchr.
+ * sysdeps/x86_64/x32/test-size_t.h: New file.
+ * sysdeps/x86_64/x32/tst-size_t-memchr.c: Likewise.
+ * sysdeps/x86_64/x32/tst-size_t-wmemchr.c: Likewise.
+
+2019-01-16 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
+
+ * math/libm-test-fma.inc (fma_test_data): Set
+ XFAIL_ROUNDING_IBM128_LIBGCC to more tests.
+
+2019-01-07 Aurelien Jarno <aurelien@aurel32.net>
+
+ [BZ #24024]
+ * Makeconfig: Build libm with -fno-math-errno but build the remaining
+ code with -fmath-errno.
+ * string/Makefile [$(build-shared)] (tests): Add test-strerror-errno.
+ [$(build-shared)] (LDLIBS-test-strerror-errno): New variable.
+ * string/test-strerror-errno.c: New file.
+
+2019-01-03 Martin Jansa <Martin.Jansa@gmail.com>
+
+ [BZ #19444]
+ * sysdeps/ieee754/soft-fp/s_fdiv.c: Include <libc-diag.h> and use
+ DIAG_PUSH_NEEDS_COMMENT, DIAG_IGNORE_NEEDS_COMMENT and
+ DIAG_POP_NEEDS_COMMENT to disable -Wmaybe-uninitialized.
+
+2019-01-02 Aurelien Jarno <aurelien@aurel32.net>
+
+ [BZ #24034]
+ * sysdeps/unix/sysv/linux/arm/atomic-machine.h
+ (__arm_assisted_compare_and_exchange_val_32_acq): Use uint32_t rather
+ than __typeof (...) for the a_ptr variable.
+
+2018-12-31 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #24022]
+ * sysdeps/unix/sysv/linux/riscv/flush-icache.c: Check if
+ <asm/syscalls.h> exists with __has_include__ before including it.
+
+2019-01-02 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #24018]
+ * intl/dcigettext.c (DCIGETTEXT): Do not return NULL on asprintf
+ failure.
+
+2018-12-31 Florian Weimer <fw@deneb.enyo.de>
+
+ [BZ #24027]
+ * malloc/malloc.c (_int_realloc): Always call memcpy for the
+ copying operation. (ncopies had the wrong type, resulting in an
+ integer wraparound and too few elements being copied.)
+
+2018-12-28 Aurelien Jarno <aurelien@aurel32.net>
+
+ * sysdeps/alpha/fpu/libm-test-ulps: Regenerated.
+
+2018-12-18 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+ James Clarke <jrtc27@jrtc27.com>
+
+ [BZ #23967]
+ * sysdeps/unix/sysv/linux/kernel_sigaction.h (HAS_SA_RESTORER):
+ Define if SA_RESTORER is defined.
+ (kernel_sigaction): Define sa_restorer if HAS_SA_RESTORER is defined.
+ (SET_SA_RESTORER, RESET_SA_RESTORER): Define iff the macro are not
+ already defined.
+ * sysdeps/unix/sysv/linux/m68k/kernel_sigaction.h (SA_RESTORER,
+ kernel_sigaction, SET_SA_RESTORER, RESET_SA_RESTORER): Remove
+ definitions.
+ (HAS_SA_RESTORER): Define.
+ * sysdeps/unix/sysv/linux/sparc/kernel_sigaction.h (SA_RESTORER,
+ SET_SA_RESTORER, RESET_SA_RESTORER): Remove definition.
+ (HAS_SA_RESTORER): Define.
+ * sysdeps/unix/sysv/linux/nios2/kernel_sigaction.h: Include generic
+ kernel_sigaction after define SET_SA_RESTORER and RESET_SA_RESTORER.
+ * sysdeps/unix/sysv/linux/powerpc/kernel_sigaction.h: Likewise.
+ * sysdeps/unix/sysv/linux/s390/kernel_sigaction.h: Likewise.
+ * sysdeps/unix/sysv/linux/x86_64/sigaction.c: Likewise.
+
+2018-10-30 Andreas Schwab <schwab@suse.de>
+
+ [BZ #23125]
+ * sysdeps/riscv/start.S (ENTRY_POINT): Mark ra as undefined.
+ Don't use tail call.
+ * elf/tst-unwind-main.c: New file.
+ * elf/Makefile (tests): Add tst-unwind-main.
+ (CFLAGS-tst-unwind-main.c): Define.
+
+2018-12-15 Florian Weimer <fweimer@redhat.com>
+
+ * support/blob_repeat.c (check_mul_overflow_size_t): New function.
+ (minimum_stride_size): Use it.
+ (support_blob_repeat_allocate): Likewise.
+
+2018-12-13 Andreas Schwab <schwab@suse.de>
+
+ [BZ #23861]
+ * nptl/pthread_rwlock_common.c: Reindent. Fix typos.
+ (__pthread_rwlock_rdlock_full): Update expected value for
+ __readers while waiting on PTHREAD_RWLOCK_RWAITING.
+ * nptl/tst-rwlock-pwn.c: New file.
+ * nptl/Makefile (tests): Add tst-rwlock-pwn.
+
+2018-12-12 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
+
+ [BZ #23614]
+ * sysdeps/powerpc/powerpc64/addmul_1.S (FUNC): Add CFI offset for
+ registers saved in the stack frame.
+ * sysdeps/powerpc/powerpc64/lshift.S (__mpn_lshift): Likewise.
+ * sysdeps/powerpc/powerpc64/mul_1.S (__mpn_mul_1): Likewise.
+
+2018-12-07 DJ Delorie <dj@redhat.com>
+
+ [BZ #23907]
+ * malloc/tst-tcfree3.c: New.
+ * malloc/Makefile: Add it.
+
+2018-12-07 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23927]
+ CVE-2018-19591
+ * inet/tst-if_index-long.c: New file.
+ * inet/Makefile (tests): Add tst-if_index-long.
+
+2018-12-07 Florian Weimer <fweimer@redhat.com>
+
+ * support/check.h (support_record_failure_is_failed): Declare.
+ * support/descriptors.h: New file.
+ * support/support_descriptors.c: Likewise.
+ * support/tst-support_descriptors.c: Likewise.
+ * support/support_record_failure.c
+ (support_record_failure_is_failed): New function.
+ * support/Makefile (libsupport-routines): Add support_descriptors.
+ (tests): Add tst-support_descriptors.
+
+2018-12-01 Florian Weimer <fweimer@redhat.com>
+
+ * support/support_capture_subprocess.c
+ (support_capture_subprocess): Check that pipe descriptors have
+ expected values. Close original pipe descriptors in subprocess.
+
+2018-11-28 Florian Weimer <fweimer@redhat.com>
+
+ * support/support.h (support_quote_string): Do not use str
+ parameter name.
+
+2018-11-27 Florian Weimer <fweimer@redhat.com>
+
+ * support/support.h (support_quote_string): Declare.
+ * support/support_quote_string.c: New file.
+ * support/tst-support_quote_string.c: Likewise.
+ * support/Makefile (libsupport-routines): Add
+ support_quote_string.
+ (tests): Add tst-support_quote_string.
+
+2018-12-10 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23972]
+ * sysdeps/unix/sysv/linux/getdents64.c (handle_overflow): Check
+ offset instead of count for clarity. Fix typo in comment.
+ (__old_getdents64): Keep track of previous offset. Use it to call
+ handle_overflow.
+ * sysdeps/unix/sysv/linux/tst-readdir64-compat.c (do_test): Check
+ that d_off is never zero.
+
+2018-11-30 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
+
+ [BZ #23690]
+ * elf/dl-runtime.c (_dl_profile_fixup): Guarantee memory
+ modification order when accessing reloc_result->addr.
+ * include/link.h (reloc_result): Add field init.
+ * nptl/Makefile (tests): Add tst-audit-threads.
+ (modules-names): Add tst-audit-threads-mod1 and
+ tst-audit-threads-mod2.
+ Add rules to build tst-audit-threads.
+ * nptl/tst-audit-threads-mod1.c: New file.
+ * nptl/tst-audit-threads-mod2.c: Likewise.
+ * nptl/tst-audit-threads.c: Likewise.
+ * nptl/tst-audit-threads.h: Likewise.
+
+2018-11-26 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23907]
+ * malloc/malloc.c (_int_free): Validate tc_idx before checking for
+ double-frees.
+
+
+2018-11-20 DJ Delorie <dj@redhat.com>
+
+ * malloc/malloc.c (tcache_entry): Add key field.
+ (tcache_put): Set it.
+ (tcache_get): Likewise.
+ (_int_free): Check for double free in tcache.
+ * malloc/tst-tcfree1.c: New.
+ * malloc/tst-tcfree2.c: New.
+ * malloc/Makefile: Run the new tests.
+ * manual/probes.texi: Document memory_tcache_double_free probe.
+
+ * dlfcn/dlerror.c (check_free): Prevent double frees.
+
+2018-11-27 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23927]
+ CVE-2018-19591
+ * sysdeps/unix/sysv/linux/if_index.c (__if_nametoindex): Avoid
+ descriptor leak in case of ENODEV error.
+
+2018-11-19 Florian Weimer <fweimer@redhat.com>
+
+ support: Print timestamps in timeout handler.
+ * support/support_test_main.c (print_timestamp): New function.
+ (signal_handler): Use it to print the termination time and the
+ time of the last write to standard output.
+
+2018-10-09 Szabolcs Nagy <szabolcs.nagy@arm.com>
+
+ * libio/tst-readline.c (TIMEOUT): Define.
+
+2018-10-22 Joseph Myers <joseph@codesourcery.com>
+
+ * sysdeps/unix/sysv/linux/syscall-names.list: Update kernel
+ version to 4.19.
+
+2018-09-18 Paul Eggert <eggert@cs.ucla.edu>
+
+ Fix tzfile low-memory assertion failure
+ [BZ #21716]
+ * time/tzfile.c (__tzfile_read): Check for memory exhaustion
+ when registering time zone abbreviations.
+
+2018-08-31 Paul Pluzhnikov <ppluzhnikov@google.com>
+
+ [BZ #20271]
+ * include/stdio.h (__libc_fatal): Mention newline in comment.
+ * grp/initgroups.c (internal_getgrouplist): Add missing newline.
+ * nptl/pthread_cond_wait.c (__pthread_cond_wait_common): Likewise.
+ * nscd/initgrcache.c (addinitgroupsX): Likewise.
+ * nss/nsswitch.c (__nss_next2): Likewise.
+ * sysdeps/aarch64/dl-irel.h (elf_irela): Likewise.
+ * sysdeps/arm/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/generic/unwind-dw2.c (execute_cfa_program): Likewise.
+ * sysdeps/i386/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/powerpc/powerpc32/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/powerpc/powerpc64/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/s390/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/sparc/sparc32/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/sparc/sparc64/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/x86_64/dl-irel.h (elf_irel): Likewise.
+ * sysdeps/nptl/futex-internal.h (futex_wake): Likewise.
+ * sysdeps/unix/sysv/linux/netlink_assert_response.c
+ (__netlink_assert_response): Likewise.
+
+2018-08-28 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23520]
+ nscd: Fix use-after-free in addgetnetgrentX and its callers.
+ * nscd/netgroupcache.c
+ (addgetnetgrentX): Add tofreep parameter. Do not free
+ heap-allocated buffer.
+ (addinnetgrX): Free buffer allocated bt addgetnetgrentX.
+ (addgetnetgrentX_ignore): New function.
+ (addgetnetgrent): Call it.
+ (readdgetnetgrent): Likewise.
+
+2018-08-16 DJ Delorie <dj@delorie.com>
+
+ * malloc/malloc.c (_int_free): Check for corrupt prev_size vs size.
+ (malloc_consolidate): Likewise.
+
+2018-08-16 Pochang Chen <johnchen902@gmail.com>
+
+ * malloc/malloc.c (_int_malloc.c): Verify size of top chunk.
+
+2018-08-13 Joseph Myers <joseph@codesourcery.com>
+
+ * sysdeps/unix/sysv/linux/syscall-names.list: Update kernel
+ version to 4.18.
+ (io_pgetevents): New syscall.
+ (rseq): Likewise.
+
+2018-11-08 Alexandra Hájková <ahajkova@redhat.com>
+
+ [BZ #17630]
+ * resolv/tst-resolv-network.c: Add test for getnetbyname.
+
+2018-11-08 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #23509]
+ * sysdeps/x86/dl-prop.h (_dl_process_cet_property_note): Skip
+ note parsing if a NT_GNU_PROPERTY_TYPE_0 note has been processed.
+ Update the l_cet field when processing NT_GNU_PROPERTY_TYPE_0 note.
+ Check multiple NT_GNU_PROPERTY_TYPE_0 notes.
+ * sysdeps/x86/link_map.h (l_cet): Expand to 3 bits, Add
+ lc_unknown.
+
+2018-11-05 Andreas Schwab <schwab@suse.de>
+
+ [BZ #22927]
+ * resolv/gai_misc.c (__gai_enqueue_request): Don't crash if
+ creating the first helper thread failed.
+
+2018-10-23 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+
+ [BZ #23709]
+ * sysdeps/x86/cpu-features.c (init_cpu_features): Set TSX bits
+ independently of other flags.
+
+2018-10-30 Florian Weimer <fweimer@redhat.com>
+
+ * stdlib/tst-strtod-overflow.c (do_test): Switch to
+ support_blob_repeat.
+
+2018-10-30 Florian Weimer <fweimer@redhat.com>
+
+ * support/blob_repeat.c (allocate_big): Call mkstemp directly.
+
+2018-10-30 Florian Weimer <fweimer@redhat.com>
+
+ * stdlib/test-bz22786.c (do_test): Additional free calls to avoid
+ memory leaks.
+
+2018-10-30 Florian Weimer <fweimer@redhat.com>
+
+ Avoid spurious test failures in stdlib/test-bz22786.
+ * support/Makefile (libsupport-routines): Add blob_repeat.
+ (tests): Add tst-support_blob_repeat.
+ * support/blob_repeat.h: New file.
+ * support/blob_repeat.c: Likewise.
+ * support/tst-support_blob_repeat.c: Likewise.
+ * stdlib/test-bz22786.c (do_test): Replace malloc and memset with
+ support_blob_repeat_allocate.
+
+2018-08-30 Stefan Liebler <stli@linux.ibm.com>
+
+ * stdlib/test-bz22786.c (do_test): Return EXIT_UNSUPPORTED
+ if malloc fails.
+
+2018-08-24 Paul Pluzhnikov <ppluzhnikov@google.com>
+
+ [BZ #23400]
+ * stdlib/test-bz22786.c (do_test): Fix undefined behavior, don't
+ create temporary files in source tree.
+
+2018-10-26 Szabolcs Nagy <szabolcs.nagy@arm.com>
+
+ [BZ #23822]
+ * sysdeps/ia64/fpu/e_exp2f.S (exp2f): Use WEAK_LIBM_ENTRY.
+ * sysdeps/ia64/fpu/e_log2f.S (log2f): Likewise.
+ * sysdeps/ia64/fpu/e_exp2f.S (powf): Likewise.
+
+2018-10-25 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23562]
+ [BZ #23821]
+ XFAIL siginfo_t si_band conform test on sparc64.
+ * sysdeps/unix/sysv/linux/sparc/bits/siginfo-arch.h
+ (__SI_BAND_TYPE): Only override long int default type on sparc64.
+ * sysdeps/unix/sysv/linux/sparc/sparc64/Makefile
+ (conformtest-xfail-conds): Add sparc64-linux.
+ * conform/data/signal.h-data (siginfo_t): XFAIL si_band test on
+ sparc64.
+ * conform/data/sys/wait.h-data (siginfo_t): Likewise.
+
+2018-10-19 Ilya Yu. Malakhov <malakhov@mcst.ru>
+
+ [BZ #23562]
+ * sysdeps/unix/sysv/linux/bits/types/siginfo_t.h
+ (struct siginfo_t): Use correct type for si_band.
+
+2018-10-17 Stefan Liebler <stli@linux.ibm.com>
+
+ [BZ #23275]
+ * nptl/tst-mutex10.c: New File.
+ * nptl/Makefile (tests): Add tst-mutex10.
+ (tst-mutex10-ENV): New variable.
+ * sysdeps/unix/sysv/linux/s390/force-elision.h: (FORCE_ELISION):
+ Ensure that elision path is used if elision is available.
+ * sysdeps/unix/sysv/linux/powerpc/force-elision.h (FORCE_ELISION):
+ Likewise.
+ * sysdeps/unix/sysv/linux/x86/force-elision.h: (FORCE_ELISION):
+ Likewise.
+ * nptl/pthreadP.h (PTHREAD_MUTEX_TYPE, PTHREAD_MUTEX_TYPE_ELISION)
+ (PTHREAD_MUTEX_PSHARED): Use atomic_load_relaxed.
+ * nptl/pthread_mutex_consistent.c (pthread_mutex_consistent): Likewise.
+ * nptl/pthread_mutex_getprioceiling.c (pthread_mutex_getprioceiling):
+ Likewise.
+ * nptl/pthread_mutex_lock.c (__pthread_mutex_lock_full)
+ (__pthread_mutex_cond_lock_adjust): Likewise.
+ * nptl/pthread_mutex_setprioceiling.c (pthread_mutex_setprioceiling):
+ Likewise.
+ * nptl/pthread_mutex_timedlock.c (__pthread_mutex_timedlock): Likewise.
+ * nptl/pthread_mutex_trylock.c (__pthread_mutex_trylock): Likewise.
+ * nptl/pthread_mutex_unlock.c (__pthread_mutex_unlock_full): Likewise.
+ * sysdeps/nptl/bits/thread-shared-types.h (struct __pthread_mutex_s):
+ Add comments.
+ * nptl/pthread_mutex_destroy.c (__pthread_mutex_destroy):
+ Use atomic_load_relaxed and atomic_store_relaxed.
+ * nptl/pthread_mutex_init.c (__pthread_mutex_init):
+ Use atomic_store_relaxed.
+
+2018-10-09 H.J. Lu <hongjiu.lu@intel.com>
+
+ [BZ #23716]
+ * sysdeps/i386/dl-cet.c: Removed.
+ * sysdeps/i386/dl-machine.h (_dl_runtime_resolve_shstk): New
+ prototype.
+ (_dl_runtime_profile_shstk): Likewise.
+ (elf_machine_runtime_setup): Use _dl_runtime_profile_shstk or
+ _dl_runtime_resolve_shstk if SHSTK is enabled by kernel.
+
+2018-10-09 Rafal Luzynski <digitalfreak@lingonborough.com>
+
+ [BZ #20209]
+ * localedata/locales/kl_GL: (abday): Fix spelling of Sun (Sunday),
+ should be "sap" rather than "sab".
+ (day): Fix spelling of Sunday, should be "sapaat" rather than
+ "sabaat".
+
+2018-09-28 Adhemerval Zanella <adhemerval.zanella@linaro.org>
+
+ [BZ #23579]
+ * misc/tst-preadvwritev2-common.c (do_test_with_invalid_fd,
+ do_test_with_invalid_iov): New tests.
+ * misc/tst-preadvwritev2.c, misc/tst-preadvwritev64v2.c (do_test):
+ Call do_test_with_invalid_fd and do_test_with_invalid_iov.
+ * sysdeps/unix/sysv/linux/preadv2.c (preadv2): Use fallback code iff
+ errno is ENOSYS.
+ * sysdeps/unix/sysv/linux/preadv64v2.c (preadv64v2): Likewise.
+ * sysdeps/unix/sysv/linux/pwritev2.c (pwritev2): Likewise.
+ * sysdeps/unix/sysv/linux/pwritev64v2.c (pwritev64v2): Likewise.
+ * NEWS: Add bug fixed.
+
+2018-09-27 Andreas Schwab <schwab@suse.de>
+
+ [BZ #23717]
+ * stdlib/tst-setcontext9.c (f1a): Make st2 static.
+ (do_test): Make st1 static.
+
+2018-09-21 H.J. Lu <hongjiu.lu@intel.com>
+ Xuepeng Guo <xuepeng.guo@intel.com>
+
+ [BZ #23606]
+ * sysdeps/i386/start.S: Include <sysdep.h>
+ (_start): Use ENTRY/END to insert ENDBR32 at entry when CET is
+ enabled. Add cfi_undefined (eip).
+
+2018-09-19 Wilco Dijkstra <wdijkstr@arm.com>
+
+ [BZ #23637]
+ * string/test-strstr.c (pr23637): New function.
+ (test_main): Add tests with longer needles.
+ * string/strcasestr.c (AVAILABLE): Fix readahead distance.
+ * string/strstr.c (AVAILABLE): Likewise.
+
+2018-09-19 Carlos O'Donell <carlos@redhat.com>
+
+ * stdlib/tst-setcontext9.c (f1): Rename to...
+ (f1a): ... this.
+ (f1b): New function implementing lower half of f1 in alternate stack.
+
+2018-09-20 Florian Weimer <fweimer@redhat.com>
+
+ * misc/tst-gethostid.c: New file.
+ * misc/Makefile [$(build-shared)] (tests): Add tst-gethostid.
+ (tst-gethostid): Link with -ldl.
+
+2018-09-20 Mingli Yu <Mingli.Yu@windriver.com>
+
+ * sysdeps/unix/sysv/linux/gethostid.c (gethostid): Check for NULL
+ value from gethostbyname_r.
+
+2018-09-06 Stefan Liebler <stli@linux.ibm.com>
+
+ * sysdeps/unix/sysv/linux/spawni.c (maybe_script_execute):
+ Increment size of new_argv by one.
+
+2018-08-28 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23578]
+ * posix/tst-regcomp-truncated.c: New file.
+ * posix/Makefile (tests): Add it.
+ (tst-regcomp-truncated.out): Depend on generated locales.
+
+2018-08-25 Paul Eggert <eggert@cs.ucla.edu>
+
+ [BZ #23578]
+ regex: fix uninitialized memory access
+ I introduced this bug into gnulib in commit
+ 8335a4d6c7b4448cd0bcb6d0bebf1d456bcfdb17 dated 2006-04-10;
+ eventually it was merged into glibc. The bug was found by
+ project-repo <bugs@feusi.co> and reported here:
+ https://lists.gnu.org/r/sed-devel/2018-08/msg00017.html
+ Diagnosis and draft fix reported by Assaf Gordon here:
+ https://lists.gnu.org/r/bug-gnulib/2018-08/msg00071.html
+ https://lists.gnu.org/r/bug-gnulib/2018-08/msg00142.html
+ * posix/regex_internal.c (build_wcs_upper_buffer):
+ Fix bug when mbrtowc returns 0.
+
+2018-08-27 Martin Kuchta <martin.kuchta@netapp.com>
+ Torvald Riegel <triegel@redhat.com>
+
+ [BZ #23538]
+ * nptl/pthread_cond_common.c (__condvar_quiesce_and_switch_g1):
+ Update r to include the set wake-request flag if waiters are
+ remaining after spinning.
+
+2018-08-03 DJ Delorie <dj@redhat.com>
+
+ * sysdeps/riscv/rvf/math_private.h (libc_feholdexcept_setround_riscv):
+ Move libc_fesetround_riscv after libc_feholdexcept_riscv.
+
+ * sysdeps/riscv/rv64/rvd/libm-test-ulps: Update.
+
+2018-08-14 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23521]
+ [BZ #23522]
+ * nss/nss_files/files-alias.c (get_next_alias): During :include:
+ processing, bail out if no room, and close the stream before
+ returning ERANGE.
+ * nss/Makefile (tests): Add tst-nss-files-alias-leak.
+ (tst-nss-files-alias-leak): Link with libdl.
+ (tst-nss-files-alias-leak.out): Depend on nss_files.
+
+ * nss/tst-nss-files-alias-leak.c: New file.
+
+2018-08-14 Florian Weimer <fweimer@redhat.com>
+
+ * nscd/nscd_conf.c (nscd_parse_file): Deallocate old storage for
+ server_user, stat_user.
+
+2018-08-13 Florian Weimer <fweimer@redhat.com>
+
+ * misc/error.c (error): Add missing va_end call.
+ (error_at_line): Likewise.
+
+2018-08-10 Florian Weimer <fweimer@redhat.com>
+
+ [BZ #23497]
+ * sysdeps/unix/sysv/linux/getdents64.c (handle_overflow): New
+ function.
+ (__old_getdents64): Use getdents64. Convert entries without
+ moving them.
+ * sysdeps/unix/sysv/linux/tst-readdir64-compat.c: New file.
+ * sysdeps/unix/sysv/linux/Makefile (tests-internal): Add
+ tst-readdir64-compat.
+
+2018-08-08 Samuel Thibault <samuel.thibault@ens-lyon.org>
+
+ * htl/Versions (__pthread_getspecific, __pthread_setspecific): Add
+ symbols.
+ * sysdeps/htl/pthreadP.h [IS_IN (libpthread)] (__pthread_getspecific,
+ __pthread_setspecific): Add hidden proto.
+ * sysdeps/htl/pt-getspecific.c (__pthread_getspecific): Add hidden def.
+ * sysdeps/htl/pt-setspecific.c (__pthread_setspecific): Add hidden def.
+
2018-08-01 Carlos O'Donel <carlos@redhat.com>
* version.h (RELEASE): Set to "stable".
# disable any optimization that assume default rounding mode.
+math-flags = -frounding-math
-# Build libc/libm using -fno-math-errno, but run testsuite with -fmath-errno.
-+extra-math-flags = $(if $(filter libnldbl nonlib testsuite,$(in-module)),-fmath-errno,-fno-math-errno)
+# Logically only "libnldbl", "nonlib" and "testsuite" should be using
+# -fno-math-errno. However due to GCC bug #88576, only "libm" can use
+# -fno-math-errno.
++extra-math-flags = $(if $(filter libm,$(in-module)),-fno-math-errno,-fmath-errno)
# We might want to compile with some stack-protection flag.
ifneq ($(stack-protector),)
Please send GNU C library bug reports via <https://sourceware.org/bugzilla/>
using `glibc' in the "product" field.
\f
+Version 2.28.1
+
+Major new features:
+
+* The entry for the new Japanese era has been added for ja_JP locale.
+
+Deprecated and removed features, and other changes affecting compatibility:
+
+* For powercp64le ABI, Transactional Lock Elision is now enabled iff kernel
+ indicates that it will abort the transaction prior to entering the kernel
+ (PPC_FEATURE2_HTM_NOSC on hwcap2). On older kernels the transaction is
+ suspended, and this caused some undefined side-effects issues by aborting
+ transactions manually. Glibc avoided it by abort transactions manually on
+ each syscall, but it lead to performance issues on newer kernels where the
+ HTM state is saved and restore lazily (the state being saved even when the
+ process actually does not use HTM).
+
+* The copy_file_range function fails with ENOSYS if the kernel does not
+ support the system call of the same name. Previously, user space
+ emulation was performed, but its behavior did not match the kernel
+ behavior, which was deemed too confusing. Applications which use the
+ copy_file_range function will have to be run on kernels which implement
+ the copy_file_range system call. Support for most architectures was added
+ in version 4.5 of the mainline Linux kernel.
+
+The following bugs are resolved with this release:
+
+ [18035] Fix pldd hang
+ [19444] build failures with -O1 due to -Wmaybe-uninitialized
+ [20018] getaddrinfo should reject IP addresses with trailing characters
+ [20209] localedata: Spelling mistake for Sunday in Greenlandic kl_GL
+ [20568] Fix crash in _IO_wfile_sync
+ [22927] libanl: properly cleanup if first helper thread creation failed
+ [23400] stdlib/test-bz22786.c creates temporary files in glibc source tree
+ [23497] readdir64@GLIBC_2.1 cannot parse the kernel directory stream
+ [23509] CET enabled glibc is incompatible with the older linker
+ [23521] nss_files aliases database file stream leak
+ [23538] pthread_cond_broadcast: Fix waiters-after-spinning case
+ [23562] signal: Use correct type for si_band in siginfo_t
+ [23578] regex: Fix memory overread in re_compile_pattern
+ [23579] libc: Errors misreported in preadv2
+ [23606] Missing ENDBR32 in sysdeps/i386/start.S
+ [23614] powerpc: missing CFI register information in __mpn_* functions
+ [23679] gethostid: Missing NULL check for gethostbyname_r result
+ [23709] Fix CPU string flags for Haswell-type CPUs
+ [23717] Fix stack overflow in stdlib/tst-setcontext9
+ [23821] si_band in siginfo_t has wrong type long int on sparc64
+ [23822] ia64 static libm.a is missing exp2f, log2f and powf symbols
+ [23864] libc: [riscv] missing kernel-features.h undefines
+ [23844] pthread_rwlock_trywrlock results in hang
+ [23927] Linux if_nametoindex() does not close descriptor (CVE-2018-19591)
+ [23972] __old_getdents64 uses wrong d_off value on overflow
+ [24018] gettext may return NULL
+ [24022] riscv may lack <asm/syscalls.h>
+ [24024] strerror() might set errno to ENOMEM due to -fno-math-error
+ [24027] malloc: Integer overflow in realloc
+ [24034] tst-cancel21-static fails with SIGBUS on pre-ARMv7 when using GCC 8
+ [24040] riscv64: unterminated call chain in __thread_start
+ [24097] Can't use 64-bit register for size_t in assembly codes for x32 (CVE-2019-6488)
+ [24155] x32 memcmp can treat positive length as 0 (if sign bit in RDX is set) (CVE-2019-7309)
+ [24161] __run_fork_handlers self-deadlocks in malloc/tst-mallocfork2
+ [24228] old x86 applications that use legacy libio crash on exit
+ [24476] dlfcn: Guard __dlerror_main_freeres with __libc_once_get (once)
+ [24744] io: Remove the copy_file_range emulation.
+
+Security related changes:
+
+ CVE-2018-19591: A file descriptor leak in if_nametoindex can lead to a
+ denial of service due to resource exhaustion when processing getaddrinfo
+ calls with crafted host names. Reported by Guido Vranken.
+
+ CVE-2019-6488: On x32, the size_t parameter may be passed in the lower
+ 32 bits of a 64-bit register with with non-zero upper 32 bit. When it
+ happened, accessing the 32-bit size_t value as the full 64-bit register
+ in the assembly string/memory functions would cause a buffer overflow.
+ Reported by H.J. Lu.
+
+ CVE-2019-7309: x86-64 memcmp used signed Jcc instructions to check
+ size. For x86-64, memcmp on an object size larger than SSIZE_MAX
+ has undefined behavior. On x32, the size_t argument may be passed
+ in the lower 32 bits of the 64-bit RDX register with non-zero upper
+ 32 bits. When it happened with the sign bit of RDX register set,
+ memcmp gave the wrong result since it treated the size argument as
+ zero. Reported by H.J. Lu.
+
+ CVE-2016-10739: The getaddrinfo function could successfully parse IPv4
+ addresses with arbitrary trailing characters, potentially leading to data
+ or command injection issues in applications.
+
+ CVE-2019-9169: Attempted case-insensitive regular-expression match
+ via proceed_next_node in posix/regexec.c leads to heap-based buffer
+ over-read. Reported by Hongxu Chen.
+\f
Version 2.28
Major new features:
[23459] libc: COMMON_CPUID_INDEX_80000001 isn't populated for Intel
processors
[23467] dynamic-link: x86/CET: A property note parser bug
+ [24112] network: Do not send DNS queries for non-host names (where all
+ answers will be rejected)
\f
Version 2.27
element siginfo_t uid_t si_uid
element siginfo_t {void*} si_addr
element siginfo_t int si_status
-element siginfo_t long si_band
+// Bug 23821: si_band has type int on sparc64.
+xfail[sparc64-linux]-element siginfo_t long si_band
# endif
# ifndef XPG42
element siginfo_t {union sigval} si_value
element siginfo_t uid_t si_uid
element siginfo_t {void*} si_addr
element siginfo_t int si_status
-element siginfo_t long si_band
+// Bug 23821: si_band has type int on sparc64.
+xfail[sparc64-linux]-element siginfo_t long si_band
# ifndef XPG42
element siginfo_t {union sigval} si_value
# endif
__libc_once (once, init);
/* Get error string. */
- result = (struct dl_action_result *) __libc_getspecific (key);
- if (result == NULL)
- result = &last_result;
+ if (static_buf != NULL)
+ result = static_buf;
+ else
+ {
+ /* init () has been run and we don't use the static buffer.
+ So we have a valid key. */
+ result = (struct dl_action_result *) __libc_getspecific (key);
+ if (result == NULL)
+ result = &last_result;
+ }
/* Test whether we already returned the string. */
if (result->returned != 0)
Dl_info info;
if (_dl_addr (check_free, &info, &map, NULL) != 0 && map->l_ns == 0)
#endif
- free ((char *) rec->errstring);
+ {
+ free ((char *) rec->errstring);
+ rec->errstring = NULL;
+ }
}
}
void
__dlerror_main_freeres (void)
{
- void *mem;
/* Free the global memory if used. */
check_free (&last_result);
- /* Free the TSD memory if used. */
- mem = __libc_getspecific (key);
- if (mem != NULL)
- free_key_mem (mem);
+
+ if (__libc_once_get (once) && static_buf == NULL)
+ {
+ /* init () has been run and we don't use the static buffer.
+ So we have a valid key. */
+ void *mem;
+ /* Free the TSD memory if used. */
+ mem = __libc_getspecific (key);
+ if (mem != NULL)
+ free_key_mem (mem);
+ }
}
struct dlfcn_hook *_dlfcn_hook __attribute__((nocommon));
tst-tlsalign tst-tlsalign-extern tst-nodelete-opened \
tst-nodelete2 tst-audit11 tst-audit12 tst-dlsym-error tst-noload \
tst-latepthread tst-tls-manydynamic tst-nodelete-dlclose \
- tst-debug1 tst-main1 tst-absolute-sym tst-absolute-zero tst-big-note
+ tst-debug1 tst-main1 tst-absolute-sym tst-absolute-zero tst-big-note \
+ tst-unwind-main
# reldep9
tests-internal += loadtest unload unload2 circleload1 \
neededtest neededtest2 neededtest3 neededtest4 \
$(objpfx)tst-libc_dlvsym-static.out: $(objpfx)tst-libc_dlvsym-dso.so
$(objpfx)tst-big-note: $(objpfx)tst-big-note-lib.so
+
+CFLAGS-tst-unwind-main.c += -funwind-tables -DUSE_PTHREADS=0
/* This is the address in the array where we store the result of previous
relocations. */
struct reloc_result *reloc_result = &l->l_reloc_result[reloc_index];
- DL_FIXUP_VALUE_TYPE *resultp = &reloc_result->addr;
- DL_FIXUP_VALUE_TYPE value = *resultp;
- if (DL_FIXUP_VALUE_CODE_ADDR (value) == 0)
+ /* CONCURRENCY NOTES:
+
+ Multiple threads may be calling the same PLT sequence and with
+ LD_AUDIT enabled they will be calling into _dl_profile_fixup to
+ update the reloc_result with the result of the lazy resolution.
+ The reloc_result guard variable is reloc_init, and we use
+ acquire/release loads and store to it to ensure that the results of
+ the structure are consistent with the loaded value of the guard.
+ This does not fix all of the data races that occur when two or more
+ threads read reloc_result->reloc_init with a value of zero and read
+ and write to that reloc_result concurrently. The expectation is
+ generally that while this is a data race it works because the
+ threads write the same values. Until the data races are fixed
+ there is a potential for problems to arise from these data races.
+ The reloc result updates should happen in parallel but there should
+ be an atomic RMW which does the final update to the real result
+ entry (see bug 23790).
+
+ The following code uses reloc_result->init set to 0 to indicate if it is
+ the first time this object is being relocated, otherwise 1 which
+ indicates the object has already been relocated.
+
+ Reading/Writing from/to reloc_result->reloc_init must not happen
+ before previous writes to reloc_result complete as they could
+ end-up with an incomplete struct. */
+ DL_FIXUP_VALUE_TYPE value;
+ unsigned int init = atomic_load_acquire (&reloc_result->init);
+
+ if (init == 0)
{
/* This is the first time we have to relocate this object. */
const ElfW(Sym) *const symtab
/* Store the result for later runs. */
if (__glibc_likely (! GLRO(dl_bind_not)))
- *resultp = value;
+ {
+ reloc_result->addr = value;
+ /* Guarantee all previous writes complete before
+ init is updated. See CONCURRENCY NOTES earlier */
+ atomic_store_release (&reloc_result->init, 1);
+ }
+ init = 1;
}
+ else
+ value = reloc_result->addr;
/* By default we do not call the pltexit function. */
long int framesize = -1;
+
#ifdef SHARED
/* Auditing checkpoint: report the PLT entering and allow the
auditors to change the value. */
- if (DL_FIXUP_VALUE_CODE_ADDR (value) != 0 && GLRO(dl_naudit) > 0
+ if (GLRO(dl_naudit) > 0
/* Don't do anything if no auditor wants to intercept this call. */
&& (reloc_result->enterexit & LA_SYMB_NOPLTENTER) == 0)
{
+ /* Sanity check: DL_FIXUP_VALUE_CODE_ADDR (value) should have been
+ initialized earlier in this function or in another thread. */
+ assert (DL_FIXUP_VALUE_CODE_ADDR (value) != 0);
ElfW(Sym) *defsym = ((ElfW(Sym) *) D_PTR (reloc_result->bound,
l_info[DT_SYMTAB])
+ reloc_result->boundndx);
#define R_AARCH64_TLSDESC 1031 /* TLS Descriptor. */
#define R_AARCH64_IRELATIVE 1032 /* STT_GNU_IFUNC relocation. */
+/* AArch64 specific values for the Dyn d_tag field. */
+#define DT_AARCH64_VARIANT_PCS (DT_LOPROC + 5)
+#define DT_AARCH64_NUM 6
+
+/* AArch64 specific values for the st_other field. */
+#define STO_AARCH64_VARIANT_PCS 0x80
+
/* ARM relocs. */
#define R_ARM_NONE 0 /* No reloc */
#define EW_(e, w, t) EW__(e, w, _##t)
#define EW__(e, w, t) e##w##t
-#define pldd_assert(name, exp) \
- typedef int __assert_##name[((exp) != 0) - 1]
-
-
struct E(link_map)
{
EW(Addr) l_addr;
EW(Addr) l_libname;
};
#if CLASS == __ELF_NATIVE_CLASS
-pldd_assert (l_addr, (offsetof (struct link_map, l_addr)
- == offsetof (struct E(link_map), l_addr)));
-pldd_assert (l_name, (offsetof (struct link_map, l_name)
- == offsetof (struct E(link_map), l_name)));
-pldd_assert (l_next, (offsetof (struct link_map, l_next)
- == offsetof (struct E(link_map), l_next)));
+_Static_assert (offsetof (struct link_map, l_addr)
+ == offsetof (struct E(link_map), l_addr), "l_addr");
+_Static_assert (offsetof (struct link_map, l_name)
+ == offsetof (struct E(link_map), l_name), "l_name");
+_Static_assert (offsetof (struct link_map, l_next)
+ == offsetof (struct E(link_map), l_next), "l_next");
#endif
EW(Addr) next;
};
#if CLASS == __ELF_NATIVE_CLASS
-pldd_assert (name, (offsetof (struct libname_list, name)
- == offsetof (struct E(libname_list), name)));
-pldd_assert (next, (offsetof (struct libname_list, next)
- == offsetof (struct E(libname_list), next)));
+_Static_assert (offsetof (struct libname_list, name)
+ == offsetof (struct E(libname_list), name), "name");
+_Static_assert (offsetof (struct libname_list, next)
+ == offsetof (struct E(libname_list), next), "next");
#endif
struct E(r_debug)
EW(Addr) r_map;
};
#if CLASS == __ELF_NATIVE_CLASS
-pldd_assert (r_version, (offsetof (struct r_debug, r_version)
- == offsetof (struct E(r_debug), r_version)));
-pldd_assert (r_map, (offsetof (struct r_debug, r_map)
- == offsetof (struct E(r_debug), r_map)));
+_Static_assert (offsetof (struct r_debug, r_version)
+ == offsetof (struct E(r_debug), r_version), "r_version");
+_Static_assert (offsetof (struct r_debug, r_map)
+ == offsetof (struct E(r_debug), r_map), "r_map");
#endif
static int
-E(find_maps) (pid_t pid, void *auxv, size_t auxv_size)
+E(find_maps) (const char *exe, int memfd, pid_t pid, void *auxv,
+ size_t auxv_size)
{
EW(Addr) phdr = 0;
unsigned int phnum = 0;
if (phdr == 0 || phnum == 0 || phent == 0)
error (EXIT_FAILURE, 0, gettext ("cannot find program header of process"));
- EW(Phdr) *p = alloca (phnum * phent);
- if (pread64 (memfd, p, phnum * phent, phdr) != phnum * phent)
- {
- error (0, 0, gettext ("cannot read program header"));
- return EXIT_FAILURE;
- }
+ EW(Phdr) *p = xmalloc (phnum * phent);
+ if (pread (memfd, p, phnum * phent, phdr) != phnum * phent)
+ error (EXIT_FAILURE, 0, gettext ("cannot read program header"));
/* Determine the load offset. We need this for interpreting the
other program header entries so we do this in a separate loop.
if (p[i].p_type == PT_DYNAMIC)
{
EW(Dyn) *dyn = xmalloc (p[i].p_filesz);
- if (pread64 (memfd, dyn, p[i].p_filesz, offset + p[i].p_vaddr)
+ if (pread (memfd, dyn, p[i].p_filesz, offset + p[i].p_vaddr)
!= p[i].p_filesz)
- {
- error (0, 0, gettext ("cannot read dynamic section"));
- return EXIT_FAILURE;
- }
+ error (EXIT_FAILURE, 0, gettext ("cannot read dynamic section"));
/* Search for the DT_DEBUG entry. */
for (unsigned int j = 0; j < p[i].p_filesz / sizeof (EW(Dyn)); ++j)
if (dyn[j].d_tag == DT_DEBUG && dyn[j].d_un.d_ptr != 0)
{
struct E(r_debug) r;
- if (pread64 (memfd, &r, sizeof (r), dyn[j].d_un.d_ptr)
+ if (pread (memfd, &r, sizeof (r), dyn[j].d_un.d_ptr)
!= sizeof (r))
- {
- error (0, 0, gettext ("cannot read r_debug"));
- return EXIT_FAILURE;
- }
+ error (EXIT_FAILURE, 0, gettext ("cannot read r_debug"));
if (r.r_map != 0)
{
}
else if (p[i].p_type == PT_INTERP)
{
- interp = alloca (p[i].p_filesz);
- if (pread64 (memfd, interp, p[i].p_filesz, offset + p[i].p_vaddr)
+ interp = xmalloc (p[i].p_filesz);
+ if (pread (memfd, interp, p[i].p_filesz, offset + p[i].p_vaddr)
!= p[i].p_filesz)
- {
- error (0, 0, gettext ("cannot read program interpreter"));
- return EXIT_FAILURE;
- }
+ error (EXIT_FAILURE, 0, gettext ("cannot read program interpreter"));
}
if (list == 0)
if (interp == NULL)
{
// XXX check whether the executable itself is the loader
- return EXIT_FAILURE;
+ exit (EXIT_FAILURE);
}
// XXX perhaps try finding ld.so and _r_debug in it
-
- return EXIT_FAILURE;
+ exit (EXIT_FAILURE);
}
+ free (p);
+ free (interp);
+
/* Print the PID and program name first. */
printf ("%lu:\t%s\n", (unsigned long int) pid, exe);
do
{
struct E(link_map) m;
- if (pread64 (memfd, &m, sizeof (m), list) != sizeof (m))
- {
- error (0, 0, gettext ("cannot read link map"));
- status = EXIT_FAILURE;
- goto out;
- }
+ if (pread (memfd, &m, sizeof (m), list) != sizeof (m))
+ error (EXIT_FAILURE, 0, gettext ("cannot read link map"));
EW(Addr) name_offset = m.l_name;
- again:
while (1)
{
- ssize_t n = pread64 (memfd, tmpbuf.data, tmpbuf.length, name_offset);
+ ssize_t n = pread (memfd, tmpbuf.data, tmpbuf.length, name_offset);
if (n == -1)
- {
- error (0, 0, gettext ("cannot read object name"));
- status = EXIT_FAILURE;
- goto out;
- }
+ error (EXIT_FAILURE, 0, gettext ("cannot read object name"));
if (memchr (tmpbuf.data, '\0', n) != NULL)
break;
if (!scratch_buffer_grow (&tmpbuf))
- {
- error (0, 0, gettext ("cannot allocate buffer for object name"));
- status = EXIT_FAILURE;
- goto out;
- }
+ error (EXIT_FAILURE, 0,
+ gettext ("cannot allocate buffer for object name"));
}
- if (((char *)tmpbuf.data)[0] == '\0' && name_offset == m.l_name
- && m.l_libname != 0)
- {
- /* Try the l_libname element. */
- struct E(libname_list) ln;
- if (pread64 (memfd, &ln, sizeof (ln), m.l_libname) == sizeof (ln))
- {
- name_offset = ln.name;
- goto again;
- }
- }
+ /* The m.l_name and m.l_libname.name for loader linkmap points to same
+ values (since BZ#387 fix). Trying to use l_libname name as the
+ shared object name might lead to an infinite loop (BZ#18035). */
/* Skip over the executable. */
if (((char *)tmpbuf.data)[0] != '\0')
}
while (list != 0);
- out:
scratch_buffer_free (&tmpbuf);
return status;
}
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-#include <alloca.h>
+#define _FILE_OFFSET_BITS 64
+
#include <argp.h>
-#include <assert.h>
#include <dirent.h>
-#include <elf.h>
-#include <errno.h>
#include <error.h>
#include <fcntl.h>
#include <libintl.h>
-#include <link.h>
-#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
-#include <string.h>
#include <unistd.h>
#include <sys/ptrace.h>
-#include <sys/stat.h>
#include <sys/wait.h>
#include <scratch_buffer.h>
options, parse_opt, args_doc, doc, NULL, more_help, NULL
};
-// File descriptor of /proc/*/mem file.
-static int memfd;
-
-/* Name of the executable */
-static char *exe;
/* Local functions. */
-static int get_process_info (int dfd, long int pid);
+static int get_process_info (const char *exe, int dfd, long int pid);
static void wait_for_ptrace_stop (long int pid);
return 1;
}
- assert (sizeof (pid_t) == sizeof (int)
- || sizeof (pid_t) == sizeof (long int));
+ _Static_assert (sizeof (pid_t) == sizeof (int)
+ || sizeof (pid_t) == sizeof (long int),
+ "sizeof (pid_t) != sizeof (int) or sizeof (long int)");
+
char *endp;
errno = 0;
long int pid = strtol (argv[remaining], &endp, 10);
if (dfd == -1)
error (EXIT_FAILURE, errno, gettext ("cannot open %s"), buf);
- struct scratch_buffer exebuf;
- scratch_buffer_init (&exebuf);
+ /* Name of the executable */
+ struct scratch_buffer exe;
+ scratch_buffer_init (&exe);
ssize_t nexe;
while ((nexe = readlinkat (dfd, "exe",
- exebuf.data, exebuf.length)) == exebuf.length)
+ exe.data, exe.length)) == exe.length)
{
- if (!scratch_buffer_grow (&exebuf))
+ if (!scratch_buffer_grow (&exe))
{
nexe = -1;
break;
}
}
if (nexe == -1)
- exe = (char *) "<program name undetermined>";
+ /* Default stack allocation is at least 1024. */
+ snprintf (exe.data, exe.length, "<program name undetermined>");
else
- {
- exe = exebuf.data;
- exe[nexe] = '\0';
- }
+ ((char*)exe.data)[nexe] = '\0';
/* Stop all threads since otherwise the list of loaded modules might
change while we are reading it. */
error (EXIT_FAILURE, errno, gettext ("cannot prepare reading %s/task"),
buf);
- struct dirent64 *d;
- while ((d = readdir64 (dir)) != NULL)
+ struct dirent *d;
+ while ((d = readdir (dir)) != NULL)
{
if (! isdigit (d->d_name[0]))
continue;
wait_for_ptrace_stop (tid);
- struct thread_list *newp = alloca (sizeof (*newp));
+ struct thread_list *newp = xmalloc (sizeof (*newp));
newp->tid = tid;
newp->next = thread_list;
thread_list = newp;
closedir (dir);
- int status = get_process_info (dfd, pid);
+ if (thread_list == NULL)
+ error (EXIT_FAILURE, 0, gettext ("no valid %s/task entries"), buf);
+
+ int status = get_process_info (exe.data, dfd, pid);
- assert (thread_list != NULL);
do
{
ptrace (PTRACE_DETACH, thread_list->tid, NULL, NULL);
+ struct thread_list *prev = thread_list;
thread_list = thread_list->next;
+ free (prev);
}
while (thread_list != NULL);
close (dfd);
+ scratch_buffer_free (&exe);
return status;
}
static int
-get_process_info (int dfd, long int pid)
+get_process_info (const char *exe, int dfd, long int pid)
{
- memfd = openat (dfd, "mem", O_RDONLY);
+ /* File descriptor of /proc/<pid>/mem file. */
+ int memfd = openat (dfd, "mem", O_RDONLY);
if (memfd == -1)
goto no_info;
int retval;
if (e_ident[EI_CLASS] == ELFCLASS32)
- retval = find_maps32 (pid, auxv, auxv_size);
+ retval = find_maps32 (exe, memfd, pid, auxv, auxv_size);
else
- retval = find_maps64 (pid, auxv, auxv_size);
+ retval = find_maps64 (exe, memfd, pid, auxv, auxv_size);
free (auxv);
close (memfd);
--- /dev/null
+/* Test unwinding through main.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <unwind.h>
+#include <unistd.h>
+#include <support/test-driver.h>
+
+#if USE_PTHREADS
+# include <pthread.h>
+# include <error.h>
+#endif
+
+static _Unwind_Reason_Code
+callback (struct _Unwind_Context *ctx, void *arg)
+{
+ return _URC_NO_REASON;
+}
+
+static void *
+func (void *a)
+{
+ /* Arrange for this test to be killed if _Unwind_Backtrace runs into an
+ endless loop. We cannot use the test driver because the complete
+ call chain needs to be compiled with -funwind-tables so that
+ _Unwind_Backtrace is able to reach the start routine. */
+ alarm (DEFAULT_TIMEOUT);
+ _Unwind_Backtrace (callback, 0);
+ return a;
+}
+
+int
+main (void)
+{
+#if USE_PTHREADS
+ pthread_t thr;
+ int rc = pthread_create (&thr, NULL, &func, NULL);
+ if (rc)
+ error (1, rc, "pthread_create");
+ rc = pthread_join (thr, NULL);
+ if (rc)
+ error (1, rc, "pthread_join");
+#else
+ func (NULL);
+#endif
+}
/* This is really only for debugging. */
if (NSS_STATUS_TRYAGAIN > status || status > NSS_STATUS_RETURN)
- __libc_fatal ("illegal status in internal_getgrouplist");
+ __libc_fatal ("Illegal status in internal_getgrouplist.\n");
/* For compatibility reason we will continue to look for more
entries using the next service even though data has already
__cthread_keycreate;
__cthread_getspecific;
__cthread_setspecific;
+ __pthread_getspecific;
+ __pthread_setspecific;
__pthread_getattr_np;
__pthread_attr_getstack;
}
#include <inet/arpa/inet.h>
#ifndef _ISOMAC
-extern int __inet_aton (const char *__cp, struct in_addr *__inp);
-libc_hidden_proto (__inet_aton)
+/* Variant of inet_aton which rejects trailing garbage. */
+extern int __inet_aton_exact (const char *__cp, struct in_addr *__inp);
+libc_hidden_proto (__inet_aton_exact)
-libc_hidden_proto (inet_aton)
libc_hidden_proto (inet_ntop)
libc_hidden_proto (inet_pton)
extern __typeof (inet_pton) __inet_pton;
unsigned int boundndx;
uint32_t enterexit;
unsigned int flags;
+ /* CONCURRENCY NOTE: This is used to guard the concurrent initialization
+ of the relocation result across multiple threads. See the more
+ detailed notes in elf/dl-runtime.c. */
+ unsigned int init;
} *l_reloc_result;
/* Pointer to the version information if available. */
do_backtrace = 1 << 1 /* Backtrace. */
};
-/* Print out MESSAGE on the error output and abort. */
+/* Print out MESSAGE (which should end with a newline) on the error output
+ and abort. */
extern void __libc_fatal (const char *__message)
__attribute__ ((__noreturn__));
extern void __libc_message (enum __libc_message_action action,
tests := htontest test_ifindex tst-ntoa tst-ether_aton tst-network \
tst-gethnm test-ifaddrs bug-if1 test-inet6_opt tst-ether_line \
tst-getni1 tst-getni2 tst-inet6_rth tst-checks tst-checks-posix \
- tst-sockaddr test-hnto-types
+ tst-sockaddr test-hnto-types tst-if_index-long
# tst-deadline must be linked statically so that we can access
# internal functions.
--- /dev/null
+/* Check for descriptor leak in if_nametoindex with a long interface name.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* This test checks for a descriptor leak in case of a long interface
+ name (CVE-2018-19591, bug 23927). */
+
+#include <errno.h>
+#include <net/if.h>
+#include <netdb.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/descriptors.h>
+#include <support/support.h>
+
+static int
+do_test (void)
+{
+ struct support_descriptors *descrs = support_descriptors_list ();
+
+ /* Prepare a name which is just as long as required for trigging the
+ bug. */
+ char name[IFNAMSIZ + 1];
+ memset (name, 'A', IFNAMSIZ);
+ name[IFNAMSIZ] = '\0';
+ TEST_COMPARE (strlen (name), IFNAMSIZ);
+ struct ifreq ifr;
+ TEST_COMPARE (strlen (name), sizeof (ifr.ifr_name));
+
+ /* Test directly via if_nametoindex. */
+ TEST_COMPARE (if_nametoindex (name), 0);
+ TEST_COMPARE (errno, ENODEV);
+ support_descriptors_check (descrs);
+
+ /* Same test via getaddrinfo. */
+ char *host = xasprintf ("fea0::%%%s", name);
+ struct addrinfo hints = { .ai_flags = AI_NUMERICHOST, };
+ struct addrinfo *ai;
+ TEST_COMPARE (getaddrinfo (host, NULL, &hints, &ai), EAI_NONAME);
+ support_descriptors_check (descrs);
+
+ support_descriptors_free (descrs);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
int ret = __asprintf (&xdirname, "%s/%s", cwd, dirname);
free (cwd);
if (ret < 0)
- return NULL;
+ goto return_untranslated;
dirname = xdirname;
}
#ifndef IN_LIBGLOCALE
tst-fts tst-fts-lfs tst-open-tmpfile \
tst-copy_file_range tst-getcwd-abspath \
-# This test includes the compat implementation of copy_file_range,
-# which uses internal, unexported libc functions.
-tests-static += tst-copy_file_range-compat
-tests-internal += tst-copy_file_range-compat
-
# Likewise for statx, but we do not need static linking here.
tests-internal += tst-statx
+++ /dev/null
-/* Emulation of copy_file_range.
- Copyright (C) 2017-2018 Free Software Foundation, Inc.
- This file is part of the GNU C Library.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <http://www.gnu.org/licenses/>. */
-
-/* The following macros should be defined before including this
- file:
-
- COPY_FILE_RANGE_DECL Declaration specifiers for the function below.
- COPY_FILE_RANGE Name of the function to define. */
-
-#include <errno.h>
-#include <fcntl.h>
-#include <inttypes.h>
-#include <limits.h>
-#include <sys/stat.h>
-#include <sys/types.h>
-#include <unistd.h>
-
-COPY_FILE_RANGE_DECL
-ssize_t
-COPY_FILE_RANGE (int infd, __off64_t *pinoff,
- int outfd, __off64_t *poutoff,
- size_t length, unsigned int flags)
-{
- if (flags != 0)
- {
- __set_errno (EINVAL);
- return -1;
- }
-
- {
- struct stat64 instat;
- struct stat64 outstat;
- if (fstat64 (infd, &instat) != 0 || fstat64 (outfd, &outstat) != 0)
- return -1;
- if (S_ISDIR (instat.st_mode) || S_ISDIR (outstat.st_mode))
- {
- __set_errno (EISDIR);
- return -1;
- }
- if (!S_ISREG (instat.st_mode) || !S_ISREG (outstat.st_mode))
- {
- /* We need a regular input file so that the we can seek
- backwards in case of a write failure. */
- __set_errno (EINVAL);
- return -1;
- }
- if (instat.st_dev != outstat.st_dev)
- {
- /* Cross-device copies are not supported. */
- __set_errno (EXDEV);
- return -1;
- }
- }
-
- /* The output descriptor must not have O_APPEND set. */
- {
- int flags = __fcntl (outfd, F_GETFL);
- if (flags & O_APPEND)
- {
- __set_errno (EBADF);
- return -1;
- }
- }
-
- /* Avoid an overflow in the result. */
- if (length > SSIZE_MAX)
- length = SSIZE_MAX;
-
- /* Main copying loop. The buffer size is arbitrary and is a
- trade-off between stack size consumption, cache usage, and
- amortization of system call overhead. */
- size_t copied = 0;
- char buf[8192];
- while (length > 0)
- {
- size_t to_read = length;
- if (to_read > sizeof (buf))
- to_read = sizeof (buf);
-
- /* Fill the buffer. */
- ssize_t read_count;
- if (pinoff == NULL)
- read_count = read (infd, buf, to_read);
- else
- read_count = __libc_pread64 (infd, buf, to_read, *pinoff);
- if (read_count == 0)
- /* End of file reached prematurely. */
- return copied;
- if (read_count < 0)
- {
- if (copied > 0)
- /* Report the number of bytes copied so far. */
- return copied;
- return -1;
- }
- if (pinoff != NULL)
- *pinoff += read_count;
-
- /* Write the buffer part which was read to the destination. */
- char *end = buf + read_count;
- for (char *p = buf; p < end; )
- {
- ssize_t write_count;
- if (poutoff == NULL)
- write_count = write (outfd, p, end - p);
- else
- write_count = __libc_pwrite64 (outfd, p, end - p, *poutoff);
- if (write_count < 0)
- {
- /* Adjust the input read position to match what we have
- written, so that the caller can pick up after the
- error. */
- size_t written = p - buf;
- /* NB: This needs to be signed so that we can form the
- negative value below. */
- ssize_t overread = read_count - written;
- if (pinoff == NULL)
- {
- if (overread > 0)
- {
- /* We are on an error recovery path, so we
- cannot deal with failure here. */
- int save_errno = errno;
- (void) __libc_lseek64 (infd, -overread, SEEK_CUR);
- __set_errno (save_errno);
- }
- }
- else /* pinoff != NULL */
- *pinoff -= overread;
-
- if (copied + written > 0)
- /* Report the number of bytes copied so far. */
- return copied + written;
- return -1;
- }
- p += write_count;
- if (poutoff != NULL)
- *poutoff += write_count;
- } /* Write loop. */
-
- copied += read_count;
- length -= read_count;
- }
- return copied;
-}
-/* Generic implementation of copy_file_range.
- Copyright (C) 2017-2018 Free Software Foundation, Inc.
+/* Stub implementation of copy_file_range.
+ Copyright (C) 2017-2019 Free Software Foundation, Inc.
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-#define COPY_FILE_RANGE_DECL
-#define COPY_FILE_RANGE copy_file_range
+#include <errno.h>
+#include <unistd.h>
-#include <io/copy_file_range-compat.c>
+ssize_t
+copy_file_range (int infd, __off64_t *pinoff,
+ int outfd, __off64_t *poutoff,
+ size_t length, unsigned int flags)
+{
+ __set_errno (ENOSYS);
+ return -1;
+}
+stub_warning (copy_file_range)
+++ /dev/null
-/* Test the fallback implementation of copy_file_range.
- Copyright (C) 2017-2018 Free Software Foundation, Inc.
- This file is part of the GNU C Library.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <http://www.gnu.org/licenses/>. */
-
-/* Get the declaration of the official copy_of_range function. */
-#include <unistd.h>
-
-/* Compile a local version of copy_file_range. */
-#define COPY_FILE_RANGE_DECL static
-#define COPY_FILE_RANGE copy_file_range_compat
-#include <io/copy_file_range-compat.c>
-
-/* Re-use the test, but run it against copy_file_range_compat defined
- above. */
-#define copy_file_range copy_file_range_compat
-#include "tst-copy_file_range.c"
/* Tests for copy_file_range.
- Copyright (C) 2017-2018 Free Software Foundation, Inc.
+ Copyright (C) 2017-2019 Free Software Foundation, Inc.
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
#include <errno.h>
#include <fcntl.h>
#include <inttypes.h>
-#include <libgen.h>
-#include <poll.h>
-#include <sched.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <support/check.h>
-#include <support/namespace.h>
#include <support/support.h>
#include <support/temp_file.h>
#include <support/test-driver.h>
#include <support/xunistd.h>
-#ifdef CLONE_NEWNS
-# include <sys/mount.h>
-#endif
/* Boolean flags which indicate whether to use pointers with explicit
output flags. */
static char *outfile;
static int outfd;
-/* Like the above, but on a different file system. xdevfile can be
- NULL if no suitable file system has been found. */
-static char *xdevfile;
-
/* Input and output offsets. Set according to do_inoff and do_outoff
before the test. The offsets themselves are always set to
zero. */
static off64_t outoff;
static off64_t *poutoff;
-/* These are a collection of copy sizes used in tests. The selection
- takes into account that the fallback implementation uses an
- internal buffer of 8192 bytes. */
+/* These are a collection of copy sizes used in tests. */
enum { maximum_size = 99999 };
static const int typical_sizes[] =
- { 0, 1, 2, 3, 1024, 2048, 4096, 8191, 8192, 8193, 16383, 16384, 16385,
- maximum_size };
+ { 0, 1, 2, 3, 1024, 2048, 4096, 8191, 8192, 8193, maximum_size };
/* The random contents of this array can be used as a pattern to check
for correct write operations. */
/* The size chosen by the test harness. */
static int current_size;
-/* Maximum writable file offset. Updated by find_maximum_offset
- below. */
-static off64_t maximum_offset;
-
-/* Error code when crossing the offset. */
-static int maximum_offset_errno;
-
-/* If true: Writes which cross the limit will fail. If false: Writes
- which cross the limit will result in a partial write. */
-static bool maximum_offset_hard_limit;
-
-/* Fills maximum_offset etc. above. Truncates outfd as a side
- effect. */
-static void
-find_maximum_offset (void)
-{
- xftruncate (outfd, 0);
- if (maximum_offset != 0)
- return;
-
- uint64_t upper = -1;
- upper >>= 1; /* Maximum of off64_t. */
- TEST_VERIFY ((off64_t) upper > 0);
- TEST_VERIFY ((off64_t) (upper + 1) < 0);
- if (lseek64 (outfd, upper, SEEK_SET) >= 0)
- {
- if (write (outfd, "", 1) == 1)
- FAIL_EXIT1 ("created a file larger than the off64_t range");
- }
-
- uint64_t lower = 1024 * 1024; /* A reasonable minimum file size. */
- /* Loop invariant: writing at lower succeeds, writing at upper fails. */
- while (lower + 1 < upper)
- {
- uint64_t middle = (lower + upper) / 2;
- if (test_verbose > 0)
- printf ("info: %s: remaining test range %" PRIu64 " .. %" PRIu64
- ", probe at %" PRIu64 "\n", __func__, lower, upper, middle);
- xftruncate (outfd, 0);
- if (lseek64 (outfd, middle, SEEK_SET) >= 0
- && write (outfd, "", 1) == 1)
- lower = middle;
- else
- upper = middle;
- }
- TEST_VERIFY (lower + 1 == upper);
- maximum_offset = lower;
- printf ("info: maximum writable file offset: %" PRIu64 " (%" PRIx64 ")\n",
- lower, lower);
-
- /* Check that writing at the valid offset actually works. */
- xftruncate (outfd, 0);
- xlseek (outfd, lower, SEEK_SET);
- TEST_COMPARE (write (outfd, "", 1), 1);
-
- /* Cross the boundary with a two-byte write. This can either result
- in a short write, or a failure. */
- xlseek (outfd, lower, SEEK_SET);
- ssize_t ret = write (outfd, " ", 2);
- if (ret < 0)
- {
- maximum_offset_errno = errno;
- maximum_offset_hard_limit = true;
- }
- else
- maximum_offset_hard_limit = false;
-
- /* Check that writing at the next offset actually fails. This also
- obtains the expected errno value. */
- xftruncate (outfd, 0);
- const char *action;
- if (lseek64 (outfd, lower + 1, SEEK_SET) != 0)
- {
- if (write (outfd, "", 1) != -1)
- FAIL_EXIT1 ("write to impossible offset %" PRIu64 " succeeded",
- lower + 1);
- action = "writing";
- int errno_copy = errno;
- if (maximum_offset_hard_limit)
- TEST_COMPARE (errno_copy, maximum_offset_errno);
- else
- maximum_offset_errno = errno_copy;
- }
- else
- {
- action = "seeking";
- maximum_offset_errno = errno;
- }
- printf ("info: %s out of range fails with %m (%d)\n",
- action, maximum_offset_errno);
-
- xftruncate (outfd, 0);
- xlseek (outfd, 0, SEEK_SET);
-}
-
/* Perform a copy of a file. */
static void
simple_file_copy (void)
free (bytes);
}
-/* Test that reading from a pipe willfails. */
-static void
-pipe_as_source (void)
-{
- int pipefds[2];
- xpipe (pipefds);
-
- for (int length = 0; length < 2; ++length)
- {
- if (test_verbose > 0)
- printf ("info: %s: length=%d\n", __func__, length);
-
- /* Make sure that there is something to copy in the pipe. */
- xwrite (pipefds[1], "@", 1);
-
- TEST_COMPARE (copy_file_range (pipefds[0], pinoff, outfd, poutoff,
- length, 0), -1);
- /* Linux 4.10 and later return EINVAL. Older kernels return
- EXDEV. */
- TEST_VERIFY (errno == EINVAL || errno == EXDEV);
- TEST_COMPARE (inoff, 0);
- TEST_COMPARE (outoff, 0);
- TEST_COMPARE (xlseek (outfd, 0, SEEK_CUR), 0);
-
- /* Make sure that nothing was read. */
- char buf = 'A';
- TEST_COMPARE (read (pipefds[0], &buf, 1), 1);
- TEST_COMPARE (buf, '@');
- }
-
- xclose (pipefds[0]);
- xclose (pipefds[1]);
-}
-
-/* Test that writing to a pipe fails. */
-static void
-pipe_as_destination (void)
-{
- /* Make sure that there is something to read in the input file. */
- xwrite (infd, "abc", 3);
- xlseek (infd, 0, SEEK_SET);
-
- int pipefds[2];
- xpipe (pipefds);
-
- for (int length = 0; length < 2; ++length)
- {
- if (test_verbose > 0)
- printf ("info: %s: length=%d\n", __func__, length);
-
- TEST_COMPARE (copy_file_range (infd, pinoff, pipefds[1], poutoff,
- length, 0), -1);
- /* Linux 4.10 and later return EINVAL. Older kernels return
- EXDEV. */
- TEST_VERIFY (errno == EINVAL || errno == EXDEV);
- TEST_COMPARE (inoff, 0);
- TEST_COMPARE (outoff, 0);
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 0);
-
- /* Make sure that nothing was written. */
- struct pollfd pollfd = { .fd = pipefds[0], .events = POLLIN, };
- TEST_COMPARE (poll (&pollfd, 1, 0), 0);
- }
-
- xclose (pipefds[0]);
- xclose (pipefds[1]);
-}
-
-/* Test a write failure after (potentially) writing some bytes.
- Failure occurs near the start of the buffer. */
-static void
-delayed_write_failure_beginning (void)
-{
- /* We need to write something to provoke the error. */
- if (current_size == 0)
- return;
- xwrite (infd, random_data, sizeof (random_data));
- xlseek (infd, 0, SEEK_SET);
-
- /* Write failure near the start. The actual error code varies among
- file systems. */
- find_maximum_offset ();
- off64_t where = maximum_offset;
-
- if (current_size == 1)
- ++where;
- outoff = where;
- if (do_outoff)
- xlseek (outfd, 1, SEEK_SET);
- else
- xlseek (outfd, where, SEEK_SET);
- if (maximum_offset_hard_limit || where > maximum_offset)
- {
- TEST_COMPARE (copy_file_range (infd, pinoff, outfd, poutoff,
- sizeof (random_data), 0), -1);
- TEST_COMPARE (errno, maximum_offset_errno);
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 0);
- TEST_COMPARE (inoff, 0);
- if (do_outoff)
- TEST_COMPARE (xlseek (outfd, 0, SEEK_CUR), 1);
- else
- TEST_COMPARE (xlseek (outfd, 0, SEEK_CUR), where);
- TEST_COMPARE (outoff, where);
- struct stat64 st;
- xfstat (outfd, &st);
- TEST_COMPARE (st.st_size, 0);
- }
- else
- {
- /* The offset is not a hard limit. This means we write one
- byte. */
- TEST_COMPARE (copy_file_range (infd, pinoff, outfd, poutoff,
- sizeof (random_data), 0), 1);
- if (do_inoff)
- {
- TEST_COMPARE (inoff, 1);
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 0);
- }
- else
- {
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 1);
- TEST_COMPARE (inoff, 0);
- }
- if (do_outoff)
- {
- TEST_COMPARE (xlseek (outfd, 0, SEEK_CUR), 1);
- TEST_COMPARE (outoff, where + 1);
- }
- else
- {
- TEST_COMPARE (xlseek (outfd, 0, SEEK_CUR), where + 1);
- TEST_COMPARE (outoff, where);
- }
- struct stat64 st;
- xfstat (outfd, &st);
- TEST_COMPARE (st.st_size, where + 1);
- }
-}
-
-/* Test a write failure after (potentially) writing some bytes.
- Failure occurs near the end of the buffer. */
-static void
-delayed_write_failure_end (void)
-{
- if (current_size <= 1)
- /* This would be same as the first test because there is not
- enough data to write to make a difference. */
- return;
- xwrite (infd, random_data, sizeof (random_data));
- xlseek (infd, 0, SEEK_SET);
-
- find_maximum_offset ();
- off64_t where = maximum_offset - current_size + 1;
- if (current_size == sizeof (random_data))
- /* Otherwise we do not reach the non-writable byte. */
- ++where;
- outoff = where;
- if (do_outoff)
- xlseek (outfd, 1, SEEK_SET);
- else
- xlseek (outfd, where, SEEK_SET);
- ssize_t ret = copy_file_range (infd, pinoff, outfd, poutoff,
- sizeof (random_data), 0);
- if (ret < 0)
- {
- TEST_COMPARE (ret, -1);
- TEST_COMPARE (errno, maximum_offset_errno);
- struct stat64 st;
- xfstat (outfd, &st);
- TEST_COMPARE (st.st_size, 0);
- }
- else
- {
- /* The first copy succeeded. This happens in the emulation
- because the internal buffer of limited size does not
- necessarily cross the off64_t boundary on the first write
- operation. */
- if (test_verbose > 0)
- printf ("info: copy_file_range (%zu) returned %zd\n",
- sizeof (random_data), ret);
- TEST_VERIFY (ret > 0);
- TEST_VERIFY (ret < maximum_size);
- struct stat64 st;
- xfstat (outfd, &st);
- TEST_COMPARE (st.st_size, where + ret);
- if (do_inoff)
- {
- TEST_COMPARE (inoff, ret);
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 0);
- }
- else
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), ret);
-
- char *buffer = xmalloc (ret);
- TEST_COMPARE (pread64 (outfd, buffer, ret, where), ret);
- TEST_VERIFY (memcmp (buffer, random_data, ret) == 0);
- free (buffer);
-
- /* The second copy fails. */
- TEST_COMPARE (copy_file_range (infd, pinoff, outfd, poutoff,
- sizeof (random_data), 0), -1);
- TEST_COMPARE (errno, maximum_offset_errno);
- }
-}
-
-/* Test a write failure across devices. */
-static void
-cross_device_failure (void)
-{
- if (xdevfile == NULL)
- /* Subtest not supported due to missing cross-device file. */
- return;
-
- /* We need something to write. */
- xwrite (infd, random_data, sizeof (random_data));
- xlseek (infd, 0, SEEK_SET);
-
- int xdevfd = xopen (xdevfile, O_RDWR | O_LARGEFILE, 0);
- TEST_COMPARE (copy_file_range (infd, pinoff, xdevfd, poutoff,
- current_size, 0), -1);
- TEST_COMPARE (errno, EXDEV);
- TEST_COMPARE (xlseek (infd, 0, SEEK_CUR), 0);
- struct stat64 st;
- xfstat (xdevfd, &st);
- TEST_COMPARE (st.st_size, 0);
-
- xclose (xdevfd);
-}
-
-/* Try to exercise ENOSPC behavior with a tempfs file system (so that
- we do not have to fill up a regular file system to get the error).
- This function runs in a subprocess, so that we do not change the
- mount namespace of the actual test process. */
-static void
-enospc_failure_1 (void *closure)
-{
-#ifdef CLONE_NEWNS
- support_become_root ();
-
- /* Make sure that we do not alter the file system mounts of the
- parents. */
- if (! support_enter_mount_namespace ())
- {
- printf ("warning: ENOSPC test skipped\n");
- return;
- }
-
- char *mountpoint = closure;
- if (mount ("none", mountpoint, "tmpfs", MS_NODEV | MS_NOEXEC,
- "size=500k") != 0)
- {
- printf ("warning: could not mount tmpfs at %s: %m\n", mountpoint);
- return;
- }
-
- /* The source file must reside on the same file system. */
- char *intmpfsfile = xasprintf ("%s/%s", mountpoint, "in");
- int intmpfsfd = xopen (intmpfsfile, O_RDWR | O_CREAT | O_LARGEFILE, 0600);
- xwrite (intmpfsfd, random_data, sizeof (random_data));
- xlseek (intmpfsfd, 1, SEEK_SET);
- inoff = 1;
-
- char *outtmpfsfile = xasprintf ("%s/%s", mountpoint, "out");
- int outtmpfsfd = xopen (outtmpfsfile, O_RDWR | O_CREAT | O_LARGEFILE, 0600);
-
- /* Fill the file with data until ENOSPC is reached. */
- while (true)
- {
- ssize_t ret = write (outtmpfsfd, random_data, sizeof (random_data));
- if (ret < 0 && errno != ENOSPC)
- FAIL_EXIT1 ("write to %s: %m", outtmpfsfile);
- if (ret < sizeof (random_data))
- break;
- }
- TEST_COMPARE (write (outtmpfsfd, "", 1), -1);
- TEST_COMPARE (errno, ENOSPC);
- off64_t maxsize = xlseek (outtmpfsfd, 0, SEEK_CUR);
- TEST_VERIFY_EXIT (maxsize > sizeof (random_data));
-
- /* Constructed the expected file contents. */
- char *expected = xmalloc (maxsize);
- TEST_COMPARE (pread64 (outtmpfsfd, expected, maxsize, 0), maxsize);
- /* Go back a little, so some bytes can be written. */
- enum { offset = 20000 };
- TEST_VERIFY_EXIT (offset < maxsize);
- TEST_VERIFY_EXIT (offset < sizeof (random_data));
- memcpy (expected + maxsize - offset, random_data + 1, offset);
-
- if (do_outoff)
- {
- outoff = maxsize - offset;
- xlseek (outtmpfsfd, 2, SEEK_SET);
- }
- else
- xlseek (outtmpfsfd, -offset, SEEK_CUR);
-
- /* First call is expected to succeed because we made room for some
- bytes. */
- TEST_COMPARE (copy_file_range (intmpfsfd, pinoff, outtmpfsfd, poutoff,
- maximum_size, 0), offset);
- if (do_inoff)
- {
- TEST_COMPARE (inoff, 1 + offset);
- TEST_COMPARE (xlseek (intmpfsfd, 0, SEEK_CUR), 1);
- }
- else
- TEST_COMPARE (xlseek (intmpfsfd, 0, SEEK_CUR), 1 + offset);
- if (do_outoff)
- {
- TEST_COMPARE (outoff, maxsize);
- TEST_COMPARE (xlseek (outtmpfsfd, 0, SEEK_CUR), 2);
- }
- else
- TEST_COMPARE (xlseek (outtmpfsfd, 0, SEEK_CUR), maxsize);
- struct stat64 st;
- xfstat (outtmpfsfd, &st);
- TEST_COMPARE (st.st_size, maxsize);
- char *actual = xmalloc (st.st_size);
- TEST_COMPARE (pread64 (outtmpfsfd, actual, st.st_size, 0), st.st_size);
- TEST_VERIFY (memcmp (expected, actual, maxsize) == 0);
-
- /* Second call should fail with ENOSPC. */
- TEST_COMPARE (copy_file_range (intmpfsfd, pinoff, outtmpfsfd, poutoff,
- maximum_size, 0), -1);
- TEST_COMPARE (errno, ENOSPC);
-
- /* Offsets should be unchanged. */
- if (do_inoff)
- {
- TEST_COMPARE (inoff, 1 + offset);
- TEST_COMPARE (xlseek (intmpfsfd, 0, SEEK_CUR), 1);
- }
- else
- TEST_COMPARE (xlseek (intmpfsfd, 0, SEEK_CUR), 1 + offset);
- if (do_outoff)
- {
- TEST_COMPARE (outoff, maxsize);
- TEST_COMPARE (xlseek (outtmpfsfd, 0, SEEK_CUR), 2);
- }
- else
- TEST_COMPARE (xlseek (outtmpfsfd, 0, SEEK_CUR), maxsize);
- TEST_COMPARE (xlseek (outtmpfsfd, 0, SEEK_END), maxsize);
- TEST_COMPARE (pread64 (outtmpfsfd, actual, maxsize, 0), maxsize);
- TEST_VERIFY (memcmp (expected, actual, maxsize) == 0);
-
- free (actual);
- free (expected);
-
- xclose (intmpfsfd);
- xclose (outtmpfsfd);
- free (intmpfsfile);
- free (outtmpfsfile);
-
-#else /* !CLONE_NEWNS */
- puts ("warning: ENOSPC test skipped (no mount namespaces)");
-#endif
-}
-
-/* Call enospc_failure_1 in a subprocess. */
-static void
-enospc_failure (void)
-{
- char *mountpoint
- = support_create_temp_directory ("tst-copy_file_range-enospc-");
- support_isolate_in_subprocess (enospc_failure_1, mountpoint);
- free (mountpoint);
-}
-
-/* The target file descriptor must have O_APPEND enabled. */
-static void
-oappend_failure (void)
-{
- /* Add data, to make sure we do not fail because there is
- insufficient input data. */
- xwrite (infd, random_data, current_size);
- xlseek (infd, 0, SEEK_SET);
-
- xclose (outfd);
- outfd = xopen (outfile, O_RDWR | O_APPEND, 0);
- TEST_COMPARE (copy_file_range (infd, pinoff, outfd, poutoff,
- current_size, 0), -1);
- TEST_COMPARE (errno, EBADF);
-}
-
/* Test that a short input file results in a shortened copy. */
static void
short_copy (void)
static struct test_case tests[] =
{
{ "simple_file_copy", simple_file_copy, .sizes = true },
- { "pipe_as_source", pipe_as_source, },
- { "pipe_as_destination", pipe_as_destination, },
- { "delayed_write_failure_beginning", delayed_write_failure_beginning,
- .sizes = true },
- { "delayed_write_failure_end", delayed_write_failure_end, .sizes = true },
- { "cross_device_failure", cross_device_failure, .sizes = true },
- { "enospc_failure", enospc_failure, },
- { "oappend_failure", oappend_failure, .sizes = true },
{ "short_copy", short_copy, .sizes = true },
};
*p = rand () >> 24;
infd = create_temp_file ("tst-copy_file_range-in-", &infile);
- xclose (create_temp_file ("tst-copy_file_range-out-", &outfile));
-
- /* Try to find a different directory from the default input/output
- file. */
+ outfd = create_temp_file ("tst-copy_file_range-out-", &outfile);
{
- struct stat64 instat;
- xfstat (infd, &instat);
- static const char *const candidates[] =
- { NULL, "/var/tmp", "/dev/shm" };
- for (const char *const *c = candidates; c < array_end (candidates); ++c)
- {
- const char *path = *c;
- char *to_free = NULL;
- if (path == NULL)
- {
- to_free = xreadlink ("/proc/self/exe");
- path = dirname (to_free);
- }
-
- struct stat64 cstat;
- xstat (path, &cstat);
- if (cstat.st_dev == instat.st_dev)
- {
- free (to_free);
- continue;
- }
-
- printf ("info: using alternate temporary files directory: %s\n", path);
- xdevfile = xasprintf ("%s/tst-copy_file_range-xdev-XXXXXX", path);
- free (to_free);
- break;
- }
- if (xdevfile != NULL)
+ ssize_t ret = copy_file_range (infd, NULL, outfd, NULL, 0, 0);
+ if (ret != 0)
{
- int xdevfd = mkstemp (xdevfile);
- if (xdevfd < 0)
- FAIL_EXIT1 ("mkstemp (\"%s\"): %m", xdevfile);
- struct stat64 xdevst;
- xfstat (xdevfd, &xdevst);
- TEST_VERIFY (xdevst.st_dev != instat.st_dev);
- add_temp_file (xdevfile);
- xclose (xdevfd);
+ if (errno == ENOSYS)
+ FAIL_UNSUPPORTED ("copy_file_range is not support on this system");
+ FAIL_EXIT1 ("copy_file_range probing call: %m");
}
- else
- puts ("warning: no alternate directory on different file system found");
}
xclose (infd);
+ xclose (outfd);
for (do_inoff = 0; do_inoff < 2; ++do_inoff)
for (do_outoff = 0; do_outoff < 2; ++do_outoff)
free (infile);
free (outfile);
- free (xdevfile);
return 0;
}
bug-memstream1 bug-wmemstream1 \
tst-setvbuf1 tst-popen1 tst-fgetwc bug-wsetpos tst-fseek \
tst-fwrite-error tst-ftell-partial-wide tst-ftell-active-handler \
- tst-ftell-append tst-fputws tst-bz22415 tst-fgetc-after-eof
+ tst-ftell-append tst-fputws tst-bz22415 tst-fgetc-after-eof \
+ tst-wfile-sync
tests-internal = tst-vtables tst-vtables-interposed tst-readline
# Add test-fopenloc only if shared library is enabled since it depends on
# shared localedata objects.
tests += tst-fopenloc
+# Add tst-bz24228 only if shared library is enabled since it can never meet its
+# objective with static linking because the relevant code just is not there.
+tests += tst-bz24228
endif
test-srcs = test-freopen
CFLAGS-tst_putwc.c += -DOBJPFX=\"$(objpfx)\"
+LDFLAGS-tst-bz24228 = -Wl,--version-script=tst-bz24228.map
+
tst_wprintf2-ARGS = "Some Text"
test-fmemopen-ENV = MALLOC_TRACE=$(objpfx)test-fmemopen.mtrace
tst-fopenloc-ENV = MALLOC_TRACE=$(objpfx)tst-fopenloc.mtrace
tst-bz22415-ENV = MALLOC_TRACE=$(objpfx)tst-bz22415.mtrace
+tst-bz24228-ENV = MALLOC_TRACE=$(objpfx)tst-bz24228.mtrace
generated += test-fmemopen.mtrace test-fmemopen.check
generated += tst-fopenloc.mtrace tst-fopenloc.check
aux := fileops genops stdfiles stdio strops
ifeq ($(build-shared),yes)
+generated += tst-bz24228.mtrace tst-bz24228.check
aux += oldfileops oldstdfiles
endif
ifeq (yes,$(build-shared))
# Run tst-fopenloc-cmp.out and tst-openloc-mem.out only if shared
# library is enabled since they depend on tst-fopenloc.out.
-tests-special += $(objpfx)tst-fopenloc-cmp.out $(objpfx)tst-fopenloc-mem.out
+tests-special += $(objpfx)tst-fopenloc-cmp.out $(objpfx)tst-fopenloc-mem.out \
+ $(objpfx)tst-bz24228-mem.out
endif
endif
$(objpfx)tst-ungetwc2.out: $(gen-locales)
$(objpfx)tst-widetext.out: $(gen-locales)
$(objpfx)tst_wprintf2.out: $(gen-locales)
+$(objpfx)tst-wfile-sync.out: $(gen-locales)
endif
$(objpfx)test-freopen.out: test-freopen.sh $(objpfx)test-freopen
$(objpfx)tst-bz22415-mem.out: $(objpfx)tst-bz22415.out
$(common-objpfx)malloc/mtrace $(objpfx)tst-bz22415.mtrace > $@; \
$(evaluate-test)
+
+$(objpfx)tst-bz24228-mem.out: $(objpfx)tst-bz24228.out
+ $(common-objpfx)malloc/mtrace $(objpfx)tst-bz24228.mtrace > $@; \
+ $(evaluate-test)
for (fp = (FILE *) _IO_list_all; fp; fp = fp->_chain)
{
+ int legacy = 0;
+
+#if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_1)
+ if (__glibc_unlikely (_IO_vtable_offset (fp) != 0))
+ legacy = 1;
+#endif
+
if (! (fp->_flags & _IO_UNBUFFERED)
/* Iff stream is un-orientated, it wasn't used. */
- && fp->_mode != 0)
+ && (legacy || fp->_mode != 0))
{
#ifdef _IO_MTSAFE_IO
int cnt;
__sched_yield ();
#endif
- if (! dealloc_buffers && !(fp->_flags & _IO_USER_BUF))
+ if (! legacy && ! dealloc_buffers && !(fp->_flags & _IO_USER_BUF))
{
fp->_flags |= _IO_USER_BUF;
_IO_SETBUF (fp, NULL, 0);
- if (fp->_mode > 0)
+ if (! legacy && fp->_mode > 0)
_IO_wsetb (fp, NULL, NULL, 0);
#ifdef _IO_MTSAFE_IO
/* Make sure that never again the wide char functions can be
used. */
- fp->_mode = -1;
+ if (! legacy)
+ fp->_mode = -1;
}
#ifdef _IO_MTSAFE_IO
--- /dev/null
+/* BZ #24228 check for memory corruption in legacy libio
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
+
+#include <mcheck.h>
+#include <support/test-driver.h>
+
+static int
+do_test (void)
+{
+ mtrace ();
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+# Hide the symbol from libc.so.6 to switch to the libio/oldfileops.c
+# implementation when it is available for the architecture.
+{
+ local: _IO_stdin_used;
+};
return 0;
}
+#define TIMEOUT 100
#define PREPARE prepare
#include <support/test-driver.c>
--- /dev/null
+/* Test that _IO_wfile_sync does not crash (bug 20568).
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <locale.h>
+#include <stdio.h>
+#include <wchar.h>
+#include <support/check.h>
+#include <support/xunistd.h>
+
+static int
+do_test (void)
+{
+ TEST_VERIFY_EXIT (setlocale (LC_ALL, "de_DE.UTF-8") != NULL);
+ /* Fill the stdio buffer and advance the read pointer. */
+ TEST_VERIFY_EXIT (fgetwc (stdin) != WEOF);
+ /* This calls _IO_wfile_sync, it should not crash. */
+ TEST_VERIFY_EXIT (setvbuf (stdin, NULL, _IONBF, 0) == 0);
+ /* Verify that the external file offset has been synchronized. */
+ TEST_COMPARE (xlseek (0, 0, SEEK_CUR), 1);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+This is a test of _IO_wfile_sync.
generate the wide characters up to the current reading
position. */
int nread;
-
+ size_t wnread = (fp->_wide_data->_IO_read_ptr
+ - fp->_wide_data->_IO_read_base);
fp->_wide_data->_IO_state = fp->_wide_data->_IO_last_state;
nread = (*cv->__codecvt_do_length) (cv, &fp->_wide_data->_IO_state,
fp->_IO_read_base,
- fp->_IO_read_end, delta);
+ fp->_IO_read_end, wnread);
fp->_IO_read_ptr = fp->_IO_read_base + nread;
delta = -(fp->_IO_read_end - fp->_IO_read_base - nread);
}
t_fmt_ampm "%p%I<U6642>%M<U5206>%S<U79D2>"
-era "+:2:1990//01//01:+*:<U5E73><U6210>:%EC%Ey<U5E74>";/
+era "+:2:2020//01//01:+*:<U4EE4><U548C>:%EC%Ey<U5E74>";/
+ "+:1:2019//05//01:2019//12//31:<U4EE4><U548C>:%EC<U5143><U5E74>";/
+ "+:2:1990//01//01:2019//04//30:<U5E73><U6210>:%EC%Ey<U5E74>";/
"+:1:1989//01//08:1989//12//31:<U5E73><U6210>:%EC<U5143><U5E74>";/
"+:2:1927//01//01:1989//01//07:<U662D><U548C>:%EC%Ey<U5E74>";/
"+:1:1926//12//25:1926//12//31:<U662D><U548C>:%EC<U5143><U5E74>";/
END LC_NUMERIC
LC_TIME
-abday "sab";"ata";/
+abday "sap";"ata";/
"mar";"pin";/
"sis";"tal";/
"arf"
-day "sabaat";/
+day "sapaat";/
"ataasinngorneq";/
"marlunngorneq";/
"pingasunngorneq";/
tst-malloc_info \
tst-malloc-too-large \
tst-malloc-stats-cancellation \
+ tst-tcfree1 tst-tcfree2 tst-tcfree3 \
tests-static := \
tst-interpose-static-nothread \
static void
munmap_chunk (mchunkptr p)
{
+ size_t pagesize = GLRO (dl_pagesize);
INTERNAL_SIZE_T size = chunksize (p);
assert (chunk_is_mmapped (p));
if (DUMPED_MAIN_ARENA_CHUNK (p))
return;
+ uintptr_t mem = (uintptr_t) chunk2mem (p);
uintptr_t block = (uintptr_t) p - prev_size (p);
size_t total_size = prev_size (p) + size;
/* Unfortunately we have to do the compilers job by hand here. Normally
page size. But gcc does not recognize the optimization possibility
(in the moment at least) so we combine the two values into one before
the bit test. */
- if (__builtin_expect (((block | total_size) & (GLRO (dl_pagesize) - 1)) != 0, 0))
+ if (__glibc_unlikely ((block | total_size) & (pagesize - 1)) != 0
+ || __glibc_unlikely (!powerof2 (mem & (pagesize - 1))))
malloc_printerr ("munmap_chunk(): invalid pointer");
atomic_decrement (&mp_.n_mmaps);
char *cp;
assert (chunk_is_mmapped (p));
- assert (((size + offset) & (GLRO (dl_pagesize) - 1)) == 0);
+
+ uintptr_t block = (uintptr_t) p - offset;
+ uintptr_t mem = (uintptr_t) chunk2mem(p);
+ size_t total_size = offset + size;
+ if (__glibc_unlikely ((block | total_size) & (pagesize - 1)) != 0
+ || __glibc_unlikely (!powerof2 (mem & (pagesize - 1))))
+ malloc_printerr("mremap_chunk(): invalid pointer");
/* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
new_size = ALIGN_UP (new_size + offset + SIZE_SZ, pagesize);
/* No need to remap if the number of pages does not change. */
- if (size + offset == new_size)
+ if (total_size == new_size)
return p;
- cp = (char *) __mremap ((char *) p - offset, size + offset, new_size,
+ cp = (char *) __mremap ((char *) block, total_size, new_size,
MREMAP_MAYMOVE);
if (cp == MAP_FAILED)
typedef struct tcache_entry
{
struct tcache_entry *next;
+ /* This field exists to detect double frees. */
+ struct tcache_perthread_struct *key;
} tcache_entry;
/* There is one of these for each thread, which contains the
tcache_entry *entries[TCACHE_MAX_BINS];
} tcache_perthread_struct;
+#define MAX_TCACHE_COUNT 127 /* Maximum value of counts[] entries. */
+
static __thread bool tcache_shutting_down = false;
static __thread tcache_perthread_struct *tcache = NULL;
{
tcache_entry *e = (tcache_entry *) chunk2mem (chunk);
assert (tc_idx < TCACHE_MAX_BINS);
+
+ /* Mark this chunk as "in the tcache" so the test in _int_free will
+ detect a double free. */
+ e->key = tcache;
+
e->next = tcache->entries[tc_idx];
tcache->entries[tc_idx] = e;
++(tcache->counts[tc_idx]);
assert (tcache->entries[tc_idx] > 0);
tcache->entries[tc_idx] = e->next;
--(tcache->counts[tc_idx]);
+ e->key = NULL;
return (void *) e;
}
while ((victim = unsorted_chunks (av)->bk) != unsorted_chunks (av))
{
bck = victim->bk;
- if (__builtin_expect (chunksize_nomask (victim) <= 2 * SIZE_SZ, 0)
- || __builtin_expect (chunksize_nomask (victim)
- > av->system_mem, 0))
- malloc_printerr ("malloc(): memory corruption");
size = chunksize (victim);
+ mchunkptr next = chunk_at_offset (victim, size);
+
+ if (__glibc_unlikely (size <= 2 * SIZE_SZ)
+ || __glibc_unlikely (size > av->system_mem))
+ malloc_printerr ("malloc(): invalid size (unsorted)");
+ if (__glibc_unlikely (chunksize_nomask (next) < 2 * SIZE_SZ)
+ || __glibc_unlikely (chunksize_nomask (next) > av->system_mem))
+ malloc_printerr ("malloc(): invalid next size (unsorted)");
+ if (__glibc_unlikely ((prev_size (next) & ~(SIZE_BITS)) != size))
+ malloc_printerr ("malloc(): mismatching next->prev_size (unsorted)");
+ if (__glibc_unlikely (bck->fd != victim)
+ || __glibc_unlikely (victim->fd != unsorted_chunks (av)))
+ malloc_printerr ("malloc(): unsorted double linked list corrupted");
+ if (__glibc_unlikely (prev_inuse (next)))
+ malloc_printerr ("malloc(): invalid next->prev_inuse (unsorted)");
/*
If a small request, try to use last remainder if it is the
{
victim->fd_nextsize = fwd;
victim->bk_nextsize = fwd->bk_nextsize;
+ if (__glibc_unlikely (fwd->bk_nextsize->fd_nextsize != fwd))
+ malloc_printerr ("malloc(): largebin double linked list corrupted (nextsize)");
fwd->bk_nextsize = victim;
victim->bk_nextsize->fd_nextsize = victim;
}
bck = fwd->bk;
+ if (bck->fd != fwd)
+ malloc_printerr ("malloc(): largebin double linked list corrupted (bk)");
}
}
else
victim = av->top;
size = chunksize (victim);
+ if (__glibc_unlikely (size > av->system_mem))
+ malloc_printerr ("malloc(): corrupted top size");
+
if ((unsigned long) (size) >= (unsigned long) (nb + MINSIZE))
{
remainder_size = size - nb;
#if USE_TCACHE
{
size_t tc_idx = csize2tidx (size);
-
- if (tcache
- && tc_idx < mp_.tcache_bins
- && tcache->counts[tc_idx] < mp_.tcache_count)
+ if (tcache != NULL && tc_idx < mp_.tcache_bins)
{
- tcache_put (p, tc_idx);
- return;
+ /* Check to see if it's already in the tcache. */
+ tcache_entry *e = (tcache_entry *) chunk2mem (p);
+
+ /* This test succeeds on double free. However, we don't 100%
+ trust it (it also matches random payload data at a 1 in
+ 2^<size_t> chance), so verify it's not an unlikely
+ coincidence before aborting. */
+ if (__glibc_unlikely (e->key == tcache))
+ {
+ tcache_entry *tmp;
+ LIBC_PROBE (memory_tcache_double_free, 2, e, tc_idx);
+ for (tmp = tcache->entries[tc_idx];
+ tmp;
+ tmp = tmp->next)
+ if (tmp == e)
+ malloc_printerr ("free(): double free detected in tcache 2");
+ /* If we get here, it was a coincidence. We've wasted a
+ few cycles, but don't abort. */
+ }
+
+ if (tcache->counts[tc_idx] < mp_.tcache_count)
+ {
+ tcache_put (p, tc_idx);
+ return;
+ }
}
}
#endif
prevsize = prev_size (p);
size += prevsize;
p = chunk_at_offset(p, -((long) prevsize));
+ if (__glibc_unlikely (chunksize(p) != prevsize))
+ malloc_printerr ("corrupted size vs. prev_size while consolidating");
unlink(av, p, bck, fwd);
}
prevsize = prev_size (p);
size += prevsize;
p = chunk_at_offset(p, -((long) prevsize));
+ if (__glibc_unlikely (chunksize(p) != prevsize))
+ malloc_printerr ("corrupted size vs. prev_size in fastbins");
unlink(av, p, bck, fwd);
}
mchunkptr bck; /* misc temp for linking */
mchunkptr fwd; /* misc temp for linking */
- unsigned long copysize; /* bytes to copy */
- unsigned int ncopies; /* INTERNAL_SIZE_T words to copy */
- INTERNAL_SIZE_T* s; /* copy source */
- INTERNAL_SIZE_T* d; /* copy destination */
-
/* oldmem size */
if (__builtin_expect (chunksize_nomask (oldp) <= 2 * SIZE_SZ, 0)
|| __builtin_expect (oldsize >= av->system_mem, 0))
}
else
{
- /*
- Unroll copy of <= 36 bytes (72 if 8byte sizes)
- We know that contents have an odd number of
- INTERNAL_SIZE_T-sized words; minimally 3.
- */
-
- copysize = oldsize - SIZE_SZ;
- s = (INTERNAL_SIZE_T *) (chunk2mem (oldp));
- d = (INTERNAL_SIZE_T *) (newmem);
- ncopies = copysize / sizeof (INTERNAL_SIZE_T);
- assert (ncopies >= 3);
-
- if (ncopies > 9)
- memcpy (d, s, copysize);
-
- else
- {
- *(d + 0) = *(s + 0);
- *(d + 1) = *(s + 1);
- *(d + 2) = *(s + 2);
- if (ncopies > 4)
- {
- *(d + 3) = *(s + 3);
- *(d + 4) = *(s + 4);
- if (ncopies > 6)
- {
- *(d + 5) = *(s + 5);
- *(d + 6) = *(s + 6);
- if (ncopies > 8)
- {
- *(d + 7) = *(s + 7);
- *(d + 8) = *(s + 8);
- }
- }
- }
- }
-
+ memcpy (newmem, chunk2mem (oldp), oldsize - SIZE_SZ);
_int_free (av, oldp, 1);
check_inuse_chunk (av, newp);
return chunk2mem (newp);
__always_inline
do_set_tcache_count (size_t value)
{
- LIBC_PROBE (memory_tunable_tcache_count, 2, value, mp_.tcache_count);
- mp_.tcache_count = value;
+ if (value <= MAX_TCACHE_COUNT)
+ {
+ LIBC_PROBE (memory_tunable_tcache_count, 2, value, mp_.tcache_count);
+ mp_.tcache_count = value;
+ }
return 1;
}
--- /dev/null
+/* Test that malloc tcache catches double free.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <errno.h>
+#include <error.h>
+#include <limits.h>
+#include <malloc.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <sys/signal.h>
+
+static int
+do_test (void)
+{
+ /* Do one allocation of any size that fits in tcache. */
+ char * volatile x = malloc (32);
+
+ free (x); // puts in tcache
+ free (x); // should abort
+
+ printf("FAIL: tcache double free not detected\n");
+ return 1;
+}
+
+#define TEST_FUNCTION do_test
+#define EXPECTED_SIGNAL SIGABRT
+#include <support/test-driver.c>
--- /dev/null
+/* Test that malloc tcache catches double free.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <errno.h>
+#include <error.h>
+#include <limits.h>
+#include <malloc.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <sys/signal.h>
+
+static int
+do_test (void)
+{
+ char * volatile ptrs[20];
+ int i;
+
+ /* Allocate enough small chunks so that when we free them all, the tcache
+ is full, and the first one we freed is at the end of its linked list. */
+#define COUNT 20
+ for (i=0; i<COUNT; i++)
+ ptrs[i] = malloc (20);
+ for (i=0; i<COUNT; i++)
+ free (ptrs[i]);
+ free (ptrs[0]);
+
+ printf("FAIL: tcache double free\n");
+ return 1;
+}
+
+#define TEST_FUNCTION do_test
+#define EXPECTED_SIGNAL SIGABRT
+#include <support/test-driver.c>
--- /dev/null
+/* Test that malloc tcache catches double free.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <malloc.h>
+#include <string.h>
+
+/* Prevent GCC from optimizing away any malloc/free pairs. */
+#pragma GCC optimize ("O0")
+
+static int
+do_test (void)
+{
+ /* Do two allocation of any size that fit in tcache, and one that
+ doesn't. */
+ int ** volatile a = malloc (32);
+ int ** volatile b = malloc (32);
+ /* This is just under the mmap threshold. */
+ int ** volatile c = malloc (127 * 1024);
+
+ /* The invalid "tcache bucket" we might dereference will likely end
+ up somewhere within this memory block, so make all the accidental
+ "next" pointers cause segfaults. BZ #23907. */
+ memset (c, 0xff, 127 * 1024);
+
+ free (a); // puts in tcache
+
+ /* A is now free and contains the key we use to detect in-tcache.
+ Copy the key to the other chunks. */
+ memcpy (b, a, 32);
+ memcpy (c, a, 32);
+
+ /* This free tests the "are we in the tcache already" loop with a
+ VALID bin but "coincidental" matching key. */
+ free (b); // should NOT abort
+ /* This free tests the "is it a valid tcache bin" test. */
+ free (c); // should NOT abort
+
+ return 0;
+}
+
+#include <support/test-driver.c>
TEST_fff_f (fma, plus_infty, plus_infty, -min_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, plus_infty, min_subnorm_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, plus_infty, -min_subnorm_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, plus_infty, plus_infty, max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, plus_infty, plus_infty, -max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
+ TEST_fff_f (fma, plus_infty, plus_infty, max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
+ TEST_fff_f (fma, plus_infty, plus_infty, -max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
TEST_fff_f (fma, plus_infty, minus_infty, plus_zero, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, minus_infty, minus_zero, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, minus_infty, min_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, minus_infty, -min_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, minus_infty, min_subnorm_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, plus_infty, minus_infty, -min_subnorm_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, plus_infty, minus_infty, max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, plus_infty, minus_infty, -max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
+ TEST_fff_f (fma, plus_infty, minus_infty, max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
+ TEST_fff_f (fma, plus_infty, minus_infty, -max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
TEST_fff_f (fma, minus_infty, plus_infty, plus_zero, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, plus_infty, minus_zero, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, plus_infty, min_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, plus_infty, -min_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, plus_infty, min_subnorm_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, plus_infty, -min_subnorm_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, minus_infty, plus_infty, max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, minus_infty, plus_infty, -max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
+ TEST_fff_f (fma, minus_infty, plus_infty, max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
+ TEST_fff_f (fma, minus_infty, plus_infty, -max_value, minus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
TEST_fff_f (fma, minus_infty, minus_infty, plus_zero, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, minus_infty, minus_zero, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, minus_infty, min_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, minus_infty, -min_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, minus_infty, min_subnorm_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
TEST_fff_f (fma, minus_infty, minus_infty, -min_subnorm_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, minus_infty, minus_infty, max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
- TEST_fff_f (fma, minus_infty, minus_infty, -max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED),
+ TEST_fff_f (fma, minus_infty, minus_infty, max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
+ TEST_fff_f (fma, minus_infty, minus_infty, -max_value, plus_infty, NO_INEXACT_EXCEPTION|ERRNO_UNCHANGED|XFAIL_ROUNDING_IBM128_LIBGCC),
AUTO_TESTS_fff_f (fma),
};
tst-preadvwritev tst-preadvwritev64 tst-makedev tst-empty \
tst-preadvwritev2 tst-preadvwritev64v2
+# Tests which need libdl.
+ifeq (yes,$(build-shared))
+tests += tst-gethostid
+endif
+
tests-internal := tst-atomic tst-atomic-long tst-allocate_once
tests-static := tst-empty
$(objpfx)tst-allocate_once-mem.out: $(objpfx)tst-allocate_once.out
$(common-objpfx)malloc/mtrace $(objpfx)tst-allocate_once.mtrace > $@; \
$(evaluate-test)
+
+$(objpfx)tst-gethostid: $(libdl)
va_start (args, message);
error_tail (status, errnum, message, args);
+ va_end (args);
#ifdef _LIBC
_IO_funlockfile (stderr);
va_start (args, message);
error_tail (status, errnum, message, args);
+ va_end (args);
#ifdef _LIBC
_IO_funlockfile (stderr);
--- /dev/null
+/* Basic test for gethostid.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <gnu/lib-names.h>
+#include <nss.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/namespace.h>
+#include <support/support.h>
+#include <support/temp_file.h>
+#include <support/xdlfcn.h>
+#include <support/xstdio.h>
+#include <support/xunistd.h>
+#include <unistd.h>
+
+/* Initial test is run outside a chroot, to increase the likelihood of
+ success. */
+static void
+outside_chroot (void *closure)
+{
+ long id = gethostid ();
+ printf ("info: host ID outside chroot: 0x%lx\n", id);
+}
+
+/* The same, but this time perform a chroot operation. */
+static void
+in_chroot (void *closure)
+{
+ const char *chroot_path = closure;
+ xchroot (chroot_path);
+ long id = gethostid ();
+ printf ("info: host ID in chroot: 0x%lx\n", id);
+}
+
+static int
+do_test (void)
+{
+ support_isolate_in_subprocess (outside_chroot, NULL);
+
+ /* Now run the test inside a chroot. */
+ support_become_root ();
+ if (!support_can_chroot ())
+ /* Cannot perform further tests. */
+ return 0;
+
+ /* Only use nss_files. */
+ __nss_configure_lookup ("hosts", "files");
+
+ /* Load the DSO outside of the chroot. */
+ xdlopen (LIBNSS_FILES_SO, RTLD_LAZY);
+
+ char *chroot_dir = support_create_temp_directory ("tst-gethostid-");
+ support_isolate_in_subprocess (in_chroot, chroot_dir);
+
+ /* Tests with /etc/hosts in the chroot. */
+ {
+ char *path = xasprintf ("%s/etc", chroot_dir);
+ add_temp_file (path);
+ xmkdir (path, 0777);
+ free (path);
+ path = xasprintf ("%s/etc/hosts", chroot_dir);
+ add_temp_file (path);
+
+ FILE *fp = xfopen (path, "w");
+ xfclose (fp);
+ printf ("info: chroot test with an empty /etc/hosts file\n");
+ support_isolate_in_subprocess (in_chroot, chroot_dir);
+
+ char hostname[1024];
+ int ret = gethostname (hostname, sizeof (hostname));
+ if (ret < 0)
+ printf ("warning: invalid result from gethostname: %d\n", ret);
+ else if (strlen (hostname) == 0)
+ puts ("warning: gethostname returned empty string");
+ else
+ {
+ printf ("info: chroot test with IPv6 address in /etc/hosts for: %s\n",
+ hostname);
+ fp = xfopen (path, "w");
+ /* Use an IPv6 address to induce another lookup failure. */
+ fprintf (fp, "2001:db8::1 %s\n", hostname);
+ xfclose (fp);
+ support_isolate_in_subprocess (in_chroot, chroot_dir);
+ }
+ free (path);
+ }
+ free (chroot_dir);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
#include <limits.h>
#include <support/check.h>
-static void
-do_test_with_invalid_flags (void)
-{
#ifndef RWF_HIPRI
# define RWF_HIPRI 0
#endif
#endif
#define RWF_SUPPORTED (RWF_HIPRI | RWF_DSYNC | RWF_SYNC | RWF_NOWAIT \
| RWF_APPEND)
+
+static void
+do_test_with_invalid_fd (void)
+{
+ char buf[256];
+ struct iovec iov = { buf, sizeof buf };
+
+ /* Check with flag being 0 to use the fallback code which calls pwritev
+ or writev. */
+ TEST_VERIFY (preadv2 (-1, &iov, 1, -1, 0) == -1);
+ TEST_COMPARE (errno, EBADF);
+ TEST_VERIFY (pwritev2 (-1, &iov, 1, -1, 0) == -1);
+ TEST_COMPARE (errno, EBADF);
+
+ /* Same tests as before but with flags being different than 0. Since
+ there is no emulation for any flag value, fallback code returns
+ ENOTSUP. This is different running on a kernel with preadv2/pwritev2
+ support, where EBADF is returned). */
+ TEST_VERIFY (preadv2 (-1, &iov, 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EBADF || errno == ENOTSUP);
+ TEST_VERIFY (pwritev2 (-1, &iov, 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EBADF || errno == ENOTSUP);
+}
+
+static void
+do_test_with_invalid_iov (void)
+{
+ {
+ char buf[256];
+ struct iovec iov;
+
+ iov.iov_base = buf;
+ iov.iov_len = (size_t)SSIZE_MAX + 1;
+
+ TEST_VERIFY (preadv2 (temp_fd, &iov, 1, 0, 0) == -1);
+ TEST_COMPARE (errno, EINVAL);
+ TEST_VERIFY (pwritev2 (temp_fd, &iov, 1, 0, 0) == -1);
+ TEST_COMPARE (errno, EINVAL);
+
+ /* Same as for invalid file descriptor tests, emulation fallback
+ first checks for flag value and return ENOTSUP. */
+ TEST_VERIFY (preadv2 (temp_fd, &iov, 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EINVAL || errno == ENOTSUP);
+ TEST_VERIFY (pwritev2 (temp_fd, &iov, 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EINVAL || errno == ENOTSUP);
+ }
+
+ {
+ /* An invalid iovec buffer should trigger an invalid memory access
+ or an error (Linux for instance returns EFAULT). */
+ struct iovec iov[IOV_MAX+1] = { 0 };
+
+ TEST_VERIFY (preadv2 (temp_fd, iov, IOV_MAX + 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EINVAL || errno == ENOTSUP);
+ TEST_VERIFY (pwritev2 (temp_fd, iov, IOV_MAX + 1, 0, RWF_HIPRI) == -1);
+ TEST_VERIFY (errno == EINVAL || errno == ENOTSUP);
+ }
+}
+
+static void
+do_test_with_invalid_flags (void)
+{
/* Set the next bit from the mask of all supported flags. */
int invalid_flag = RWF_SUPPORTED != 0 ? __builtin_clz (RWF_SUPPORTED) : 2;
invalid_flag = 0x1 << ((sizeof (int) * CHAR_BIT) - invalid_flag);
{
do_test_with_invalid_flags ();
do_test_without_offset ();
+ do_test_with_invalid_fd ();
+ do_test_with_invalid_iov ();
return do_test_with_offset (0);
}
{
do_test_with_invalid_flags ();
do_test_without_offset ();
+ do_test_with_invalid_fd ();
+ do_test_with_invalid_iov ();
return do_test_with_offset (0);
}
tests = tst-attr1 tst-attr2 tst-attr3 tst-default-attr \
tst-mutex1 tst-mutex2 tst-mutex3 tst-mutex4 tst-mutex5 tst-mutex6 \
- tst-mutex7 tst-mutex9 tst-mutex5a tst-mutex7a tst-mutex7robust \
- tst-mutexpi1 tst-mutexpi2 tst-mutexpi3 tst-mutexpi4 tst-mutexpi5 \
- tst-mutexpi5a tst-mutexpi6 tst-mutexpi7 tst-mutexpi7a \
+ tst-mutex7 tst-mutex9 tst-mutex10 tst-mutex5a tst-mutex7a \
+ tst-mutex7robust tst-mutexpi1 tst-mutexpi2 tst-mutexpi3 tst-mutexpi4 \
+ tst-mutexpi5 tst-mutexpi5a tst-mutexpi6 tst-mutexpi7 tst-mutexpi7a \
tst-mutexpi9 \
tst-spin1 tst-spin2 tst-spin3 tst-spin4 \
tst-cond1 tst-cond2 tst-cond3 tst-cond4 tst-cond5 tst-cond6 tst-cond7 \
tst-minstack-throw \
tst-cnd-basic tst-mtx-trylock tst-cnd-broadcast \
tst-cnd-timedwait tst-thrd-detach tst-mtx-basic tst-thrd-sleep \
- tst-mtx-recursive tst-tss-basic tst-call-once tst-mtx-timedlock
+ tst-mtx-recursive tst-tss-basic tst-call-once tst-mtx-timedlock \
+ tst-rwlock-pwn \
+ tst-rwlock-tryrdlock-stall tst-rwlock-trywrlock-stall \
+ tst-unwind-thread
tests-internal := tst-rwlock19 tst-rwlock20 \
tst-sem11 tst-sem12 tst-sem13 \
tst-cleanupx0 tst-cleanupx1 tst-cleanupx2 tst-cleanupx3 tst-cleanupx4 \
tst-oncex3 tst-oncex4
ifeq ($(build-shared),yes)
-tests += tst-atfork2 tst-tls4 tst-_res1 tst-fini1 tst-compat-forwarder
+tests += tst-atfork2 tst-tls4 tst-_res1 tst-fini1 tst-compat-forwarder \
+ tst-audit-threads
tests-internal += tst-tls3 tst-tls3-malloc tst-tls5 tst-stackguard1
tests-nolibpthread += tst-fini1
ifeq ($(have-z-execstack),yes)
tst-tls5mod tst-tls5moda tst-tls5modb tst-tls5modc \
tst-tls5modd tst-tls5mode tst-tls5modf tst-stack4mod \
tst-_res1mod1 tst-_res1mod2 tst-execstack-mod tst-fini1mod \
- tst-join7mod tst-compat-forwarder-mod
+ tst-join7mod tst-compat-forwarder-mod tst-audit-threads-mod1 \
+ tst-audit-threads-mod2
extra-test-objs += $(addsuffix .os,$(strip $(modules-names))) \
tst-cleanup4aux.o tst-cleanupx4aux.o
test-extras += tst-cleanup4aux tst-cleanupx4aux
$(objpfx)tst-compat-forwarder: $(objpfx)tst-compat-forwarder-mod.so
+tst-mutex10-ENV = GLIBC_TUNABLES=glibc.elision.enable=1
+
+# Protect against a build using -Wl,-z,now.
+LDFLAGS-tst-audit-threads-mod1.so = -Wl,-z,lazy
+LDFLAGS-tst-audit-threads-mod2.so = -Wl,-z,lazy
+LDFLAGS-tst-audit-threads = -Wl,-z,lazy
+$(objpfx)tst-audit-threads: $(objpfx)tst-audit-threads-mod2.so
+$(objpfx)tst-audit-threads.out: $(objpfx)tst-audit-threads-mod1.so
+tst-audit-threads-ENV = LD_AUDIT=$(objpfx)tst-audit-threads-mod1.so
+
+CFLAGS-tst-unwind-thread.c += -funwind-tables
+
# The tests here better do not run in parallel
ifneq ($(filter %tests,$(MAKECMDGOALS)),)
.NOTPARALLEL:
};
#define PTHREAD_MUTEX_PSHARED_BIT 128
+/* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
#define PTHREAD_MUTEX_TYPE(m) \
- ((m)->__data.__kind & 127)
+ (atomic_load_relaxed (&((m)->__data.__kind)) & 127)
/* Don't include NO_ELISION, as that type is always the same
as the underlying lock type. */
#define PTHREAD_MUTEX_TYPE_ELISION(m) \
- ((m)->__data.__kind & (127|PTHREAD_MUTEX_ELISION_NP))
+ (atomic_load_relaxed (&((m)->__data.__kind)) \
+ & (127 | PTHREAD_MUTEX_ELISION_NP))
#if LLL_PRIVATE == 0 && LLL_SHARED == 128
# define PTHREAD_MUTEX_PSHARED(m) \
- ((m)->__data.__kind & 128)
+ (atomic_load_relaxed (&((m)->__data.__kind)) & 128)
#else
# define PTHREAD_MUTEX_PSHARED(m) \
- (((m)->__data.__kind & 128) ? LLL_SHARED : LLL_PRIVATE)
+ ((atomic_load_relaxed (&((m)->__data.__kind)) & 128) \
+ ? LLL_SHARED : LLL_PRIVATE)
#endif
/* The kernel when waking robust mutexes on exit never uses
{
/* There is still a waiter after spinning. Set the wake-request
flag and block. Relaxed MO is fine because this is just about
- this futex word. */
- r = atomic_fetch_or_relaxed (cond->__data.__g_refs + g1, 1);
+ this futex word.
+
+ Update r to include the set wake-request flag so that the upcoming
+ futex_wait only blocks if the flag is still set (otherwise, we'd
+ violate the basic client-side futex protocol). */
+ r = atomic_fetch_or_relaxed (cond->__data.__g_refs + g1, 1) | 1;
if ((r >> 1) > 0)
futex_wait_simple (cond->__data.__g_refs + g1, r, private);
struct timespec rt;
if (__clock_gettime (CLOCK_MONOTONIC, &rt) != 0)
__libc_fatal ("clock_gettime does not support "
- "CLOCK_MONOTONIC");
+ "CLOCK_MONOTONIC\n");
/* Convert the absolute timeout value to a relative
timeout. */
rt.tv_sec = abstime->tv_sec - rt.tv_sec;
int
pthread_mutex_consistent (pthread_mutex_t *mutex)
{
- /* Test whether this is a robust mutex with a dead owner. */
- if ((mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0
+ /* Test whether this is a robust mutex with a dead owner.
+ See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if ((atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0
|| mutex->__data.__owner != PTHREAD_MUTEX_INCONSISTENT)
return EINVAL;
{
LIBC_PROBE (mutex_destroy, 1, mutex);
- if ((mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if ((atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0
&& mutex->__data.__nusers != 0)
return EBUSY;
- /* Set to an invalid value. */
- mutex->__data.__kind = -1;
+ /* Set to an invalid value. Relaxed MO is enough as it is undefined behavior
+ if the mutex is used after it has been destroyed. But you can reinitialize
+ it with pthread_mutex_init. */
+ atomic_store_relaxed (&(mutex->__data.__kind), -1);
return 0;
}
int
pthread_mutex_getprioceiling (const pthread_mutex_t *mutex, int *prioceiling)
{
- if (__builtin_expect ((mutex->__data.__kind
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if (__builtin_expect ((atomic_load_relaxed (&(mutex->__data.__kind))
& PTHREAD_MUTEX_PRIO_PROTECT_NP) == 0, 0))
return EINVAL;
memset (mutex, '\0', __SIZEOF_PTHREAD_MUTEX_T);
/* Copy the values from the attribute. */
- mutex->__data.__kind = imutexattr->mutexkind & ~PTHREAD_MUTEXATTR_FLAG_BITS;
+ int mutex_kind = imutexattr->mutexkind & ~PTHREAD_MUTEXATTR_FLAG_BITS;
if ((imutexattr->mutexkind & PTHREAD_MUTEXATTR_FLAG_ROBUST) != 0)
{
return ENOTSUP;
#endif
- mutex->__data.__kind |= PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ mutex_kind |= PTHREAD_MUTEX_ROBUST_NORMAL_NP;
}
switch (imutexattr->mutexkind & PTHREAD_MUTEXATTR_PROTOCOL_MASK)
{
case PTHREAD_PRIO_INHERIT << PTHREAD_MUTEXATTR_PROTOCOL_SHIFT:
- mutex->__data.__kind |= PTHREAD_MUTEX_PRIO_INHERIT_NP;
+ mutex_kind |= PTHREAD_MUTEX_PRIO_INHERIT_NP;
break;
case PTHREAD_PRIO_PROTECT << PTHREAD_MUTEXATTR_PROTOCOL_SHIFT:
- mutex->__data.__kind |= PTHREAD_MUTEX_PRIO_PROTECT_NP;
+ mutex_kind |= PTHREAD_MUTEX_PRIO_PROTECT_NP;
int ceiling = (imutexattr->mutexkind
& PTHREAD_MUTEXATTR_PRIO_CEILING_MASK)
FUTEX_PRIVATE_FLAG FUTEX_WAKE. */
if ((imutexattr->mutexkind & (PTHREAD_MUTEXATTR_FLAG_PSHARED
| PTHREAD_MUTEXATTR_FLAG_ROBUST)) != 0)
- mutex->__data.__kind |= PTHREAD_MUTEX_PSHARED_BIT;
+ mutex_kind |= PTHREAD_MUTEX_PSHARED_BIT;
+
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ atomic_store_relaxed (&(mutex->__data.__kind), mutex_kind);
/* Default values: mutex not used yet. */
// mutex->__count = 0; already done by memset
int
__pthread_mutex_lock (pthread_mutex_t *mutex)
{
+ /* See concurrency notes regarding mutex type which is loaded from __kind
+ in struct __pthread_mutex_s in sysdeps/nptl/bits/thread-shared-types.h. */
unsigned int type = PTHREAD_MUTEX_TYPE_ELISION (mutex);
LIBC_PROBE (mutex_entry, 1, mutex);
case PTHREAD_MUTEX_PI_ROBUST_NORMAL_NP:
case PTHREAD_MUTEX_PI_ROBUST_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
- int robust = mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ int kind, robust;
+ {
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int mutex_kind = atomic_load_relaxed (&(mutex->__data.__kind));
+ kind = mutex_kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ robust = mutex_kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ }
if (robust)
{
case PTHREAD_MUTEX_PP_NORMAL_NP:
case PTHREAD_MUTEX_PP_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int kind = atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_KIND_MASK_NP;
oldval = mutex->__data.__lock;
void
__pthread_mutex_cond_lock_adjust (pthread_mutex_t *mutex)
{
- assert ((mutex->__data.__kind & PTHREAD_MUTEX_PRIO_INHERIT_NP) != 0);
- assert ((mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0);
- assert ((mutex->__data.__kind & PTHREAD_MUTEX_PSHARED_BIT) == 0);
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int mutex_kind = atomic_load_relaxed (&(mutex->__data.__kind));
+ assert ((mutex_kind & PTHREAD_MUTEX_PRIO_INHERIT_NP) != 0);
+ assert ((mutex_kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) == 0);
+ assert ((mutex_kind & PTHREAD_MUTEX_PSHARED_BIT) == 0);
/* Record the ownership. */
pid_t id = THREAD_GETMEM (THREAD_SELF, tid);
mutex->__data.__owner = id;
- if (mutex->__data.__kind == PTHREAD_MUTEX_PI_RECURSIVE_NP)
+ if (mutex_kind == PTHREAD_MUTEX_PI_RECURSIVE_NP)
++mutex->__data.__count;
}
#endif
pthread_mutex_setprioceiling (pthread_mutex_t *mutex, int prioceiling,
int *old_ceiling)
{
- /* The low bits of __kind aren't ever changed after pthread_mutex_init,
- so we don't need a lock yet. */
- if ((mutex->__data.__kind & PTHREAD_MUTEX_PRIO_PROTECT_NP) == 0)
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if ((atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_PRIO_PROTECT_NP) == 0)
return EINVAL;
/* See __init_sched_fifo_prio. */
/* We must not check ABSTIME here. If the thread does not block
abstime must not be checked for a valid value. */
+ /* See concurrency notes regarding mutex type which is loaded from __kind
+ in struct __pthread_mutex_s in sysdeps/nptl/bits/thread-shared-types.h. */
switch (__builtin_expect (PTHREAD_MUTEX_TYPE_ELISION (mutex),
PTHREAD_MUTEX_TIMED_NP))
{
case PTHREAD_MUTEX_PI_ROBUST_NORMAL_NP:
case PTHREAD_MUTEX_PI_ROBUST_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
- int robust = mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ int kind, robust;
+ {
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int mutex_kind = atomic_load_relaxed (&(mutex->__data.__kind));
+ kind = mutex_kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ robust = mutex_kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ }
if (robust)
{
case PTHREAD_MUTEX_PP_NORMAL_NP:
case PTHREAD_MUTEX_PP_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int kind = atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_KIND_MASK_NP;
oldval = mutex->__data.__lock;
int oldval;
pid_t id = THREAD_GETMEM (THREAD_SELF, tid);
+ /* See concurrency notes regarding mutex type which is loaded from __kind
+ in struct __pthread_mutex_s in sysdeps/nptl/bits/thread-shared-types.h. */
switch (__builtin_expect (PTHREAD_MUTEX_TYPE_ELISION (mutex),
PTHREAD_MUTEX_TIMED_NP))
{
case PTHREAD_MUTEX_ROBUST_ADAPTIVE_NP:
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending,
&mutex->__data.__list.__next);
+ /* We need to set op_pending before starting the operation. Also
+ see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
oldval = mutex->__data.__lock;
do
/* But it is inconsistent unless marked otherwise. */
mutex->__data.__owner = PTHREAD_MUTEX_INCONSISTENT;
+ /* We must not enqueue the mutex before we have acquired it.
+ Also see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
ENQUEUE_MUTEX (mutex);
+ /* We need to clear op_pending after we enqueue the mutex. */
+ __asm ("" ::: "memory");
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
/* Note that we deliberately exist here. If we fall
int kind = PTHREAD_MUTEX_TYPE (mutex);
if (kind == PTHREAD_MUTEX_ROBUST_ERRORCHECK_NP)
{
+ /* We do not need to ensure ordering wrt another memory
+ access. Also see comments at ENQUEUE_MUTEX. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending,
NULL);
return EDEADLK;
if (kind == PTHREAD_MUTEX_ROBUST_RECURSIVE_NP)
{
+ /* We do not need to ensure ordering wrt another memory
+ access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending,
NULL);
id, 0);
if (oldval != 0 && (oldval & FUTEX_OWNER_DIED) == 0)
{
+ /* We haven't acquired the lock as it is already acquired by
+ another owner. We do not need to ensure ordering wrt another
+ memory access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return EBUSY;
if (oldval == id)
lll_unlock (mutex->__data.__lock,
PTHREAD_ROBUST_MUTEX_PSHARED (mutex));
+ /* FIXME This violates the mutex destruction requirements. See
+ __pthread_mutex_unlock_full. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return ENOTRECOVERABLE;
}
}
while ((oldval & FUTEX_OWNER_DIED) != 0);
+ /* We must not enqueue the mutex before we have acquired it.
+ Also see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
ENQUEUE_MUTEX (mutex);
+ /* We need to clear op_pending after we enqueue the mutex. */
+ __asm ("" ::: "memory");
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
mutex->__data.__owner = id;
case PTHREAD_MUTEX_PI_ROBUST_NORMAL_NP:
case PTHREAD_MUTEX_PI_ROBUST_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
- int robust = mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ int kind, robust;
+ {
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int mutex_kind = atomic_load_relaxed (&(mutex->__data.__kind));
+ kind = mutex_kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ robust = mutex_kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ }
if (robust)
- /* Note: robust PI futexes are signaled by setting bit 0. */
- THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending,
- (void *) (((uintptr_t) &mutex->__data.__list.__next)
- | 1));
+ {
+ /* Note: robust PI futexes are signaled by setting bit 0. */
+ THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending,
+ (void *) (((uintptr_t) &mutex->__data.__list.__next)
+ | 1));
+ /* We need to set op_pending before starting the operation. Also
+ see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
+ }
oldval = mutex->__data.__lock;
{
if (kind == PTHREAD_MUTEX_ERRORCHECK_NP)
{
+ /* We do not need to ensure ordering wrt another memory
+ access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return EDEADLK;
}
if (kind == PTHREAD_MUTEX_RECURSIVE_NP)
{
+ /* We do not need to ensure ordering wrt another memory
+ access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
/* Just bump the counter. */
{
if ((oldval & FUTEX_OWNER_DIED) == 0)
{
+ /* We haven't acquired the lock as it is already acquired by
+ another owner. We do not need to ensure ordering wrt another
+ memory access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return EBUSY;
if (INTERNAL_SYSCALL_ERROR_P (e, __err)
&& INTERNAL_SYSCALL_ERRNO (e, __err) == EWOULDBLOCK)
{
+ /* The kernel has not yet finished the mutex owner death.
+ We do not need to ensure ordering wrt another memory
+ access. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return EBUSY;
/* But it is inconsistent unless marked otherwise. */
mutex->__data.__owner = PTHREAD_MUTEX_INCONSISTENT;
+ /* We must not enqueue the mutex before we have acquired it.
+ Also see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
ENQUEUE_MUTEX (mutex);
+ /* We need to clear op_pending after we enqueue the mutex. */
+ __asm ("" ::: "memory");
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
/* Note that we deliberately exit here. If we fall
PTHREAD_ROBUST_MUTEX_PSHARED (mutex)),
0, 0);
+ /* To the kernel, this will be visible after the kernel has
+ acquired the mutex in the syscall. */
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
return ENOTRECOVERABLE;
}
if (robust)
{
+ /* We must not enqueue the mutex before we have acquired it.
+ Also see comments at ENQUEUE_MUTEX. */
+ __asm ("" ::: "memory");
ENQUEUE_MUTEX_PI (mutex);
+ /* We need to clear op_pending after we enqueue the mutex. */
+ __asm ("" ::: "memory");
THREAD_SETMEM (THREAD_SELF, robust_head.list_op_pending, NULL);
}
case PTHREAD_MUTEX_PP_NORMAL_NP:
case PTHREAD_MUTEX_PP_ADAPTIVE_NP:
{
- int kind = mutex->__data.__kind & PTHREAD_MUTEX_KIND_MASK_NP;
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int kind = atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_KIND_MASK_NP;
oldval = mutex->__data.__lock;
attribute_hidden
__pthread_mutex_unlock_usercnt (pthread_mutex_t *mutex, int decr)
{
+ /* See concurrency notes regarding mutex type which is loaded from __kind
+ in struct __pthread_mutex_s in sysdeps/nptl/bits/thread-shared-types.h. */
int type = PTHREAD_MUTEX_TYPE_ELISION (mutex);
if (__builtin_expect (type &
~(PTHREAD_MUTEX_KIND_MASK_NP|PTHREAD_MUTEX_ELISION_FLAGS_NP), 0))
/* If the previous owner died and the caller did not succeed in
making the state consistent, mark the mutex as unrecoverable
and make all waiters. */
- if ((mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) != 0
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if ((atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_ROBUST_NORMAL_NP) != 0
&& __builtin_expect (mutex->__data.__owner
== PTHREAD_MUTEX_INCONSISTENT, 0))
pi_notrecoverable:
newowner = PTHREAD_MUTEX_NOTRECOVERABLE;
- if ((mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP) != 0)
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ if ((atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_ROBUST_NORMAL_NP) != 0)
{
continue_pi_robust:
/* Remove mutex from the list.
/* Unlock. Load all necessary mutex data before releasing the mutex
to not violate the mutex destruction requirements (see
lll_unlock). */
- int robust = mutex->__data.__kind & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
+ /* See concurrency notes regarding __kind in struct __pthread_mutex_s
+ in sysdeps/nptl/bits/thread-shared-types.h. */
+ int robust = atomic_load_relaxed (&(mutex->__data.__kind))
+ & PTHREAD_MUTEX_ROBUST_NORMAL_NP;
private = (robust
? PTHREAD_ROBUST_MUTEX_PSHARED (mutex)
: PTHREAD_MUTEX_PSHARED (mutex));
harmless because the flag is just about the state of
__readers, and all threads set the flag under the same
conditions. */
- while ((atomic_load_relaxed (&rwlock->__data.__readers)
- & PTHREAD_RWLOCK_RWAITING) != 0)
+ while (((r = atomic_load_relaxed (&rwlock->__data.__readers))
+ & PTHREAD_RWLOCK_RWAITING) != 0)
{
int private = __pthread_rwlock_get_private (rwlock);
int err = futex_abstimed_wait (&rwlock->__data.__readers,
/* Same as in __pthread_rwlock_rdlock_full:
We started the read phase, so we are also responsible for
updating the write-phase futex. Relaxed MO is sufficient.
- Note that there can be no other reader that we have to wake
- because all other readers will see the read phase started by us
- (or they will try to start it themselves); if a writer started
- the read phase, we cannot have started it. Furthermore, we
- cannot discard a PTHREAD_RWLOCK_FUTEX_USED flag because we will
- overwrite the value set by the most recent writer (or the readers
- before it in case of explicit hand-over) and we know that there
- are no waiting readers. */
- atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 0);
+ We have to do the same steps as a writer would when handing over the
+ read phase to use because other readers cannot distinguish between
+ us and the writer.
+ Note that __pthread_rwlock_tryrdlock callers will not have to be
+ woken up because they will either see the read phase started by us
+ or they will try to start it themselves; however, callers of
+ __pthread_rwlock_rdlock_full just increase the reader count and then
+ check what state the lock is in, so they cannot distinguish between
+ us and a writer that acquired and released the lock in the
+ meantime. */
+ if ((atomic_exchange_relaxed (&rwlock->__data.__wrphase_futex, 0)
+ & PTHREAD_RWLOCK_FUTEX_USED) != 0)
+ {
+ int private = __pthread_rwlock_get_private (rwlock);
+ futex_wake (&rwlock->__data.__wrphase_futex, INT_MAX, private);
+ }
}
return 0;
&rwlock->__data.__readers, &r,
r | PTHREAD_RWLOCK_WRPHASE | PTHREAD_RWLOCK_WRLOCKED))
{
+ /* We have become the primary writer and we cannot have shared
+ the PTHREAD_RWLOCK_FUTEX_USED flag with someone else, so we
+ can simply enable blocking (see full wrlock code). */
atomic_store_relaxed (&rwlock->__data.__writers_futex, 1);
- atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 1);
+ /* If we started a write phase, we need to enable readers to
+ wait. If we did not, we must not change it because other threads
+ may have set the PTHREAD_RWLOCK_FUTEX_USED in the meantime. */
+ if ((r & PTHREAD_RWLOCK_WRPHASE) == 0)
+ atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 1);
atomic_store_relaxed (&rwlock->__data.__cur_writer,
THREAD_GETMEM (THREAD_SELF, tid));
return 0;
}
void
-__run_fork_handlers (enum __run_fork_handler_type who)
+__run_fork_handlers (enum __run_fork_handler_type who, _Bool do_locking)
{
struct fork_handler *runp;
if (who == atfork_run_prepare)
{
- lll_lock (atfork_lock, LLL_PRIVATE);
+ if (do_locking)
+ lll_lock (atfork_lock, LLL_PRIVATE);
size_t sl = fork_handler_list_size (&fork_handlers);
for (size_t i = sl; i > 0; i--)
{
else if (who == atfork_run_parent && runp->parent_handler)
runp->parent_handler ();
}
- lll_unlock (atfork_lock, LLL_PRIVATE);
+ if (do_locking)
+ lll_unlock (atfork_lock, LLL_PRIVATE);
}
}
--- /dev/null
+/* Dummy audit library for test-audit-threads.
+
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <elf.h>
+#include <link.h>
+#include <stdio.h>
+#include <assert.h>
+#include <string.h>
+
+/* We must use a dummy LD_AUDIT module to force the dynamic loader to
+ *not* update the real PLT, and instead use a cached value for the
+ lazy resolution result. It is the update of that cached value that
+ we are testing for correctness by doing this. */
+
+/* Library to be audited. */
+#define LIB "tst-audit-threads-mod2.so"
+/* CALLNUM is the number of retNum functions. */
+#define CALLNUM 7999
+
+#define CONCATX(a, b) __CONCAT (a, b)
+
+static int previous = 0;
+
+unsigned int
+la_version (unsigned int ver)
+{
+ return 1;
+}
+
+unsigned int
+la_objopen (struct link_map *map, Lmid_t lmid, uintptr_t *cookie)
+{
+ return LA_FLG_BINDTO | LA_FLG_BINDFROM;
+}
+
+uintptr_t
+CONCATX(la_symbind, __ELF_NATIVE_CLASS) (ElfW(Sym) *sym,
+ unsigned int ndx,
+ uintptr_t *refcook,
+ uintptr_t *defcook,
+ unsigned int *flags,
+ const char *symname)
+{
+ const char * retnum = "retNum";
+ char * num = strstr (symname, retnum);
+ int n;
+ /* Validate if the symbols are getting called in the correct order.
+ This code is here to verify binutils does not optimize out the PLT
+ entries that require the symbol binding. */
+ if (num != NULL)
+ {
+ n = atoi (num);
+ assert (n >= previous);
+ assert (n <= CALLNUM);
+ previous = n;
+ }
+ return sym->st_value;
+}
--- /dev/null
+/* Shared object with a huge number of functions for test-audit-threads.
+
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* Define all the retNumN functions in a library. */
+#define definenum
+#include "tst-audit-threads.h"
--- /dev/null
+/* Test multi-threading using LD_AUDIT.
+
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* This test uses a dummy LD_AUDIT library (test-audit-threads-mod1) and a
+ library with a huge number of functions in order to validate lazy symbol
+ binding with an audit library. We use one thread per CPU to test that
+ concurrent lazy resolution does not have any defects which would cause
+ the process to fail. We use an LD_AUDIT library to force the testing of
+ the relocation resolution caching code in the dynamic loader i.e.
+ _dl_runtime_profile and _dl_profile_fixup. */
+
+#include <support/xthread.h>
+#include <strings.h>
+#include <stdlib.h>
+#include <sys/sysinfo.h>
+
+static int do_test (void);
+
+/* This test usually takes less than 3s to run. However, there are cases that
+ take up to 30s. */
+#define TIMEOUT 60
+#define TEST_FUNCTION do_test ()
+#include "../test-skeleton.c"
+
+/* Declare the functions we are going to call. */
+#define externnum
+#include "tst-audit-threads.h"
+#undef externnum
+
+int num_threads;
+pthread_barrier_t barrier;
+
+void
+sync_all (int num)
+{
+ pthread_barrier_wait (&barrier);
+}
+
+void
+call_all_ret_nums (void)
+{
+ /* Call each function one at a time from all threads. */
+#define callnum
+#include "tst-audit-threads.h"
+#undef callnum
+}
+
+void *
+thread_main (void *unused)
+{
+ call_all_ret_nums ();
+ return NULL;
+}
+
+#define STR2(X) #X
+#define STR(X) STR2(X)
+
+static int
+do_test (void)
+{
+ int i;
+ pthread_t *threads;
+
+ num_threads = get_nprocs ();
+ if (num_threads <= 1)
+ num_threads = 2;
+
+ /* Used to synchronize all the threads after calling each retNumN. */
+ xpthread_barrier_init (&barrier, NULL, num_threads);
+
+ threads = (pthread_t *) xcalloc (num_threads, sizeof(pthread_t));
+ for (i = 0; i < num_threads; i++)
+ threads[i] = xpthread_create(NULL, thread_main, NULL);
+
+ for (i = 0; i < num_threads; i++)
+ xpthread_join(threads[i]);
+
+ free (threads);
+
+ return 0;
+}
--- /dev/null
+/* Helper header for test-audit-threads.
+
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* We use this helper to create a large number of functions, all of
+ which will be resolved lazily and thus have their PLT updated.
+ This is done to provide enough functions that we can statistically
+ observe a thread vs. PLT resolution failure if one exists. */
+
+#define CONCAT(a, b) a ## b
+#define NUM(x, y) CONCAT (x, y)
+
+#define FUNC10(x) \
+ FUNC (NUM (x, 0)); \
+ FUNC (NUM (x, 1)); \
+ FUNC (NUM (x, 2)); \
+ FUNC (NUM (x, 3)); \
+ FUNC (NUM (x, 4)); \
+ FUNC (NUM (x, 5)); \
+ FUNC (NUM (x, 6)); \
+ FUNC (NUM (x, 7)); \
+ FUNC (NUM (x, 8)); \
+ FUNC (NUM (x, 9))
+
+#define FUNC100(x) \
+ FUNC10 (NUM (x, 0)); \
+ FUNC10 (NUM (x, 1)); \
+ FUNC10 (NUM (x, 2)); \
+ FUNC10 (NUM (x, 3)); \
+ FUNC10 (NUM (x, 4)); \
+ FUNC10 (NUM (x, 5)); \
+ FUNC10 (NUM (x, 6)); \
+ FUNC10 (NUM (x, 7)); \
+ FUNC10 (NUM (x, 8)); \
+ FUNC10 (NUM (x, 9))
+
+#define FUNC1000(x) \
+ FUNC100 (NUM (x, 0)); \
+ FUNC100 (NUM (x, 1)); \
+ FUNC100 (NUM (x, 2)); \
+ FUNC100 (NUM (x, 3)); \
+ FUNC100 (NUM (x, 4)); \
+ FUNC100 (NUM (x, 5)); \
+ FUNC100 (NUM (x, 6)); \
+ FUNC100 (NUM (x, 7)); \
+ FUNC100 (NUM (x, 8)); \
+ FUNC100 (NUM (x, 9))
+
+#define FUNC7000() \
+ FUNC1000 (1); \
+ FUNC1000 (2); \
+ FUNC1000 (3); \
+ FUNC1000 (4); \
+ FUNC1000 (5); \
+ FUNC1000 (6); \
+ FUNC1000 (7);
+
+#ifdef FUNC
+# undef FUNC
+#endif
+
+#ifdef externnum
+# define FUNC(x) extern int CONCAT (retNum, x) (void)
+#endif
+
+#ifdef definenum
+# define FUNC(x) int CONCAT (retNum, x) (void) { return x; }
+#endif
+
+#ifdef callnum
+# define FUNC(x) CONCAT (retNum, x) (); sync_all (x)
+#endif
+
+/* A value of 7000 functions is chosen as an arbitrarily large
+ number of functions that will allow us enough attempts to
+ verify lazy resolution operation. */
+FUNC7000 ();
--- /dev/null
+/* Testing race while enabling lock elision.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <pthread.h>
+#include <unistd.h>
+#include <getopt.h>
+#include <support/support.h>
+#include <support/xthread.h>
+
+static pthread_barrier_t barrier;
+static pthread_mutex_t mutex;
+static long long int iteration_count = 1000000;
+static unsigned int thread_count = 3;
+
+static void *
+thr_func (void *arg)
+{
+ long long int i;
+ for (i = 0; i < iteration_count; i++)
+ {
+ if ((uintptr_t) arg == 0)
+ {
+ xpthread_mutex_destroy (&mutex);
+ xpthread_mutex_init (&mutex, NULL);
+ }
+
+ xpthread_barrier_wait (&barrier);
+
+ /* Test if enabling lock elision works if it is enabled concurrently.
+ There was a race in FORCE_ELISION macro which leads to either
+ pthread_mutex_destroy returning EBUSY as the owner was recorded
+ by pthread_mutex_lock - in "normal mutex" code path - but was not
+ resetted in pthread_mutex_unlock - in "elision" code path.
+ Or it leads to the assertion in nptl/pthread_mutex_lock.c:
+ assert (mutex->__data.__owner == 0);
+ Please ensure that the test is run with lock elision:
+ export GLIBC_TUNABLES=glibc.elision.enable=1 */
+ xpthread_mutex_lock (&mutex);
+ xpthread_mutex_unlock (&mutex);
+
+ xpthread_barrier_wait (&barrier);
+ }
+ return NULL;
+}
+
+static int
+do_test (void)
+{
+ unsigned int i;
+ printf ("Starting %d threads to run %lld iterations.\n",
+ thread_count, iteration_count);
+
+ pthread_t *threads = xmalloc (thread_count * sizeof (pthread_t));
+ xpthread_barrier_init (&barrier, NULL, thread_count);
+ xpthread_mutex_init (&mutex, NULL);
+
+ for (i = 0; i < thread_count; i++)
+ threads[i] = xpthread_create (NULL, thr_func, (void *) (uintptr_t) i);
+
+ for (i = 0; i < thread_count; i++)
+ xpthread_join (threads[i]);
+
+ xpthread_barrier_destroy (&barrier);
+ free (threads);
+
+ return EXIT_SUCCESS;
+}
+
+#define OPT_ITERATIONS 10000
+#define OPT_THREADS 10001
+#define CMDLINE_OPTIONS \
+ { "iterations", required_argument, NULL, OPT_ITERATIONS }, \
+ { "threads", required_argument, NULL, OPT_THREADS },
+static void
+cmdline_process (int c)
+{
+ long long int arg = strtoll (optarg, NULL, 0);
+ switch (c)
+ {
+ case OPT_ITERATIONS:
+ if (arg > 0)
+ iteration_count = arg;
+ break;
+ case OPT_THREADS:
+ if (arg > 0 && arg < 100)
+ thread_count = arg;
+ break;
+ }
+}
+#define CMDLINE_PROCESS cmdline_process
+#define TIMEOUT 50
+#include <support/test-driver.c>
--- /dev/null
+/* Test rwlock with PREFER_WRITER_NONRECURSIVE_NP (bug 23861).
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <support/xthread.h>
+
+/* We choose 10 iterations because this happens to be able to trigger the
+ stall on contemporary hardware. */
+#define LOOPS 10
+/* We need 3 threads to trigger bug 23861. One thread as a writer, and
+ two reader threads. The test verifies that the second-to-last reader
+ is able to notify the *last* reader that it should be done waiting.
+ If the second-to-last reader fails to notify the last reader or does
+ so incorrectly then the last reader may stall indefinitely. */
+#define NTHREADS 3
+
+_Atomic int do_exit;
+pthread_rwlockattr_t mylock_attr;
+pthread_rwlock_t mylock;
+
+void *
+run_loop (void *a)
+{
+ while (!do_exit)
+ {
+ if (random () & 1)
+ {
+ xpthread_rwlock_wrlock (&mylock);
+ xpthread_rwlock_unlock (&mylock);
+ }
+ else
+ {
+ xpthread_rwlock_rdlock (&mylock);
+ xpthread_rwlock_unlock (&mylock);
+ }
+ }
+ return NULL;
+}
+
+int
+do_test (void)
+{
+ xpthread_rwlockattr_init (&mylock_attr);
+ xpthread_rwlockattr_setkind_np (&mylock_attr,
+ PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP);
+ xpthread_rwlock_init (&mylock, &mylock_attr);
+
+ for (int n = 0; n < LOOPS; n++)
+ {
+ pthread_t tids[NTHREADS];
+ do_exit = 0;
+ for (int i = 0; i < NTHREADS; i++)
+ tids[i] = xpthread_create (NULL, run_loop, NULL);
+ /* Let the threads run for some time. */
+ sleep (1);
+ printf ("Exiting...");
+ fflush (stdout);
+ do_exit = 1;
+ for (int i = 0; i < NTHREADS; i++)
+ xpthread_join (tids[i]);
+ printf ("done.\n");
+ }
+ pthread_rwlock_destroy (&mylock);
+ pthread_rwlockattr_destroy (&mylock_attr);
+ return 0;
+}
+
+#define TIMEOUT (DEFAULT_TIMEOUT + 3 * LOOPS)
+#include <support/test-driver.c>
--- /dev/null
+/* Bug 23844: Test for pthread_rwlock_tryrdlock stalls.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* For a full analysis see comment:
+ https://sourceware.org/bugzilla/show_bug.cgi?id=23844#c14
+
+ Provided here for reference:
+
+ --- Analysis of pthread_rwlock_tryrdlock() stall ---
+ A read lock begins to execute.
+
+ In __pthread_rwlock_rdlock_full:
+
+ We can attempt a read lock, but find that the lock is
+ in a write phase (PTHREAD_RWLOCK_WRPHASE, or WP-bit
+ is set), and the lock is held by a primary writer
+ (PTHREAD_RWLOCK_WRLOCKED is set). In this case we must
+ wait for explicit hand over from the writer to us or
+ one of the other waiters. The read lock threads are
+ about to execute:
+
+ 341 r = (atomic_fetch_add_acquire (&rwlock->__data.__readers,
+ 342 (1 << PTHREAD_RWLOCK_READER_SHIFT))
+ 343 + (1 << PTHREAD_RWLOCK_READER_SHIFT));
+
+ An unlock beings to execute.
+
+ Then in __pthread_rwlock_wrunlock:
+
+ 547 unsigned int r = atomic_load_relaxed (&rwlock->__data.__readers);
+ ...
+ 549 while (!atomic_compare_exchange_weak_release
+ 550 (&rwlock->__data.__readers, &r,
+ 551 ((r ^ PTHREAD_RWLOCK_WRLOCKED)
+ 552 ^ ((r >> PTHREAD_RWLOCK_READER_SHIFT) == 0 ? 0
+ 553 : PTHREAD_RWLOCK_WRPHASE))))
+ 554 {
+ ...
+ 556 }
+
+ We clear PTHREAD_RWLOCK_WRLOCKED, and if there are
+ no readers so we leave the lock in PTHRAD_RWLOCK_WRPHASE.
+
+ Back in the read lock.
+
+ The read lock adjusts __readres as above.
+
+ 383 while ((r & PTHREAD_RWLOCK_WRPHASE) != 0
+ 384 && (r & PTHREAD_RWLOCK_WRLOCKED) == 0)
+ 385 {
+ ...
+ 390 if (atomic_compare_exchange_weak_acquire (&rwlock->__data.__readers, &r,
+ 391 r ^ PTHREAD_RWLOCK_WRPHASE))
+ 392 {
+
+ And then attemps to start the read phase.
+
+ Assume there happens to be a tryrdlock at this point, noting
+ that PTHREAD_RWLOCK_WRLOCKED is clear, and PTHREAD_RWLOCK_WRPHASE
+ is 1. So the try lock attemps to start the read phase.
+
+ In __pthread_rwlock_tryrdlock:
+
+ 44 if ((r & PTHREAD_RWLOCK_WRPHASE) == 0)
+ 45 {
+ ...
+ 49 if (((r & PTHREAD_RWLOCK_WRLOCKED) != 0)
+ 50 && (rwlock->__data.__flags
+ 51 == PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP))
+ 52 return EBUSY;
+ 53 rnew = r + (1 << PTHREAD_RWLOCK_READER_SHIFT);
+ 54 }
+ ...
+ 89 while (!atomic_compare_exchange_weak_acquire (&rwlock->__data.__readers,
+ 90 &r, rnew));
+
+ And succeeds.
+
+ Back in the write unlock:
+
+ 557 if ((r >> PTHREAD_RWLOCK_READER_SHIFT) != 0)
+ 558 {
+ ...
+ 563 if ((atomic_exchange_relaxed (&rwlock->__data.__wrphase_futex, 0)
+ 564 & PTHREAD_RWLOCK_FUTEX_USED) != 0)
+ 565 futex_wake (&rwlock->__data.__wrphase_futex, INT_MAX, private);
+ 566 }
+
+ We note that PTHREAD_RWLOCK_FUTEX_USED is non-zero
+ and don't wake anyone. This is OK because we handed
+ over to the trylock. It will be the trylock's responsibility
+ to wake any waiters.
+
+ Back in the read lock:
+
+ The read lock fails to install PTHRAD_REWLOCK_WRPHASE as 0 because
+ the __readers value was adjusted by the trylock, and so it falls through
+ to waiting on the lock for explicit handover from either a new writer
+ or a new reader.
+
+ 448 int err = futex_abstimed_wait (&rwlock->__data.__wrphase_futex,
+ 449 1 | PTHREAD_RWLOCK_FUTEX_USED,
+ 450 abstime, private);
+
+ We use PTHREAD_RWLOCK_FUTEX_USED to indicate the futex
+ is in use.
+
+ At this point we have readers waiting on the read lock
+ to unlock. The wrlock is done. The trylock is finishing
+ the installation of the read phase.
+
+ 92 if ((r & PTHREAD_RWLOCK_WRPHASE) != 0)
+ 93 {
+ ...
+ 105 atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 0);
+ 106 }
+
+ The trylock does note that we were the one that
+ installed the read phase, but the comments are not
+ correct, the execution ordering above shows that
+ readers might indeed be waiting, and they are.
+
+ The atomic_store_relaxed throws away PTHREAD_RWLOCK_FUTEX_USED,
+ and the waiting reader is never worken becuase as noted
+ above it is conditional on the futex being used.
+
+ The solution is for the trylock thread to inspect
+ PTHREAD_RWLOCK_FUTEX_USED and wake the waiting readers.
+
+ --- Analysis of pthread_rwlock_trywrlock() stall ---
+
+ A write lock begins to execute, takes the write lock,
+ and then releases the lock...
+
+ In pthread_rwlock_wrunlock():
+
+ 547 unsigned int r = atomic_load_relaxed (&rwlock->__data.__readers);
+ ...
+ 549 while (!atomic_compare_exchange_weak_release
+ 550 (&rwlock->__data.__readers, &r,
+ 551 ((r ^ PTHREAD_RWLOCK_WRLOCKED)
+ 552 ^ ((r >> PTHREAD_RWLOCK_READER_SHIFT) == 0 ? 0
+ 553 : PTHREAD_RWLOCK_WRPHASE))))
+ 554 {
+ ...
+ 556 }
+
+ ... leaving it in the write phase with zero readers
+ (the case where we leave the write phase in place
+ during a write unlock).
+
+ A write trylock begins to execute.
+
+ In __pthread_rwlock_trywrlock:
+
+ 40 while (((r & PTHREAD_RWLOCK_WRLOCKED) == 0)
+ 41 && (((r >> PTHREAD_RWLOCK_READER_SHIFT) == 0)
+ 42 || (prefer_writer && ((r & PTHREAD_RWLOCK_WRPHASE) != 0))))
+ 43 {
+
+ The lock is not locked.
+
+ There are no readers.
+
+ 45 if (atomic_compare_exchange_weak_acquire (
+ 46 &rwlock->__data.__readers, &r,
+ 47 r | PTHREAD_RWLOCK_WRPHASE | PTHREAD_RWLOCK_WRLOCKED))
+
+ We atomically install the write phase and we take the
+ exclusive write lock.
+
+ 48 {
+ 49 atomic_store_relaxed (&rwlock->__data.__writers_futex, 1);
+
+ We get this far.
+
+ A reader lock begins to execute.
+
+ In pthread_rwlock_rdlock:
+
+ 437 for (;;)
+ 438 {
+ 439 while (((wpf = atomic_load_relaxed (&rwlock->__data.__wrphase_futex))
+ 440 | PTHREAD_RWLOCK_FUTEX_USED) == (1 | PTHREAD_RWLOCK_FUTEX_USED))
+ 441 {
+ 442 int private = __pthread_rwlock_get_private (rwlock);
+ 443 if (((wpf & PTHREAD_RWLOCK_FUTEX_USED) == 0)
+ 444 && (!atomic_compare_exchange_weak_relaxed
+ 445 (&rwlock->__data.__wrphase_futex,
+ 446 &wpf, wpf | PTHREAD_RWLOCK_FUTEX_USED)))
+ 447 continue;
+ 448 int err = futex_abstimed_wait (&rwlock->__data.__wrphase_futex,
+ 449 1 | PTHREAD_RWLOCK_FUTEX_USED,
+ 450 abstime, private);
+
+ We are in a write phase, so the while() on line 439 is true.
+
+ The value of wpf does not have PTHREAD_RWLOCK_FUTEX_USED set
+ since this is the first reader to lock.
+
+ The atomic operation sets wpf with PTHREAD_RELOCK_FUTEX_USED
+ on the expectation that this reader will be woken during
+ the handoff.
+
+ Back in pthread_rwlock_trywrlock:
+
+ 50 atomic_store_relaxed (&rwlock->__data.__wrphase_futex, 1);
+ 51 atomic_store_relaxed (&rwlock->__data.__cur_writer,
+ 52 THREAD_GETMEM (THREAD_SELF, tid));
+ 53 return 0;
+ 54 }
+ ...
+ 57 }
+
+ We write 1 to __wrphase_futex discarding PTHREAD_RWLOCK_FUTEX_USED,
+ and so in the unlock we will not awaken the waiting reader.
+
+ The solution to this is to realize that if we did not start the write
+ phase we need not write 1 or any other value to __wrphase_futex.
+ This ensures that any readers (which saw __wrphase_futex != 0) can
+ set PTHREAD_RWLOCK_FUTEX_USED and this can be used at unlock to
+ wake them.
+
+ If we installed the write phase then all other readers are looping
+ here:
+
+ In __pthread_rwlock_rdlock_full:
+
+ 437 for (;;)
+ 438 {
+ 439 while (((wpf = atomic_load_relaxed (&rwlock->__data.__wrphase_futex))
+ 440 | PTHREAD_RWLOCK_FUTEX_USED) == (1 | PTHREAD_RWLOCK_FUTEX_USED))
+ 441 {
+ ...
+ 508 }
+
+ waiting for the write phase to be installed or removed before they
+ can begin waiting on __wrphase_futex (part of the algorithm), or
+ taking a concurrent read lock, and thus we can safely write 1 to
+ __wrphase_futex.
+
+ If we did not install the write phase then the readers may already
+ be waiting on the futex, the original writer wrote 1 to __wrphase_futex
+ as part of starting the write phase, and we cannot also write 1
+ without loosing the PTHREAD_RWLOCK_FUTEX_USED bit.
+
+ ---
+
+ Summary for the pthread_rwlock_tryrdlock() stall:
+
+ The stall is caused by pthread_rwlock_tryrdlock failing to check
+ that PTHREAD_RWLOCK_FUTEX_USED is set in the __wrphase_futex futex
+ and then waking the futex.
+
+ The fix for bug 23844 ensures that waiters on __wrphase_futex are
+ correctly woken. Before the fix the test stalls as readers can
+ wait forever on __wrphase_futex. */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <support/xthread.h>
+#include <errno.h>
+
+/* We need only one lock to reproduce the issue. We will need multiple
+ threads to get the exact case where we have a read, try, and unlock
+ all interleaving to produce the case where the readers are waiting
+ and the try fails to wake them. */
+pthread_rwlock_t onelock;
+
+/* The number of threads is arbitrary but empirically chosen to have
+ enough threads that we see the condition where waiting readers are
+ not woken by a successful tryrdlock. */
+#define NTHREADS 32
+
+_Atomic int do_exit;
+
+void *
+run_loop (void *arg)
+{
+ int i = 0, ret;
+ while (!do_exit)
+ {
+ /* Arbitrarily choose if we are the writer or reader. Choose a
+ high enough ratio of readers to writers to make it likely
+ that readers block (and eventually are susceptable to
+ stalling).
+
+ If we are a writer, take the write lock, and then unlock.
+ If we are a reader, try the lock, then lock, then unlock. */
+ if ((i % 8) != 0)
+ xpthread_rwlock_wrlock (&onelock);
+ else
+ {
+ if ((ret = pthread_rwlock_tryrdlock (&onelock)) != 0)
+ {
+ if (ret == EBUSY)
+ xpthread_rwlock_rdlock (&onelock);
+ else
+ exit (EXIT_FAILURE);
+ }
+ }
+ /* Thread does some work and then unlocks. */
+ xpthread_rwlock_unlock (&onelock);
+ i++;
+ }
+ return NULL;
+}
+
+int
+do_test (void)
+{
+ int i;
+ pthread_t tids[NTHREADS];
+ xpthread_rwlock_init (&onelock, NULL);
+ for (i = 0; i < NTHREADS; i++)
+ tids[i] = xpthread_create (NULL, run_loop, NULL);
+ /* Run for some amount of time. Empirically speaking exercising
+ the stall via pthread_rwlock_tryrdlock is much harder, and on
+ a 3.5GHz 4 core x86_64 VM system it takes somewhere around
+ 20-200s to stall, approaching 100% stall past 200s. We can't
+ wait that long for a regression test so we just test for 20s,
+ and expect the stall to happen with a 5-10% chance (enough for
+ developers to see). */
+ sleep (20);
+ /* Then exit. */
+ printf ("INFO: Exiting...\n");
+ do_exit = 1;
+ /* If any readers stalled then we will timeout waiting for them. */
+ for (i = 0; i < NTHREADS; i++)
+ xpthread_join (tids[i]);
+ printf ("INFO: Done.\n");
+ xpthread_rwlock_destroy (&onelock);
+ printf ("PASS: No pthread_rwlock_tryrdlock stalls detected.\n");
+ return 0;
+}
+
+#define TIMEOUT 30
+#include <support/test-driver.c>
--- /dev/null
+/* Bug 23844: Test for pthread_rwlock_trywrlock stalls.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* For a full analysis see comments in tst-rwlock-tryrdlock-stall.c.
+
+ Summary for the pthread_rwlock_trywrlock() stall:
+
+ The stall is caused by pthread_rwlock_trywrlock setting
+ __wrphase_futex futex to 1 and loosing the
+ PTHREAD_RWLOCK_FUTEX_USED bit.
+
+ The fix for bug 23844 ensures that waiters on __wrphase_futex are
+ correctly woken. Before the fix the test stalls as readers can
+ wait forever on __wrphase_futex. */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <support/xthread.h>
+#include <errno.h>
+
+/* We need only one lock to reproduce the issue. We will need multiple
+ threads to get the exact case where we have a read, try, and unlock
+ all interleaving to produce the case where the readers are waiting
+ and the try clears the PTHREAD_RWLOCK_FUTEX_USED bit and a
+ subsequent unlock fails to wake them. */
+pthread_rwlock_t onelock;
+
+/* The number of threads is arbitrary but empirically chosen to have
+ enough threads that we see the condition where waiting readers are
+ not woken by a successful unlock. */
+#define NTHREADS 32
+
+_Atomic int do_exit;
+
+void *
+run_loop (void *arg)
+{
+ int i = 0, ret;
+ while (!do_exit)
+ {
+ /* Arbitrarily choose if we are the writer or reader. Choose a
+ high enough ratio of readers to writers to make it likely
+ that readers block (and eventually are susceptable to
+ stalling).
+
+ If we are a writer, take the write lock, and then unlock.
+ If we are a reader, try the lock, then lock, then unlock. */
+ if ((i % 8) != 0)
+ {
+ if ((ret = pthread_rwlock_trywrlock (&onelock)) != 0)
+ {
+ if (ret == EBUSY)
+ xpthread_rwlock_wrlock (&onelock);
+ else
+ exit (EXIT_FAILURE);
+ }
+ }
+ else
+ xpthread_rwlock_rdlock (&onelock);
+ /* Thread does some work and then unlocks. */
+ xpthread_rwlock_unlock (&onelock);
+ i++;
+ }
+ return NULL;
+}
+
+int
+do_test (void)
+{
+ int i;
+ pthread_t tids[NTHREADS];
+ xpthread_rwlock_init (&onelock, NULL);
+ for (i = 0; i < NTHREADS; i++)
+ tids[i] = xpthread_create (NULL, run_loop, NULL);
+ /* Run for some amount of time. The pthread_rwlock_tryrwlock stall
+ is very easy to trigger and happens in seconds under the test
+ conditions. */
+ sleep (10);
+ /* Then exit. */
+ printf ("INFO: Exiting...\n");
+ do_exit = 1;
+ /* If any readers stalled then we will timeout waiting for them. */
+ for (i = 0; i < NTHREADS; i++)
+ xpthread_join (tids[i]);
+ printf ("INFO: Done.\n");
+ xpthread_rwlock_destroy (&onelock);
+ printf ("PASS: No pthread_rwlock_tryrwlock stalls detected.\n");
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+#define USE_PTHREADS 1
+#include "../elf/tst-unwind-main.c"
getsrvbynm_r getsrvbypt_r servicescache \
dbg_log nscd_conf nscd_stat cache mem nscd_setup_thread \
xmalloc xstrdup aicache initgrcache gai res_hconf \
- netgroupcache
+ netgroupcache nscd-inet_addr
ifeq ($(build-nscd)$(have-thread-library),yesyes)
/* This file uses the getaddrinfo code but it compiles it without NSCD
support. We just need a few symbol renames. */
-#define __inet_aton inet_aton
#define __ioctl ioctl
#define __getsockname getsockname
#define __socket socket
#define __getifaddrs getifaddrs
#define __freeifaddrs freeifaddrs
+/* We do not want to export __inet_aton_exact. Get the prototype and
+ change its visibility to hidden. */
+#include <arpa/inet.h>
+__typeof__ (__inet_aton_exact) __inet_aton_exact
+ __attribute__ ((visibility ("hidden")));
+
/* We are nscd, so we don't want to be talking to ourselves. */
#undef USE_NSCD
#define HAVE_LOOKUP_BUFFER 1
#define HAVE_AF 1
-#define __inet_aton inet_aton
-
/* We are nscd, so we don't want to be talking to ourselves. */
#undef USE_NSCD
/* This is really only for debugging. */
if (NSS_STATUS_TRYAGAIN > status || status > NSS_STATUS_RETURN)
- __libc_fatal ("illegal status in internal_getgrouplist");
+ __libc_fatal ("Illegal status in internal_getgrouplist.\n");
any_success |= status == NSS_STATUS_SUCCESS;
static time_t
addgetnetgrentX (struct database_dyn *db, int fd, request_header *req,
const char *key, uid_t uid, struct hashentry *he,
- struct datahead *dh, struct dataset **resultp)
+ struct datahead *dh, struct dataset **resultp,
+ void **tofreep)
{
if (__glibc_unlikely (debug_level > 0))
{
size_t group_len = strlen (key) + 1;
struct name_list *first_needed
= alloca (sizeof (struct name_list) + group_len);
+ *tofreep = NULL;
if (netgroup_database == NULL
&& __nss_database_lookup ("netgroup", NULL, NULL, &netgroup_database))
memset (&data, '\0', sizeof (data));
buffer = xmalloc (buflen);
+ *tofreep = buffer;
first_needed->next = first_needed;
memcpy (first_needed->name, key, group_len);
data.needed_groups = first_needed;
}
out:
- free (buffer);
-
*resultp = dataset;
return timeout;
group, group_len,
db, uid);
time_t timeout;
+ void *tofree;
if (result != NULL)
- timeout = result->head.timeout;
+ {
+ timeout = result->head.timeout;
+ tofree = NULL;
+ }
else
{
request_header req_get =
.key_len = group_len
};
timeout = addgetnetgrentX (db, -1, &req_get, group, uid, NULL, NULL,
- &result);
+ &result, &tofree);
}
struct indataset
++dh->nreloads;
if (cacheable)
pthread_rwlock_unlock (&db->lock);
- return timeout;
+ goto out;
}
if (he == NULL)
dh->usable = false;
}
+ out:
+ free (tofree);
return timeout;
}
+static time_t
+addgetnetgrentX_ignore (struct database_dyn *db, int fd, request_header *req,
+ const char *key, uid_t uid, struct hashentry *he,
+ struct datahead *dh)
+{
+ struct dataset *ignore;
+ void *tofree;
+ time_t timeout = addgetnetgrentX (db, fd, req, key, uid, he, dh,
+ &ignore, &tofree);
+ free (tofree);
+ return timeout;
+}
+
void
addgetnetgrent (struct database_dyn *db, int fd, request_header *req,
void *key, uid_t uid)
{
- struct dataset *ignore;
-
- addgetnetgrentX (db, fd, req, key, uid, NULL, NULL, &ignore);
+ addgetnetgrentX_ignore (db, fd, req, key, uid, NULL, NULL);
}
.type = GETNETGRENT,
.key_len = he->len
};
- struct dataset *ignore;
-
- return addgetnetgrentX (db, -1, &req, db->data + he->key, he->owner, he, dh,
- &ignore);
+ return addgetnetgrentX_ignore
+ (db, -1, &req, db->data + he->key, he->owner, he, dh);
}
--- /dev/null
+/* Legacy IPv4 text-to-address functions. Version for nscd.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <arpa/inet.h>
+
+/* We do not want to export __inet_aton_exact. Get the prototype and
+ change the visibility to hidden. */
+#include <arpa/inet.h>
+__typeof__ (__inet_aton_exact) __inet_aton_exact
+ __attribute__ ((visibility ("hidden")));
+
+/* Do not provide definitions of the public symbols exported from
+ libc. */
+#undef weak_alias
+#define weak_alias(from, to)
+
+#include <resolv/inet_addr.c>
if (!arg1)
error (0, 0, _("Must specify user name for server-user option"));
else
- server_user = xstrdup (arg1);
+ {
+ free ((char *) server_user);
+ server_user = xstrdup (arg1);
+ }
}
else if (strcmp (entry, "stat-user") == 0)
{
error (0, 0, _("Must specify user name for stat-user option"));
else
{
+ free ((char *) stat_user);
stat_user = xstrdup (arg1);
struct passwd *pw = getpwnam (stat_user);
tests += tst-nss-files-hosts-erange
tests += tst-nss-files-hosts-multi
tests += tst-nss-files-hosts-getent
+tests += tst-nss-files-alias-leak
endif
# If we have a thread library then we can test cancellation against
$(objpfx)tst-nss-files-hosts-erange: $(libdl)
$(objpfx)tst-nss-files-hosts-multi: $(libdl)
$(objpfx)tst-nss-files-hosts-getent: $(libdl)
+$(objpfx)tst-nss-files-alias-leak: $(libdl)
+$(objpfx)tst-nss-files-alias-leak.out: $(objpfx)/libnss_files.so
#include "nsswitch.h"
#ifdef USE_NSCD
-# define inet_aton __inet_aton
# include <nscd/nscd_proto.h>
#endif
255.255.255.255? The test below will succeed
spuriously... ??? */
if (af == AF_INET)
- ok = __inet_aton (name, (struct in_addr *) host_addr);
+ ok = __inet_aton_exact (name, (struct in_addr *) host_addr);
else
{
assert (af == AF_INET6);
{
while (! feof_unlocked (listfile))
{
+ if (room_left < 2)
+ {
+ free (old_line);
+ fclose (listfile);
+ goto no_more_room;
+ }
+
first_unused[room_left - 1] = '\xff';
line = fgets_unlocked (first_unused, room_left,
listfile);
if (first_unused[room_left - 1] != '\xff')
{
free (old_line);
+ fclose (listfile);
goto no_more_room;
}
+ __alignof__ (char *)))
{
free (old_line);
+ fclose (listfile);
goto no_more_room;
}
room_left -= ((first_unused - cp)
/* This is really only for debugging. */
if (__builtin_expect (NSS_STATUS_TRYAGAIN > status
|| status > NSS_STATUS_RETURN, 0))
- __libc_fatal ("illegal status in __nss_next");
+ __libc_fatal ("Illegal status in __nss_next.\n");
if (nss_next_action (*ni, status) == NSS_ACTION_RETURN)
return 1;
--- /dev/null
+/* Check for file descriptor leak in alias :include: processing (bug 23521).
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <aliases.h>
+#include <array_length.h>
+#include <dlfcn.h>
+#include <errno.h>
+#include <gnu/lib-names.h>
+#include <nss.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/namespace.h>
+#include <support/support.h>
+#include <support/temp_file.h>
+#include <support/test-driver.h>
+#include <support/xstdio.h>
+#include <support/xunistd.h>
+
+static struct support_chroot *chroot_env;
+
+/* Number of the aliases for the "many" user. This must be large
+ enough to trigger reallocation for the pointer array, but result in
+ answers below the maximum size tried in do_test. */
+enum { many_aliases = 30 };
+
+static void
+prepare (int argc, char **argv)
+{
+ chroot_env = support_chroot_create
+ ((struct support_chroot_configuration) { } );
+
+ char *path = xasprintf ("%s/etc/aliases", chroot_env->path_chroot);
+ add_temp_file (path);
+ support_write_file_string
+ (path,
+ "user1: :include:/etc/aliases.user1\n"
+ "user2: :include:/etc/aliases.user2\n"
+ "comment: comment1, :include:/etc/aliases.comment\n"
+ "many: :include:/etc/aliases.many\n");
+ free (path);
+
+ path = xasprintf ("%s/etc/aliases.user1", chroot_env->path_chroot);
+ add_temp_file (path);
+ support_write_file_string (path, "alias1\n");
+ free (path);
+
+ path = xasprintf ("%s/etc/aliases.user2", chroot_env->path_chroot);
+ add_temp_file (path);
+ support_write_file_string (path, "alias1a, alias2\n");
+ free (path);
+
+ path = xasprintf ("%s/etc/aliases.comment", chroot_env->path_chroot);
+ add_temp_file (path);
+ support_write_file_string
+ (path,
+ /* The line must be longer than the line with the :include:
+ directive in /etc/aliases. */
+ "# Long line. ##############################################\n"
+ "comment2\n");
+ free (path);
+
+ path = xasprintf ("%s/etc/aliases.many", chroot_env->path_chroot);
+ add_temp_file (path);
+ FILE *fp = xfopen (path, "w");
+ for (int i = 0; i < many_aliases; ++i)
+ fprintf (fp, "a%d\n", i);
+ TEST_VERIFY_EXIT (! ferror (fp));
+ xfclose (fp);
+ free (path);
+}
+
+/* The names of the users to test. */
+static const char *users[] = { "user1", "user2", "comment", "many" };
+
+static void
+check_aliases (int id, const struct aliasent *e)
+{
+ TEST_VERIFY_EXIT (id >= 0 || id < array_length (users));
+ const char *name = users[id];
+ TEST_COMPARE_BLOB (e->alias_name, strlen (e->alias_name),
+ name, strlen (name));
+
+ switch (id)
+ {
+ case 0:
+ TEST_COMPARE (e->alias_members_len, 1);
+ TEST_COMPARE_BLOB (e->alias_members[0], strlen (e->alias_members[0]),
+ "alias1", strlen ("alias1"));
+ break;
+
+ case 1:
+ TEST_COMPARE (e->alias_members_len, 2);
+ TEST_COMPARE_BLOB (e->alias_members[0], strlen (e->alias_members[0]),
+ "alias1a", strlen ("alias1a"));
+ TEST_COMPARE_BLOB (e->alias_members[1], strlen (e->alias_members[1]),
+ "alias2", strlen ("alias2"));
+ break;
+
+ case 2:
+ TEST_COMPARE (e->alias_members_len, 2);
+ TEST_COMPARE_BLOB (e->alias_members[0], strlen (e->alias_members[0]),
+ "comment1", strlen ("comment1"));
+ TEST_COMPARE_BLOB (e->alias_members[1], strlen (e->alias_members[1]),
+ "comment2", strlen ("comment2"));
+ break;
+
+ case 3:
+ TEST_COMPARE (e->alias_members_len, many_aliases);
+ for (int i = 0; i < e->alias_members_len; ++i)
+ {
+ char alias[30];
+ int len = snprintf (alias, sizeof (alias), "a%d", i);
+ TEST_VERIFY_EXIT (len > 0);
+ TEST_COMPARE_BLOB (e->alias_members[i], strlen (e->alias_members[i]),
+ alias, len);
+ }
+ break;
+ }
+}
+
+static int
+do_test (void)
+{
+ /* Make sure we don't try to load the module in the chroot. */
+ if (dlopen (LIBNSS_FILES_SO, RTLD_NOW) == NULL)
+ FAIL_EXIT1 ("could not load " LIBNSS_FILES_SO ": %s", dlerror ());
+
+ /* Some of these descriptors will become unavailable if there is a
+ file descriptor leak. 10 is chosen somewhat arbitrarily. The
+ array must be longer than the number of files opened by nss_files
+ at the same time (currently that number is 2). */
+ int next_descriptors[10];
+ for (size_t i = 0; i < array_length (next_descriptors); ++i)
+ {
+ next_descriptors[i] = dup (0);
+ TEST_VERIFY_EXIT (next_descriptors[i] > 0);
+ }
+ for (size_t i = 0; i < array_length (next_descriptors); ++i)
+ xclose (next_descriptors[i]);
+
+ support_become_root ();
+ if (!support_can_chroot ())
+ return EXIT_UNSUPPORTED;
+
+ __nss_configure_lookup ("aliases", "files");
+
+ xchroot (chroot_env->path_chroot);
+
+ /* Attempt various buffer sizes. If the operation succeeds, we
+ expect correct data. */
+ for (int id = 0; id < array_length (users); ++id)
+ {
+ bool found = false;
+ for (size_t size = 1; size <= 1000; ++size)
+ {
+ void *buffer = malloc (size);
+ struct aliasent result;
+ struct aliasent *res;
+ errno = EINVAL;
+ int ret = getaliasbyname_r (users[id], &result, buffer, size, &res);
+ if (ret == 0)
+ {
+ if (res != NULL)
+ {
+ found = true;
+ check_aliases (id, res);
+ }
+ else
+ {
+ support_record_failure ();
+ printf ("error: failed lookup for user \"%s\", size %zu\n",
+ users[id], size);
+ }
+ }
+ else if (ret != ERANGE)
+ {
+ support_record_failure ();
+ printf ("error: invalid return code %d (user \%s\", size %zu)\n",
+ ret, users[id], size);
+ }
+ free (buffer);
+
+ /* Make sure that we did not have a file descriptor leak. */
+ for (size_t i = 0; i < array_length (next_descriptors); ++i)
+ {
+ int new_fd = dup (0);
+ if (new_fd != next_descriptors[i])
+ {
+ support_record_failure ();
+ printf ("error: descriptor %d at index %zu leaked"
+ " (user \"%s\", size %zu)\n",
+ next_descriptors[i], i, users[id], size);
+
+ /* Close unexpected descriptor, the leak probing
+ descriptors, and the leaked descriptor
+ next_descriptors[i]. */
+ xclose (new_fd);
+ for (size_t j = 0; j <= i; ++j)
+ xclose (next_descriptors[j]);
+ goto next_size;
+ }
+ }
+ for (size_t i = 0; i < array_length (next_descriptors); ++i)
+ xclose (next_descriptors[i]);
+
+ next_size:
+ ;
+ }
+ if (!found)
+ {
+ support_record_failure ();
+ printf ("error: user %s not found\n", users[id]);
+ }
+ }
+
+ support_chroot_free (chroot_env);
+ return 0;
+}
+
+#define PREPARE prepare
+#include <support/test-driver.c>
tst-posix_fadvise tst-posix_fadvise64 \
tst-sysconf-empty-chroot tst-glob_symlinks tst-fexecve \
tst-glob-tilde test-ssize-max tst-spawn4 bug-regex37 \
- bug-regex38
+ bug-regex38 tst-regcomp-truncated
tests-internal := bug-regex5 bug-regex20 bug-regex33 \
tst-rfc3484 tst-rfc3484-2 tst-rfc3484-3 \
tst-glob_lstat_compat tst-spawn4-compat
$(objpfx)tst-regexloc.out: $(gen-locales)
$(objpfx)tst-rxspencer.out: $(gen-locales)
$(objpfx)tst-rxspencer-no-utf8.out: $(gen-locales)
+$(objpfx)tst-regcomp-truncated.out: $(gen-locales)
endif
# If we will use the generic uname implementation, we must figure out what
mbclen = __mbrtowc (&wc,
((const char *) pstr->raw_mbs + pstr->raw_mbs_idx
+ byte_idx), remain_len, &pstr->cur_state);
- if (BE (mbclen < (size_t) -2, 1))
+ if (BE (0 < mbclen && mbclen < (size_t) -2, 1))
{
wchar_t wcu = __towupper (wc);
if (wcu != wc)
else
p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + src_idx;
mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state);
- if (BE (mbclen < (size_t) -2, 1))
+ if (BE (0 < mbclen && mbclen < (size_t) -2, 1))
{
wchar_t wcu = __towupper (wc);
if (wcu != wc)
else if (naccepted)
{
char *buf = (char *) re_string_get_buffer (&mctx->input);
- if (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx,
- naccepted) != 0)
+ if (mctx->input.valid_len - *pidx < naccepted
+ || (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx,
+ naccepted)
+ != 0))
return -1;
}
}
-/* BZ #18877 and #21270 mmap offset test.
+/* BZ #18877, BZ #21270, and BZ #24699 mmap offset test.
Copyright (C) 2015-2018 Free Software Foundation, Inc.
This file is part of the GNU C Library.
#include <unistd.h>
#include <errno.h>
#include <sys/mman.h>
+#include <mmap_info.h>
#include <support/check.h>
/* Check if invalid offset are handled correctly by mmap. */
static int
-do_test_bz21270 (void)
+do_test_large_offset (void)
{
/* For architectures with sizeof (off_t) < sizeof (off64_t) mmap is
implemented with __SYS_mmap2 syscall and the offset is represented in
const size_t length = 4096;
void *addr = mmap64 (NULL, length, prot, flags, fd, offset);
- if (sizeof (off_t) < sizeof (off64_t))
+ if (mmap64_maximum_offset (page_shift) < UINT64_MAX)
{
if ((addr != MAP_FAILED) && (errno != EINVAL))
FAIL_RET ("mmap succeed");
int ret = 0;
ret += do_test_bz18877 ();
- ret += do_test_bz21270 ();
+ ret += do_test_large_offset ();
return ret;
}
--- /dev/null
+/* Test compilation of truncated regular expressions.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* This test constructs various patterns in an attempt to trigger
+ over-reading the regular expression compiler, such as bug
+ 23578. */
+
+#include <array_length.h>
+#include <errno.h>
+#include <locale.h>
+#include <regex.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/next_to_fault.h>
+#include <support/support.h>
+#include <support/test-driver.h>
+#include <wchar.h>
+
+/* Locales to test. */
+static const char locales[][17] =
+ {
+ "C",
+ "en_US.UTF-8",
+ "de_DE.ISO-8859-1",
+ };
+
+/* Syntax options. Will be combined with other flags. */
+static const reg_syntax_t syntaxes[] =
+ {
+ RE_SYNTAX_EMACS,
+ RE_SYNTAX_AWK,
+ RE_SYNTAX_GNU_AWK,
+ RE_SYNTAX_POSIX_AWK,
+ RE_SYNTAX_GREP,
+ RE_SYNTAX_EGREP,
+ RE_SYNTAX_POSIX_EGREP,
+ RE_SYNTAX_POSIX_BASIC,
+ RE_SYNTAX_POSIX_EXTENDED,
+ RE_SYNTAX_POSIX_MINIMAL_EXTENDED,
+ };
+
+/* Trailing characters placed after the initial character. */
+static const char trailing_strings[][4] =
+ {
+ "",
+ "[",
+ "\\",
+ "[\\",
+ "(",
+ "(\\",
+ "\\(",
+ };
+
+static int
+do_test (void)
+{
+ /* Staging buffer for the constructed regular expression. */
+ char buffer[16];
+
+ /* Allocation used to detect over-reading by the regular expression
+ compiler. */
+ struct support_next_to_fault ntf
+ = support_next_to_fault_allocate (sizeof (buffer));
+
+ /* Arbitrary Unicode codepoint at which we stop generating
+ characters. We do not probe the whole range because that would
+ take too long due to combinatorical exploision as the result of
+ combination with other flags. */
+ static const wchar_t last_character = 0xfff;
+
+ for (size_t locale_idx = 0; locale_idx < array_length (locales);
+ ++ locale_idx)
+ {
+ if (setlocale (LC_ALL, locales[locale_idx]) == NULL)
+ {
+ support_record_failure ();
+ printf ("error: setlocale (\"%s\"): %m", locales[locale_idx]);
+ continue;
+ }
+ if (test_verbose > 0)
+ printf ("info: testing locale \"%s\"\n", locales[locale_idx]);
+
+ for (wchar_t wc = 0; wc <= last_character; ++wc)
+ {
+ char *after_wc;
+ if (wc == 0)
+ {
+ /* wcrtomb treats L'\0' in a special way. */
+ *buffer = '\0';
+ after_wc = &buffer[1];
+ }
+ else
+ {
+ mbstate_t ps = { };
+ size_t ret = wcrtomb (buffer, wc, &ps);
+ if (ret == (size_t) -1)
+ {
+ /* EILSEQ means that the target character set
+ cannot encode the character. */
+ if (errno != EILSEQ)
+ {
+ support_record_failure ();
+ printf ("error: wcrtomb (0x%x) failed: %m\n",
+ (unsigned) wc);
+ }
+ continue;
+ }
+ TEST_VERIFY_EXIT (ret != 0);
+ after_wc = &buffer[ret];
+ }
+
+ for (size_t trailing_idx = 0;
+ trailing_idx < array_length (trailing_strings);
+ ++trailing_idx)
+ {
+ char *after_trailing
+ = stpcpy (after_wc, trailing_strings[trailing_idx]);
+
+ for (int do_nul = 0; do_nul < 2; ++do_nul)
+ {
+ char *after_nul;
+ if (do_nul)
+ {
+ *after_trailing = '\0';
+ after_nul = &after_trailing[1];
+ }
+ else
+ after_nul = after_trailing;
+
+ size_t length = after_nul - buffer;
+
+ /* Make sure that the faulting region starts
+ after the used portion of the buffer. */
+ char *ntf_start = ntf.buffer + sizeof (buffer) - length;
+ memcpy (ntf_start, buffer, length);
+
+ for (const reg_syntax_t *psyntax = syntaxes;
+ psyntax < array_end (syntaxes); ++psyntax)
+ for (int do_icase = 0; do_icase < 2; ++do_icase)
+ {
+ re_syntax_options = *psyntax;
+ if (do_icase)
+ re_syntax_options |= RE_ICASE;
+
+ regex_t reg;
+ memset (®, 0, sizeof (reg));
+ const char *msg = re_compile_pattern
+ (ntf_start, length, ®);
+ if (msg != NULL)
+ {
+ if (test_verbose > 0)
+ {
+ char *quoted = support_quote_blob
+ (buffer, length);
+ printf ("info: compilation failed for pattern"
+ " \"%s\", syntax 0x%lx: %s\n",
+ quoted, re_syntax_options, msg);
+ free (quoted);
+ }
+ }
+ else
+ regfree (®);
+ }
+ }
+ }
+ }
+ }
+
+ support_next_to_fault_free (&ntf);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
tests = tst-aton tst-leaks tst-inet_ntop
xtests = tst-leaks2
+tests-internal += tst-inet_aton_exact
+
+
generate := mtrace-tst-leaks.out tst-leaks.mtrace tst-leaks2.mtrace
extra-libs := libresolv libnss_dns
tst-resolv-binary \
tst-resolv-edns \
tst-resolv-network \
+ tst-resolv-nondecimal \
tst-resolv-res_init-multi \
tst-resolv-search \
+ tst-resolv-trailing \
# These tests need libdl.
ifeq (yes,$(build-shared))
$(shared-thread-library)
$(objpfx)tst-resolv-res_init-thread: $(libdl) $(objpfx)libresolv.so \
$(shared-thread-library)
+$(objpfx)tst-resolv-nondecimal: $(objpfx)libresolv.so $(shared-thread-library)
$(objpfx)tst-resolv-qtypes: $(objpfx)libresolv.so $(shared-thread-library)
$(objpfx)tst-resolv-rotate: $(objpfx)libresolv.so $(shared-thread-library)
$(objpfx)tst-resolv-search: $(objpfx)libresolv.so $(shared-thread-library)
+$(objpfx)tst-resolv-trailing: $(objpfx)libresolv.so $(shared-thread-library)
$(objpfx)tst-resolv-threads: \
$(libdl) $(objpfx)libresolv.so $(shared-thread-library)
$(objpfx)tst-resolv-canonname: \
__h_errno; __resp;
__res_iclose;
+ __inet_aton_exact;
__inet_pton_length;
__resolv_context_get;
__resolv_context_get_preinit;
/* We cannot create a thread in the moment and there is
also no thread running. This is a problem. `errno' is
set to EAGAIN if this is only a temporary problem. */
- assert (lastp->next == newp);
- lastp->next = NULL;
+ assert (requests == newp || lastp->next == newp);
+ if (lastp != NULL)
+ lastp->next = NULL;
+ else
+ requests = NULL;
requests_tail = lastp;
newp->next = freelist;
+/* Legacy IPv4 text-to-address functions.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
/*
* Copyright (c) 1983, 1990, 1993
* The Regents of the University of California. All rights reserved.
#include <limits.h>
#include <errno.h>
-/*
- * Ascii internet address interpretation routine.
- * The value returned is in network order.
- */
-in_addr_t
-__inet_addr(const char *cp) {
- struct in_addr val;
-
- if (__inet_aton(cp, &val))
- return (val.s_addr);
- return (INADDR_NONE);
+/* Check whether "cp" is a valid ASCII representation of an IPv4
+ Internet address and convert it to a binary address. Returns 1 if
+ the address is valid, 0 if not. This replaces inet_addr, the
+ return value from which cannot distinguish between failure and a
+ local broadcast address. Write a pointer to the first
+ non-converted character to *endp. */
+static int
+inet_aton_end (const char *cp, struct in_addr *addr, const char **endp)
+{
+ static const in_addr_t max[4] = { 0xffffffff, 0xffffff, 0xffff, 0xff };
+ in_addr_t val;
+ char c;
+ union iaddr
+ {
+ uint8_t bytes[4];
+ uint32_t word;
+ } res;
+ uint8_t *pp = res.bytes;
+ int digit;
+
+ int saved_errno = errno;
+ __set_errno (0);
+
+ res.word = 0;
+
+ c = *cp;
+ for (;;)
+ {
+ /* Collect number up to ``.''. Values are specified as for C:
+ 0x=hex, 0=octal, isdigit=decimal. */
+ if (!isdigit (c))
+ goto ret_0;
+ {
+ char *endp;
+ unsigned long ul = strtoul (cp, &endp, 0);
+ if (ul == ULONG_MAX && errno == ERANGE)
+ goto ret_0;
+ if (ul > 0xfffffffful)
+ goto ret_0;
+ val = ul;
+ digit = cp != endp;
+ cp = endp;
+ }
+ c = *cp;
+ if (c == '.')
+ {
+ /* Internet format:
+ a.b.c.d
+ a.b.c (with c treated as 16 bits)
+ a.b (with b treated as 24 bits). */
+ if (pp > res.bytes + 2 || val > 0xff)
+ goto ret_0;
+ *pp++ = val;
+ c = *++cp;
+ }
+ else
+ break;
+ }
+ /* Check for trailing characters. */
+ if (c != '\0' && (!isascii (c) || !isspace (c)))
+ goto ret_0;
+ /* Did we get a valid digit? */
+ if (!digit)
+ goto ret_0;
+
+ /* Check whether the last part is in its limits depending on the
+ number of parts in total. */
+ if (val > max[pp - res.bytes])
+ goto ret_0;
+
+ if (addr != NULL)
+ addr->s_addr = res.word | htonl (val);
+ *endp = cp;
+
+ __set_errno (saved_errno);
+ return 1;
+
+ ret_0:
+ __set_errno (saved_errno);
+ return 0;
}
-weak_alias (__inet_addr, inet_addr)
-/*
- * Check whether "cp" is a valid ascii representation
- * of an Internet address and convert to a binary address.
- * Returns 1 if the address is valid, 0 if not.
- * This replaces inet_addr, the return value from which
- * cannot distinguish between failure and a local broadcast address.
- */
int
-__inet_aton(const char *cp, struct in_addr *addr)
+__inet_aton_exact (const char *cp, struct in_addr *addr)
{
- static const in_addr_t max[4] = { 0xffffffff, 0xffffff, 0xffff, 0xff };
- in_addr_t val;
- char c;
- union iaddr {
- uint8_t bytes[4];
- uint32_t word;
- } res;
- uint8_t *pp = res.bytes;
- int digit;
-
- int saved_errno = errno;
- __set_errno (0);
-
- res.word = 0;
-
- c = *cp;
- for (;;) {
- /*
- * Collect number up to ``.''.
- * Values are specified as for C:
- * 0x=hex, 0=octal, isdigit=decimal.
- */
- if (!isdigit(c))
- goto ret_0;
- {
- char *endp;
- unsigned long ul = strtoul (cp, (char **) &endp, 0);
- if (ul == ULONG_MAX && errno == ERANGE)
- goto ret_0;
- if (ul > 0xfffffffful)
- goto ret_0;
- val = ul;
- digit = cp != endp;
- cp = endp;
- }
- c = *cp;
- if (c == '.') {
- /*
- * Internet format:
- * a.b.c.d
- * a.b.c (with c treated as 16 bits)
- * a.b (with b treated as 24 bits)
- */
- if (pp > res.bytes + 2 || val > 0xff)
- goto ret_0;
- *pp++ = val;
- c = *++cp;
- } else
- break;
- }
- /*
- * Check for trailing characters.
- */
- if (c != '\0' && (!isascii(c) || !isspace(c)))
- goto ret_0;
- /*
- * Did we get a valid digit?
- */
- if (!digit)
- goto ret_0;
-
- /* Check whether the last part is in its limits depending on
- the number of parts in total. */
- if (val > max[pp - res.bytes])
- goto ret_0;
-
- if (addr != NULL)
- addr->s_addr = res.word | htonl (val);
+ struct in_addr val;
+ const char *endp;
+ /* Check that inet_aton_end parsed the entire string. */
+ if (inet_aton_end (cp, &val, &endp) != 0 && *endp == 0)
+ {
+ *addr = val;
+ return 1;
+ }
+ else
+ return 0;
+}
+libc_hidden_def (__inet_aton_exact)
- __set_errno (saved_errno);
- return (1);
+/* inet_aton ignores trailing garbage. */
+int
+__inet_aton_ignore_trailing (const char *cp, struct in_addr *addr)
+{
+ const char *endp;
+ return inet_aton_end (cp, addr, &endp);
+}
+weak_alias (__inet_aton_ignore_trailing, inet_aton)
-ret_0:
- __set_errno (saved_errno);
- return (0);
+/* ASCII IPv4 Internet address interpretation routine. The value
+ returned is in network order. */
+in_addr_t
+__inet_addr (const char *cp)
+{
+ struct in_addr val;
+ const char *endp;
+ if (inet_aton_end (cp, &val, &endp))
+ return val.s_addr;
+ return INADDR_NONE;
}
-weak_alias (__inet_aton, inet_aton)
-libc_hidden_def (__inet_aton)
-libc_hidden_weak (inet_aton)
+weak_alias (__inet_addr, inet_addr)
return status;
}
+/* Verify that the name looks like a host name. There is no point in
+ sending a query which will not produce a usable name in the
+ response. */
+static enum nss_status
+check_name (const char *name, int *h_errnop)
+{
+ if (res_hnok (name))
+ return NSS_STATUS_SUCCESS;
+ *h_errnop = HOST_NOT_FOUND;
+ return NSS_STATUS_NOTFOUND;
+}
+
enum nss_status
_nss_dns_gethostbyname2_r (const char *name, int af, struct hostent *result,
char *buffer, size_t buflen, int *errnop,
int *h_errnop)
{
+ enum nss_status status = check_name (name, h_errnop);
+ if (status != NSS_STATUS_SUCCESS)
+ return status;
return _nss_dns_gethostbyname3_r (name, af, result, buffer, buflen, errnop,
h_errnop, NULL, NULL);
}
char *buffer, size_t buflen, int *errnop,
int *h_errnop)
{
+ enum nss_status status = check_name (name, h_errnop);
+ if (status != NSS_STATUS_SUCCESS)
+ return status;
struct resolv_context *ctx = __resolv_context_get ();
if (ctx == NULL)
{
*h_errnop = NETDB_INTERNAL;
return NSS_STATUS_UNAVAIL;
}
- enum nss_status status = NSS_STATUS_NOTFOUND;
+ status = NSS_STATUS_NOTFOUND;
if (res_use_inet6 ())
status = gethostbyname3_context (ctx, name, AF_INET6, result, buffer,
buflen, errnop, h_errnop, NULL, NULL);
char *buffer, size_t buflen, int *errnop,
int *herrnop, int32_t *ttlp)
{
+ enum nss_status status = check_name (name, herrnop);
+ if (status != NSS_STATUS_SUCCESS)
+ return status;
struct resolv_context *ctx = __resolv_context_get ();
if (ctx == NULL)
{
int ans2p_malloced = 0;
int olderr = errno;
- enum nss_status status;
int n = __res_context_search (ctx, name, C_IN, T_QUERY_A_AND_AAAA,
host_buffer.buf->buf, 2048, &host_buffer.ptr,
&ans2p, &nans2p, &resplen2, &ans2p_malloced);
cp = parser->buffer + sizeof ("nameserver") - 1;
while (*cp == ' ' || *cp == '\t')
cp++;
+
+ /* Ignore trailing contents on the name server line. */
+ {
+ char *el;
+ if ((el = strpbrk (cp, " \t\n")) != NULL)
+ *el = '\0';
+ }
+
struct sockaddr *sa;
- if ((*cp != '\0') && (*cp != '\n') && __inet_aton (cp, &a))
+ if ((*cp != '\0') && (*cp != '\n') && __inet_aton_exact (cp, &a))
{
sa = allocate_address_v4 (a, NAMESERVER_PORT);
if (sa == NULL)
{
struct in6_addr a6;
char *el;
-
- if ((el = strpbrk (cp, " \t\n")) != NULL)
- *el = '\0';
if ((el = strchr (cp, SCOPE_DELIMITER)) != NULL)
*el = '\0';
if ((*cp != '\0') && (__inet_pton (AF_INET6, cp, &a6) > 0))
char separator = *cp;
*cp = 0;
struct resolv_sortlist_entry e;
- if (__inet_aton (net, &a))
+ if (__inet_aton_exact (net, &a))
{
e.addr = a;
if (is_sort_mask (separator))
cp++;
separator = *cp;
*cp = 0;
- if (__inet_aton (net, &a))
+ if (__inet_aton_exact (net, &a))
e.mask = a.s_addr;
else
e.mask = net_mask (e.addr);
+/* Test legacy IPv4 text-to-address function inet_aton.
+ Copyright (C) 1998-2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <array_length.h>
#include <stdio.h>
#include <stdint.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
-
-static struct tests
+static const struct tests
{
const char *input;
int valid;
{ "-1", 0, 0 },
{ "256", 1, 0x00000100 },
{ "256.", 0, 0 },
+ { "255a", 0, 0 },
{ "256a", 0, 0 },
{ "0x100", 1, 0x00000100 },
{ "0200.0x123456", 1, 0x80123456 },
{ "1.2.256.4", 0, 0 },
{ "1.2.3.0x100", 0, 0 },
{ "323543357756889", 0, 0 },
- { "10.1.2.3.4", 0, 0},
+ { "10.1.2.3.4", 0, 0 },
+ { "192.0.2.1", 1, 0xc0000201 },
+ { "192.0.2.2\nX", 1, 0xc0000202 },
+ { "192.0.2.3 Y", 1, 0xc0000203 },
+ { "192.0.2.3Z", 0, 0 },
+ { "192.000.002.010", 1, 0xc0000208 },
};
int result = 0;
size_t cnt;
- for (cnt = 0; cnt < sizeof (tests) / sizeof (tests[0]); ++cnt)
+ for (cnt = 0; cnt < array_length (tests); ++cnt)
{
struct in_addr addr;
return result;
}
-#define TEST_FUNCTION do_test ()
-#include "../test-skeleton.c"
+#include <support/test-driver.c>
--- /dev/null
+/* Test internal legacy IPv4 text-to-address function __inet_aton_exact.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <arpa/inet.h>
+#include <support/check.h>
+
+static int
+do_test (void)
+{
+ struct in_addr addr = { };
+
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.1", &addr), 1);
+ TEST_COMPARE (ntohl (addr.s_addr), 0xC0000201);
+
+ TEST_COMPARE (__inet_aton_exact ("192.000.002.010", &addr), 1);
+ TEST_COMPARE (ntohl (addr.s_addr), 0xC0000208);
+ TEST_COMPARE (__inet_aton_exact ("0xC0000234", &addr), 1);
+ TEST_COMPARE (ntohl (addr.s_addr), 0xC0000234);
+
+ /* Trailing content is not accepted. */
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.2X", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.3 Y", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.4\nZ", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.5\tT", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.6 Y", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.7\n", &addr), 0);
+ TEST_COMPARE (__inet_aton_exact ("192.0.2.8\t", &addr), 0);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
resolv_response_add_data (b, &rrtype, sizeof (rrtype));
}
break;
+ case 104:
+ send_ptr (b, qname, qclass, qtype, "host.example");
+ break;
default:
FAIL_EXIT1 ("invalid QNAME: %s (code %d)", qname, code);
}
"error: TRY_AGAIN\n");
check_netent ("code103.example", getnetbyname ("code103.example"),
"error: NO_RECOVERY\n");
+ /* Test bug #17630. */
+ check_netent ("code104.example", getnetbyname ("code104.example"),
+ "error: TRY_AGAIN\n");
/* Lookup by address, success cases. */
check_reverse (1,
--- /dev/null
+/* Test name resolution behavior for octal, hexadecimal IPv4 addresses.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <netdb.h>
+#include <stdlib.h>
+#include <support/check.h>
+#include <support/check_nss.h>
+#include <support/resolv_test.h>
+#include <support/support.h>
+
+static void
+response (const struct resolv_response_context *ctx,
+ struct resolv_response_builder *b,
+ const char *qname, uint16_t qclass, uint16_t qtype)
+{
+ /* The tests are not supposed send any DNS queries. */
+ FAIL_EXIT1 ("unexpected DNS query for %s/%d/%d", qname, qclass, qtype);
+}
+
+static void
+run_query_addrinfo (const char *query, const char *address)
+{
+ char *quoted_query = support_quote_string (query);
+
+ struct addrinfo *ai;
+ struct addrinfo hints =
+ {
+ .ai_socktype = SOCK_STREAM,
+ .ai_protocol = IPPROTO_TCP,
+ };
+
+ char *context = xasprintf ("getaddrinfo \"%s\" AF_INET", quoted_query);
+ char *expected = xasprintf ("address: STREAM/TCP %s 80\n", address);
+ hints.ai_family = AF_INET;
+ int ret = getaddrinfo (query, "80", &hints, &ai);
+ check_addrinfo (context, ai, ret, expected);
+ if (ret == 0)
+ freeaddrinfo (ai);
+ free (context);
+
+ context = xasprintf ("getaddrinfo \"%s\" AF_UNSPEC", quoted_query);
+ hints.ai_family = AF_UNSPEC;
+ ret = getaddrinfo (query, "80", &hints, &ai);
+ check_addrinfo (context, ai, ret, expected);
+ if (ret == 0)
+ freeaddrinfo (ai);
+ free (expected);
+ free (context);
+
+ context = xasprintf ("getaddrinfo \"%s\" AF_INET6", quoted_query);
+ expected = xasprintf ("flags: AI_V4MAPPED\n"
+ "address: STREAM/TCP ::ffff:%s 80\n",
+ address);
+ hints.ai_family = AF_INET6;
+ hints.ai_flags = AI_V4MAPPED;
+ ret = getaddrinfo (query, "80", &hints, &ai);
+ check_addrinfo (context, ai, ret, expected);
+ if (ret == 0)
+ freeaddrinfo (ai);
+ free (expected);
+ free (context);
+
+ free (quoted_query);
+}
+
+static void
+run_query (const char *query, const char *address)
+{
+ char *quoted_query = support_quote_string (query);
+ char *context = xasprintf ("gethostbyname (\"%s\")", quoted_query);
+ char *expected = xasprintf ("name: %s\n"
+ "address: %s\n", query, address);
+ check_hostent (context, gethostbyname (query), expected);
+ free (context);
+
+ context = xasprintf ("gethostbyname_r \"%s\"", quoted_query);
+ struct hostent storage;
+ char buf[4096];
+ struct hostent *e = NULL;
+ TEST_COMPARE (gethostbyname_r (query, &storage, buf, sizeof (buf),
+ &e, &h_errno), 0);
+ check_hostent (context, e, expected);
+ free (context);
+
+ context = xasprintf ("gethostbyname2 (\"%s\", AF_INET)", quoted_query);
+ check_hostent (context, gethostbyname2 (query, AF_INET), expected);
+ free (context);
+
+ context = xasprintf ("gethostbyname2_r \"%s\" AF_INET", quoted_query);
+ e = NULL;
+ TEST_COMPARE (gethostbyname2_r (query, AF_INET, &storage, buf, sizeof (buf),
+ &e, &h_errno), 0);
+ check_hostent (context, e, expected);
+ free (context);
+ free (expected);
+
+ free (quoted_query);
+
+ /* The gethostbyname tests are always valid for getaddrinfo, but not
+ vice versa. */
+ run_query_addrinfo (query, address);
+}
+
+static int
+do_test (void)
+{
+ struct resolv_test *aux = resolv_test_start
+ ((struct resolv_redirect_config)
+ {
+ .response_callback = response,
+ });
+
+ run_query ("192.000.002.010", "192.0.2.8");
+
+ /* Hexadecimal numbers are not accepted by gethostbyname. */
+ run_query_addrinfo ("0xc0000210", "192.0.2.16");
+ run_query_addrinfo ("192.0x234", "192.0.2.52");
+
+ resolv_test_end (aux);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test name resolution behavior with trailing characters.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <array_length.h>
+#include <netdb.h>
+#include <support/check.h>
+#include <support/check_nss.h>
+#include <support/resolv_test.h>
+#include <support/support.h>
+
+static void
+response (const struct resolv_response_context *ctx,
+ struct resolv_response_builder *b,
+ const char *qname, uint16_t qclass, uint16_t qtype)
+{
+ /* The tests are not supposed send any DNS queries. */
+ FAIL_EXIT1 ("unexpected DNS query for %s/%d/%d", qname, qclass, qtype);
+}
+
+static int
+do_test (void)
+{
+ struct resolv_test *aux = resolv_test_start
+ ((struct resolv_redirect_config)
+ {
+ .response_callback = response,
+ });
+
+ static const char *const queries[] =
+ {
+ "192.0.2.1 ",
+ "192.0.2.2\t",
+ "192.0.2.3\n",
+ "192.0.2.4 X",
+ "192.0.2.5\tY",
+ "192.0.2.6\nZ",
+ "192.0.2. ",
+ "192.0.2.\t",
+ "192.0.2.\n",
+ "192.0.2. X",
+ "192.0.2.\tY",
+ "192.0.2.\nZ",
+ "2001:db8::1 ",
+ "2001:db8::2\t",
+ "2001:db8::3\n",
+ "2001:db8::4 X",
+ "2001:db8::5\tY",
+ "2001:db8::6\nZ",
+ };
+ for (size_t query_idx = 0; query_idx < array_length (queries); ++query_idx)
+ {
+ const char *query = queries[query_idx];
+ struct hostent storage;
+ char buf[4096];
+ struct hostent *e;
+
+ h_errno = 0;
+ TEST_VERIFY (gethostbyname (query) == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ h_errno = 0;
+ e = NULL;
+ TEST_COMPARE (gethostbyname_r (query, &storage, buf, sizeof (buf),
+ &e, &h_errno), 0);
+ TEST_VERIFY (e == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ h_errno = 0;
+ TEST_VERIFY (gethostbyname2 (query, AF_INET) == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ h_errno = 0;
+ e = NULL;
+ TEST_COMPARE (gethostbyname2_r (query, AF_INET,
+ &storage, buf, sizeof (buf),
+ &e, &h_errno), 0);
+ TEST_VERIFY (e == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ h_errno = 0;
+ TEST_VERIFY (gethostbyname2 (query, AF_INET6) == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ h_errno = 0;
+ e = NULL;
+ TEST_COMPARE (gethostbyname2_r (query, AF_INET6,
+ &storage, buf, sizeof (buf),
+ &e, &h_errno), 0);
+ TEST_VERIFY (e == NULL);
+ TEST_COMPARE (h_errno, HOST_NOT_FOUND);
+
+ static const int gai_flags[] =
+ {
+ 0,
+ AI_ADDRCONFIG,
+ AI_NUMERICHOST,
+ AI_IDN,
+ AI_IDN | AI_NUMERICHOST,
+ AI_V4MAPPED,
+ AI_V4MAPPED | AI_NUMERICHOST,
+ };
+ for (size_t gai_flags_idx; gai_flags_idx < array_length (gai_flags);
+ ++gai_flags_idx)
+ {
+ struct addrinfo hints = { .ai_flags = gai_flags[gai_flags_idx], };
+ struct addrinfo *ai;
+ hints.ai_family = AF_INET;
+ TEST_COMPARE (getaddrinfo (query, "80", &hints, &ai), EAI_NONAME);
+ hints.ai_family = AF_INET6;
+ TEST_COMPARE (getaddrinfo (query, "80", &hints, &ai), EAI_NONAME);
+ hints.ai_family = AF_UNSPEC;
+ TEST_COMPARE (getaddrinfo (query, "80", &hints, &ai), EAI_NONAME);
+ }
+ };
+
+ resolv_test_end (aux);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
#include <unistd.h>
#include <sys/stat.h>
#include <sys/types.h>
+#include <support/blob_repeat.h>
+#include <support/check.h>
+#include <support/support.h>
+#include <support/temp_file.h>
#include <support/test-driver.h>
#include <libc-diag.h>
static int
do_test (void)
{
- const char dir[] = "bz22786";
- const char lnk[] = "bz22786/symlink";
-
- rmdir (dir);
- if (mkdir (dir, 0755) != 0 && errno != EEXIST)
- {
- printf ("mkdir %s: %m\n", dir);
- return EXIT_FAILURE;
- }
- if (symlink (".", lnk) != 0 && errno != EEXIST)
- {
- printf ("symlink (%s, %s): %m\n", dir, lnk);
- return EXIT_FAILURE;
- }
-
- const size_t path_len = (size_t) INT_MAX + 1;
-
- DIAG_PUSH_NEEDS_COMMENT;
-#if __GNUC_PREREQ (7, 0)
- /* GCC 7 warns about too-large allocations; here we need such
- allocation to succeed for the test to work. */
- DIAG_IGNORE_NEEDS_COMMENT (7, "-Walloc-size-larger-than=");
-#endif
- char *path = malloc (path_len);
- DIAG_POP_NEEDS_COMMENT;
+ char *dir = support_create_temp_directory ("bz22786.");
+ char *lnk = xasprintf ("%s/symlink", dir);
+ const size_t path_len = (size_t) INT_MAX + strlen (lnk) + 1;
+ struct support_blob_repeat repeat
+ = support_blob_repeat_allocate ("a", 1, path_len);
+ char *path = repeat.start;
if (path == NULL)
{
- printf ("malloc (%zu): %m\n", path_len);
+ printf ("Repeated allocation (%zu bytes): %m\n", path_len);
+ /* On 31-bit s390 the malloc will always fail as we do not have
+ so much memory, and we want to mark the test unsupported.
+ Likewise on systems with little physical memory the test will
+ fail and should be unsupported. */
return EXIT_UNSUPPORTED;
}
- /* Construct very long path = "bz22786/symlink/aaaa....." */
- char *p = mempcpy (path, lnk, sizeof (lnk) - 1);
+ TEST_VERIFY_EXIT (symlink (".", lnk) == 0);
+
+ /* Construct very long path = "/tmp/bz22786.XXXX/symlink/aaaa....." */
+ char *p = mempcpy (path, lnk, strlen (lnk));
*(p++) = '/';
- memset (p, 'a', path_len - (path - p) - 2);
- p[path_len - (path - p) - 1] = '\0';
+ p[path_len - (p - path) - 1] = '\0';
/* This call crashes before the fix for bz22786 on 32-bit platforms. */
p = realpath (path, NULL);
/* Cleanup. */
unlink (lnk);
- rmdir (dir);
+ support_blob_repeat_free (&repeat);
+ free (lnk);
+ free (dir);
return 0;
}
}
static void
-f1 (void)
+f1b (void)
{
- puts ("start f1");
- if (getcontext (&ctx[2]) != 0)
- {
- printf ("%s: getcontext: %m\n", __FUNCTION__);
- exit (EXIT_FAILURE);
- }
if (done)
{
- puts ("set context in f1");
+ puts ("set context in f1b");
if (setcontext (&ctx[3]) != 0)
{
printf ("%s: setcontext: %m\n", __FUNCTION__);
exit (EXIT_FAILURE);
}
}
+ exit (EXIT_FAILURE);
+}
+
+static void
+f1a (void)
+{
+ static char st2[32768];
+ puts ("start f1a");
+ if (getcontext (&ctx[2]) != 0)
+ {
+ printf ("%s: getcontext: %m\n", __FUNCTION__);
+ exit (EXIT_FAILURE);
+ }
+ ctx[2].uc_stack.ss_sp = st2;
+ ctx[2].uc_stack.ss_size = sizeof st2;
+ ctx[2].uc_link = &ctx[0];
+ makecontext (&ctx[2], (void (*) (void)) f1b, 0);
f2 ();
}
+/* The execution path through the test looks like this:
+ do_test (call)
+ -> "making contexts"
+ -> "swap contexts"
+ f1a (via swapcontext to ctx[1], with alternate stack)
+ -> "start f1a"
+ f2 (call)
+ -> "swap contexts in f2"
+ f1b (via swapcontext to ctx[2], with alternate stack)
+ -> "set context in f1b"
+ do_test (via setcontext to ctx[3], main stack)
+ -> "setcontext"
+ f2 (via setcontext to ctx[4], with alternate stack)
+ -> "end f2"
+
+ We must use an alternate stack for f1b, because if we don't then the
+ result of executing an earlier caller may overwrite registers
+ spilled to the stack in f2. */
static int
do_test (void)
{
- char st1[32768];
+ static char st1[32768];
puts ("making contexts");
if (getcontext (&ctx[0]) != 0)
{
ctx[1].uc_stack.ss_sp = st1;
ctx[1].uc_stack.ss_size = sizeof st1;
ctx[1].uc_link = &ctx[0];
- makecontext (&ctx[1], (void (*) (void)) f1, 0);
+ makecontext (&ctx[1], (void (*) (void)) f1a, 0);
puts ("swap contexts");
if (swapcontext (&ctx[3], &ctx[1]) != 0)
{
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
+#include <support/blob_repeat.h>
+#include <support/test-driver.h>
#define EXPONENT "e-2147483649"
#define SIZE 214748364
static int
do_test (void)
{
- char *p = malloc (1 + SIZE + sizeof (EXPONENT));
- if (p == NULL)
+ struct support_blob_repeat repeat = support_blob_repeat_allocate
+ ("0", 1, 1 + SIZE + sizeof (EXPONENT));
+ if (repeat.size == 0)
{
- puts ("malloc failed, cannot test for overflow");
- return 0;
+ puts ("warning: memory allocation failed, cannot test for overflow");
+ return EXIT_UNSUPPORTED;
}
+ char *p = repeat.start;
p[0] = '1';
- memset (p + 1, '0', SIZE);
memcpy (p + 1 + SIZE, EXPONENT, sizeof (EXPONENT));
double d = strtod (p, NULL);
if (d != 0)
{
- printf ("strtod returned wrong value: %a\n", d);
+ printf ("error: strtod returned wrong value: %a\n", d);
return 1;
}
+ support_blob_repeat_free (&repeat);
return 0;
}
# This test allocates a lot of memory and can run for a long time.
xtests = tst-strcoll-overflow
+# This test needs libdl.
+ifeq (yes,$(build-shared))
+tests += test-strerror-errno
+LDLIBS-test-strerror-errno = $(libdl)
+endif
+
ifeq ($(run-built-tests),yes)
tests-special += $(objpfx)tst-svc-cmp.out
endif
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-/* This particular implementation was written by Eric Blake, 2008. */
-
#ifndef _LIBC
# include <config.h>
#endif
-/* Specification of memmem. */
#include <string.h>
#ifndef _LIBC
-# define __builtin_expect(expr, val) (expr)
# define __memmem memmem
#endif
#undef memmem
-/* Return the first occurrence of NEEDLE in HAYSTACK. Return HAYSTACK
- if NEEDLE_LEN is 0, otherwise NULL if NEEDLE is not found in
- HAYSTACK. */
+/* Hash character pairs so a small shift table can be used. All bits of
+ p[0] are included, but not all bits from p[-1]. So if two equal hashes
+ match on p[-1], p[0] matches too. Hash collisions are harmless and result
+ in smaller shifts. */
+#define hash2(p) (((size_t)(p)[0] - ((size_t)(p)[-1] << 3)) % sizeof (shift))
+
+/* Fast memmem algorithm with guaranteed linear-time performance.
+ Small needles up to size 2 use a dedicated linear search. Longer needles
+ up to size 256 use a novel modified Horspool algorithm. It hashes pairs
+ of characters to quickly skip past mismatches. The main search loop only
+ exits if the last 2 characters match, avoiding unnecessary calls to memcmp
+ and allowing for a larger skip if there is no match. A self-adapting
+ filtering check is used to quickly detect mismatches in long needles.
+ By limiting the needle length to 256, the shift table can be reduced to 8
+ bits per entry, lowering preprocessing overhead and minimizing cache effects.
+ The limit also implies worst-case performance is linear.
+ Needles larger than 256 characters use the linear-time Two-Way algorithm. */
void *
-__memmem (const void *haystack_start, size_t haystack_len,
- const void *needle_start, size_t needle_len)
+__memmem (const void *haystack, size_t hs_len,
+ const void *needle, size_t ne_len)
{
- /* Abstract memory is considered to be an array of 'unsigned char' values,
- not an array of 'char' values. See ISO C 99 section 6.2.6.1. */
- const unsigned char *haystack = (const unsigned char *) haystack_start;
- const unsigned char *needle = (const unsigned char *) needle_start;
-
- if (needle_len == 0)
- /* The first occurrence of the empty string is deemed to occur at
- the beginning of the string. */
- return (void *) haystack;
-
- /* Sanity check, otherwise the loop might search through the whole
- memory. */
- if (__glibc_unlikely (haystack_len < needle_len))
+ const unsigned char *hs = (const unsigned char *) haystack;
+ const unsigned char *ne = (const unsigned char *) needle;
+
+ if (ne_len == 0)
+ return (void *) hs;
+ if (ne_len == 1)
+ return (void *) memchr (hs, ne[0], hs_len);
+
+ /* Ensure haystack length is >= needle length. */
+ if (hs_len < ne_len)
return NULL;
- /* Use optimizations in memchr when possible, to reduce the search
- size of haystack using a linear algorithm with a smaller
- coefficient. However, avoid memchr for long needles, since we
- can often achieve sublinear performance. */
- if (needle_len < LONG_NEEDLE_THRESHOLD)
+ const unsigned char *end = hs + hs_len - ne_len;
+
+ if (ne_len == 2)
+ {
+ uint32_t nw = ne[0] << 16 | ne[1], hw = hs[0] << 16 | hs[1];
+ for (hs++; hs <= end && hw != nw; )
+ hw = hw << 16 | *++hs;
+ return hw == nw ? (void *)hs - 1 : NULL;
+ }
+
+ /* Use Two-Way algorithm for very long needles. */
+ if (__builtin_expect (ne_len > 256, 0))
+ return two_way_long_needle (hs, hs_len, ne, ne_len);
+
+ uint8_t shift[256];
+ size_t tmp, shift1;
+ size_t m1 = ne_len - 1;
+ size_t offset = 0;
+
+ memset (shift, 0, sizeof (shift));
+ for (int i = 1; i < m1; i++)
+ shift[hash2 (ne + i)] = i;
+ /* Shift1 is the amount we can skip after matching the hash of the
+ needle end but not the full needle. */
+ shift1 = m1 - shift[hash2 (ne + m1)];
+ shift[hash2 (ne + m1)] = m1;
+
+ for ( ; hs <= end; )
{
- haystack = memchr (haystack, *needle, haystack_len);
- if (!haystack || __builtin_expect (needle_len == 1, 0))
- return (void *) haystack;
- haystack_len -= haystack - (const unsigned char *) haystack_start;
- if (haystack_len < needle_len)
- return NULL;
- return two_way_short_needle (haystack, haystack_len, needle, needle_len);
+ /* Skip past character pairs not in the needle. */
+ do
+ {
+ hs += m1;
+ tmp = shift[hash2 (hs)];
+ }
+ while (tmp == 0 && hs <= end);
+
+ /* If the match is not at the end of the needle, shift to the end
+ and continue until we match the hash of the needle end. */
+ hs -= tmp;
+ if (tmp < m1)
+ continue;
+
+ /* Hash of the last 2 characters matches. If the needle is long,
+ try to quickly filter out mismatches. */
+ if (m1 < 15 || memcmp (hs + offset, ne + offset, 8) == 0)
+ {
+ if (memcmp (hs, ne, m1) == 0)
+ return (void *) hs;
+
+ /* Adjust filter offset when it doesn't find the mismatch. */
+ offset = (offset >= 8 ? offset : m1) - 8;
+ }
+
+ /* Skip based on matching the hash of the needle end. */
+ hs += shift1;
}
- else
- return two_way_long_needle (haystack, haystack_len, needle, needle_len);
+ return NULL;
}
libc_hidden_def (__memmem)
weak_alias (__memmem, memmem)
libc_hidden_weak (memmem)
-
-#undef LONG_NEEDLE_THRESHOLD
most 2 * HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching.
If AVAILABLE modifies HAYSTACK_LEN (as in strstr), then at most 3 *
HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching. */
-static RETURN_TYPE
+static inline RETURN_TYPE
two_way_short_needle (const unsigned char *haystack, size_t haystack_len,
const unsigned char *needle, size_t needle_len)
{
and sublinear performance O(HAYSTACK_LEN / NEEDLE_LEN) is possible.
If AVAILABLE modifies HAYSTACK_LEN (as in strstr), then at most 3 *
HAYSTACK_LEN - NEEDLE_LEN comparisons occur in searching, and
- sublinear performance is not possible. */
-static RETURN_TYPE
+ sublinear performance is not possible.
+
+ Since this function is large and complex, block inlining to avoid
+ slowing down the common case of small needles. */
+__attribute__((noinline)) static RETURN_TYPE
two_way_long_needle (const unsigned char *haystack, size_t haystack_len,
const unsigned char *needle, size_t needle_len)
{
/* Two-Way algorithm. */
#define RETURN_TYPE char *
#define AVAILABLE(h, h_l, j, n_l) \
- (((j) + (n_l) <= (h_l)) || ((h_l) += __strnlen ((void*)((h) + (h_l)), 512), \
- (j) + (n_l) <= (h_l)))
+ (((j) + (n_l) <= (h_l)) \
+ || ((h_l) += __strnlen ((void*)((h) + (h_l)), (n_l) + 512), \
+ (j) + (n_l) <= (h_l)))
#define CHECK_EOL (1)
#define RET0_IF_0(a) if (!a) goto ret0
#define CANON_ELEMENT(c) TOLOWER (c)
case-insensitive comparison. This function gives unspecified
results in multibyte locales. */
char *
-STRCASESTR (const char *haystack_start, const char *needle_start)
+STRCASESTR (const char *haystack, const char *needle)
{
- const char *haystack = haystack_start;
- const char *needle = needle_start;
size_t needle_len; /* Length of NEEDLE. */
size_t haystack_len; /* Known minimum length of HAYSTACK. */
- bool ok = true; /* True if NEEDLE is prefix of HAYSTACK. */
-
- /* Determine length of NEEDLE, and in the process, make sure
- HAYSTACK is at least as long (no point processing all of a long
- NEEDLE if HAYSTACK is too short). */
- while (*haystack && *needle)
- {
- ok &= (TOLOWER ((unsigned char) *haystack)
- == TOLOWER ((unsigned char) *needle));
- haystack++;
- needle++;
- }
- if (*needle)
+
+ /* Handle empty NEEDLE special case. */
+ if (needle[0] == '\0')
+ return (char *) haystack;
+
+ /* Ensure HAYSTACK length is at least as long as NEEDLE length.
+ Since a match may occur early on in a huge HAYSTACK, use strnlen
+ and read ahead a few cachelines for improved performance. */
+ needle_len = strlen (needle);
+ haystack_len = __strnlen (haystack, needle_len + 256);
+ if (haystack_len < needle_len)
return NULL;
- if (ok)
- return (char *) haystack_start;
- needle_len = needle - needle_start;
- haystack = haystack_start + 1;
- haystack_len = needle_len - 1;
/* Perform the search. Abstract memory is considered to be an array
of 'unsigned char' values, not an array of 'char' values. See
if (needle_len < LONG_NEEDLE_THRESHOLD)
return two_way_short_needle ((const unsigned char *) haystack,
haystack_len,
- (const unsigned char *) needle_start,
+ (const unsigned char *) needle,
needle_len);
return two_way_long_needle ((const unsigned char *) haystack, haystack_len,
- (const unsigned char *) needle_start,
+ (const unsigned char *) needle,
needle_len);
}
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
-/* This particular implementation was written by Eric Blake, 2008. */
-
#ifndef _LIBC
# include <config.h>
#endif
-/* Specification of strstr. */
#include <string.h>
-#include <stdbool.h>
-
-#ifndef _LIBC
-# define __builtin_expect(expr, val) (expr)
-#endif
-
#define RETURN_TYPE char *
#define AVAILABLE(h, h_l, j, n_l) \
- (((j) + (n_l) <= (h_l)) || ((h_l) += __strnlen ((void*)((h) + (h_l)), 512), \
- (j) + (n_l) <= (h_l)))
-#define CHECK_EOL (1)
-#define RET0_IF_0(a) if (!a) goto ret0
-#define FASTSEARCH(S,C,N) (void*) strchr ((void*)(S), (C))
+ (((j) + (n_l) <= (h_l)) \
+ || ((h_l) += __strnlen ((void*)((h) + (h_l)), (n_l) + 512), \
+ (j) + (n_l) <= (h_l)))
#include "str-two-way.h"
#undef strstr
#define STRSTR strstr
#endif
-/* Return the first occurrence of NEEDLE in HAYSTACK. Return HAYSTACK
- if NEEDLE is empty, otherwise NULL if NEEDLE is not found in
- HAYSTACK. */
+static inline char *
+strstr2 (const unsigned char *hs, const unsigned char *ne)
+{
+ uint32_t h1 = (ne[0] << 16) | ne[1];
+ uint32_t h2 = 0;
+ for (int c = hs[0]; h1 != h2 && c != 0; c = *++hs)
+ h2 = (h2 << 16) | c;
+ return h1 == h2 ? (char *)hs - 2 : NULL;
+}
+
+static inline char *
+strstr3 (const unsigned char *hs, const unsigned char *ne)
+{
+ uint32_t h1 = ((uint32_t)ne[0] << 24) | (ne[1] << 16) | (ne[2] << 8);
+ uint32_t h2 = 0;
+ for (int c = hs[0]; h1 != h2 && c != 0; c = *++hs)
+ h2 = (h2 | c) << 8;
+ return h1 == h2 ? (char *)hs - 3 : NULL;
+}
+
+/* Hash character pairs so a small shift table can be used. All bits of
+ p[0] are included, but not all bits from p[-1]. So if two equal hashes
+ match on p[-1], p[0] matches too. Hash collisions are harmless and result
+ in smaller shifts. */
+#define hash2(p) (((size_t)(p)[0] - ((size_t)(p)[-1] << 3)) % sizeof (shift))
+
+/* Fast strstr algorithm with guaranteed linear-time performance.
+ Small needles up to size 3 use a dedicated linear search. Longer needles
+ up to size 256 use a novel modified Horspool algorithm. It hashes pairs
+ of characters to quickly skip past mismatches. The main search loop only
+ exits if the last 2 characters match, avoiding unnecessary calls to memcmp
+ and allowing for a larger skip if there is no match. A self-adapting
+ filtering check is used to quickly detect mismatches in long needles.
+ By limiting the needle length to 256, the shift table can be reduced to 8
+ bits per entry, lowering preprocessing overhead and minimizing cache effects.
+ The limit also implies worst-case performance is linear.
+ Needles larger than 256 characters use the linear-time Two-Way algorithm. */
char *
-STRSTR (const char *haystack_start, const char *needle_start)
+STRSTR (const char *haystack, const char *needle)
{
- const char *haystack = haystack_start;
- const char *needle = needle_start;
- size_t needle_len; /* Length of NEEDLE. */
- size_t haystack_len; /* Known minimum length of HAYSTACK. */
- bool ok = true; /* True if NEEDLE is prefix of HAYSTACK. */
-
- /* Determine length of NEEDLE, and in the process, make sure
- HAYSTACK is at least as long (no point processing all of a long
- NEEDLE if HAYSTACK is too short). */
- while (*haystack && *needle)
- ok &= *haystack++ == *needle++;
- if (*needle)
+ const unsigned char *hs = (const unsigned char *) haystack;
+ const unsigned char *ne = (const unsigned char *) needle;
+
+ /* Handle short needle special cases first. */
+ if (ne[0] == '\0')
+ return (char *)hs;
+ hs = (const unsigned char *)strchr ((const char*)hs, ne[0]);
+ if (hs == NULL || ne[1] == '\0')
+ return (char*)hs;
+ if (ne[2] == '\0')
+ return strstr2 (hs, ne);
+ if (ne[3] == '\0')
+ return strstr3 (hs, ne);
+
+ /* Ensure haystack length is at least as long as needle length.
+ Since a match may occur early on in a huge haystack, use strnlen
+ and read ahead a few cachelines for improved performance. */
+ size_t ne_len = strlen ((const char*)ne);
+ size_t hs_len = __strnlen ((const char*)hs, ne_len | 512);
+ if (hs_len < ne_len)
return NULL;
- if (ok)
- return (char *) haystack_start;
-
- /* Reduce the size of haystack using strchr, since it has a smaller
- linear coefficient than the Two-Way algorithm. */
- needle_len = needle - needle_start;
- haystack = strchr (haystack_start + 1, *needle_start);
- if (!haystack || __builtin_expect (needle_len == 1, 0))
- return (char *) haystack;
- needle -= needle_len;
- haystack_len = (haystack > haystack_start + needle_len ? 1
- : needle_len + haystack_start - haystack);
-
- /* Perform the search. Abstract memory is considered to be an array
- of 'unsigned char' values, not an array of 'char' values. See
- ISO C 99 section 6.2.6.1. */
- if (needle_len < LONG_NEEDLE_THRESHOLD)
- return two_way_short_needle ((const unsigned char *) haystack,
- haystack_len,
- (const unsigned char *) needle, needle_len);
- return two_way_long_needle ((const unsigned char *) haystack, haystack_len,
- (const unsigned char *) needle, needle_len);
+
+ /* Check whether we have a match. This improves performance since we
+ avoid initialization overheads. */
+ if (memcmp (hs, ne, ne_len) == 0)
+ return (char *) hs;
+
+ /* Use Two-Way algorithm for very long needles. */
+ if (__glibc_unlikely (ne_len > 256))
+ return two_way_long_needle (hs, hs_len, ne, ne_len);
+
+ const unsigned char *end = hs + hs_len - ne_len;
+ uint8_t shift[256];
+ size_t tmp, shift1;
+ size_t m1 = ne_len - 1;
+ size_t offset = 0;
+
+ /* Initialize bad character shift hash table. */
+ memset (shift, 0, sizeof (shift));
+ for (int i = 1; i < m1; i++)
+ shift[hash2 (ne + i)] = i;
+ /* Shift1 is the amount we can skip after matching the hash of the
+ needle end but not the full needle. */
+ shift1 = m1 - shift[hash2 (ne + m1)];
+ shift[hash2 (ne + m1)] = m1;
+
+ while (1)
+ {
+ if (__glibc_unlikely (hs > end))
+ {
+ end += __strnlen ((const char*)end + m1 + 1, 2048);
+ if (hs > end)
+ return NULL;
+ }
+
+ /* Skip past character pairs not in the needle. */
+ do
+ {
+ hs += m1;
+ tmp = shift[hash2 (hs)];
+ }
+ while (tmp == 0 && hs <= end);
+
+ /* If the match is not at the end of the needle, shift to the end
+ and continue until we match the hash of the needle end. */
+ hs -= tmp;
+ if (tmp < m1)
+ continue;
+
+ /* Hash of the last 2 characters matches. If the needle is long,
+ try to quickly filter out mismatches. */
+ if (m1 < 15 || memcmp (hs + offset, ne + offset, 8) == 0)
+ {
+ if (memcmp (hs, ne, m1) == 0)
+ return (void *) hs;
+
+ /* Adjust filter offset when it doesn't find the mismatch. */
+ offset = (offset >= 8 ? offset : m1) - 8;
+ }
+
+ /* Skip based on matching the hash of the needle end. */
+ hs += shift1;
+ }
}
libc_hidden_builtin_def (strstr)
-
-#undef LONG_NEEDLE_THRESHOLD
--- /dev/null
+/* BZ #24024 strerror and errno test.
+
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <dlfcn.h>
+#include <errno.h>
+#include <string.h>
+
+#include <support/check.h>
+#include <support/support.h>
+
+/* malloc is allowed to change errno to a value different than 0, even when
+ there is no actual error. This happens for example when the memory
+ allocation through sbrk fails. Simulate this by interposing our own
+ malloc implementation which sets errno to ENOMEM and calls the original
+ malloc. */
+void
+*malloc (size_t size)
+{
+ static void *(*real_malloc) (size_t size);
+
+ if (!real_malloc)
+ real_malloc = dlsym (RTLD_NEXT, "malloc");
+
+ errno = ENOMEM;
+
+ return (*real_malloc) (size);
+}
+
+/* strerror must not change the value of errno. Unfortunately due to GCC bug
+ #88576, this happens when -fmath-errno is used. This simple test checks
+ that it doesn't happen. */
+static int
+do_test (void)
+{
+ char *msg;
+
+ errno = 0;
+ msg = strerror (-3);
+ (void) msg;
+ TEST_COMPARE (errno, 0);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
}
}
+#define N 1024
+
+static void
+pr23637 (void)
+{
+ char *h = (char*) buf1;
+ char *n = (char*) buf2;
+
+ for (int i = 0; i < N; i++)
+ {
+ n[i] = 'x';
+ h[i] = ' ';
+ h[i + N] = 'x';
+ }
+
+ n[N] = '\0';
+ h[N * 2] = '\0';
+
+ /* Ensure we don't match at the first 'x'. */
+ h[0] = 'x';
+
+ char *exp_result = stupid_strstr (h, n);
+ FOR_EACH_IMPL (impl, 0)
+ check_result (impl, h, n, exp_result);
+}
+
static int
test_main (void)
{
check1 ();
check2 ();
+ pr23637 ();
printf ("%23s", "");
FOR_EACH_IMPL (impl, 0)
do_test (15, 9, hlen, klen, 1);
do_test (15, 15, hlen, klen, 0);
do_test (15, 15, hlen, klen, 1);
+
+ do_test (15, 15, hlen + klen * 4, klen * 4, 0);
+ do_test (15, 15, hlen + klen * 4, klen * 4, 1);
}
do_test (0, 0, page_size - 1, 16, 0);
extra-libs-noinstall := $(extra-libs)
libsupport-routines = \
+ blob_repeat \
check \
check_addrinfo \
check_dns_packet \
support_capture_subprocess \
support_capture_subprocess_check \
support_chroot \
+ support_descriptors \
support_enter_mount_namespace \
support_enter_network_namespace \
support_format_address_family \
support_isolate_in_subprocess \
support_openpty \
support_quote_blob \
+ support_quote_string \
support_record_failure \
support_run_diff \
support_shared_allocate \
xpthread_mutexattr_settype \
xpthread_once \
xpthread_rwlock_init \
+ xpthread_rwlock_destroy \
xpthread_rwlock_rdlock \
xpthread_rwlock_unlock \
xpthread_rwlock_wrlock \
tests = \
README-testing \
tst-support-namespace \
+ tst-support_blob_repeat \
tst-support_capture_subprocess \
+ tst-support_descriptors \
tst-support_format_dns_packet \
tst-support_quote_blob \
+ tst-support_quote_string \
tst-support_record_failure \
tst-test_compare \
tst-test_compare_blob \
--- /dev/null
+/* Repeating a memory blob, with alias mapping optimization.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/blob_repeat.h>
+#include <support/check.h>
+#include <support/test-driver.h>
+#include <support/support.h>
+#include <support/xunistd.h>
+#include <sys/mman.h>
+#include <unistd.h>
+#include <wchar.h>
+
+/* Small allocations should use malloc directly instead of the mmap
+ optimization because mappings carry a lot of overhead. */
+static const size_t maximum_small_size = 4 * 1024 * 1024;
+
+/* Set *RESULT to LEFT * RIGHT. Return true if the multiplication
+ overflowed. See <malloc/malloc-internal.h>. */
+static inline bool
+check_mul_overflow_size_t (size_t left, size_t right, size_t *result)
+{
+#if __GNUC__ >= 5
+ return __builtin_mul_overflow (left, right, result);
+#else
+ /* size_t is unsigned so the behavior on overflow is defined. */
+ *result = left * right;
+ size_t half_size_t = ((size_t) 1) << (8 * sizeof (size_t) / 2);
+ if (__glibc_unlikely ((left | right) >= half_size_t))
+ {
+ if (__glibc_unlikely (right != 0 && *result / right != left))
+ return true;
+ }
+ return false;
+#endif
+}
+
+/* Internal helper for fill. */
+static void
+fill0 (char *target, const char *element, size_t element_size,
+ size_t count)
+{
+ while (count > 0)
+ {
+ memcpy (target, element, element_size);
+ target += element_size;
+ --count;
+ }
+}
+
+/* Fill the buffer at TARGET with COUNT copies of the ELEMENT_SIZE
+ bytes starting at ELEMENT. */
+static void
+fill (char *target, const char *element, size_t element_size,
+ size_t count)
+{
+ if (element_size == 0 || count == 0)
+ return;
+ else if (element_size == 1)
+ memset (target, element[0], count);
+ else if (element_size == sizeof (wchar_t))
+ {
+ wchar_t wc;
+ memcpy (&wc, element, sizeof (wc));
+ wmemset ((wchar_t *) target, wc, count);
+ }
+ else if (element_size < 1024 && count > 4096)
+ {
+ /* Use larger copies for really small element sizes. */
+ char buffer[8192];
+ size_t buffer_count = sizeof (buffer) / element_size;
+ fill0 (buffer, element, element_size, buffer_count);
+ while (count > 0)
+ {
+ size_t copy_count = buffer_count;
+ if (copy_count > count)
+ copy_count = count;
+ size_t copy_bytes = copy_count * element_size;
+ memcpy (target, buffer, copy_bytes);
+ target += copy_bytes;
+ count -= copy_count;
+ }
+ }
+ else
+ fill0 (target, element, element_size, count);
+}
+
+/* Use malloc instead of mmap for small allocations and unusual size
+ combinations. */
+static struct support_blob_repeat
+allocate_malloc (size_t total_size, const void *element, size_t element_size,
+ size_t count)
+{
+ void *buffer = malloc (total_size);
+ if (buffer == NULL)
+ return (struct support_blob_repeat) { 0 };
+ fill (buffer, element, element_size, count);
+ return (struct support_blob_repeat)
+ {
+ .start = buffer,
+ .size = total_size,
+ .use_malloc = true
+ };
+}
+
+/* Return the least common multiple of PAGE_SIZE and ELEMENT_SIZE,
+ avoiding overflow. This assumes that PAGE_SIZE is a power of
+ two. */
+static size_t
+minimum_stride_size (size_t page_size, size_t element_size)
+{
+ TEST_VERIFY_EXIT (page_size > 0);
+ TEST_VERIFY_EXIT (element_size > 0);
+
+ /* Compute the number of trailing zeros common to both sizes. */
+ unsigned int common_zeros = __builtin_ctzll (page_size | element_size);
+
+ /* In the product, this power of two appears twice, but in the least
+ common multiple, it appears only once. Therefore, shift one
+ factor. */
+ size_t multiple;
+ if (check_mul_overflow_size_t (page_size >> common_zeros, element_size,
+ &multiple))
+ return 0;
+ return multiple;
+}
+
+/* Allocations larger than maximum_small_size potentially use mmap
+ with alias mappings. */
+static struct support_blob_repeat
+allocate_big (size_t total_size, const void *element, size_t element_size,
+ size_t count)
+{
+ unsigned long page_size = xsysconf (_SC_PAGESIZE);
+ size_t stride_size = minimum_stride_size (page_size, element_size);
+ if (stride_size == 0)
+ {
+ errno = EOVERFLOW;
+ return (struct support_blob_repeat) { 0 };
+ }
+
+ /* Ensure that the stride size is at least maximum_small_size. This
+ is necessary to reduce the number of distinct mappings. */
+ if (stride_size < maximum_small_size)
+ stride_size
+ = ((maximum_small_size + stride_size - 1) / stride_size) * stride_size;
+
+ if (stride_size > total_size)
+ /* The mmap optimization would not save anything. */
+ return allocate_malloc (total_size, element, element_size, count);
+
+ /* Reserve the memory region. If we cannot create the mapping,
+ there is no reason to set up the backing file. */
+ void *target = mmap (NULL, total_size, PROT_NONE,
+ MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+ if (target == MAP_FAILED)
+ return (struct support_blob_repeat) { 0 };
+
+ /* Create the backing file for the repeated mapping. Call mkstemp
+ directly to remove the resources backing the temporary file
+ immediately, once support_blob_repeat_free is called. Using
+ create_temp_file would result in a warning during post-test
+ cleanup. */
+ int fd;
+ {
+ char *temppath = xasprintf ("%s/support_blob_repeat-XXXXXX", test_dir);
+ fd = mkstemp (temppath);
+ if (fd < 0)
+ FAIL_EXIT1 ("mkstemp (\"%s\"): %m", temppath);
+ xunlink (temppath);
+ free (temppath);
+ }
+
+ /* Make sure that there is backing storage, so that the fill
+ operation will not fault. */
+ if (posix_fallocate (fd, 0, stride_size) != 0)
+ FAIL_EXIT1 ("posix_fallocate (%zu): %m", stride_size);
+
+ /* The stride size must still be a multiple of the page size and
+ element size. */
+ TEST_VERIFY_EXIT ((stride_size % page_size) == 0);
+ TEST_VERIFY_EXIT ((stride_size % element_size) == 0);
+
+ /* Fill the backing store. */
+ {
+ void *ptr = mmap (target, stride_size, PROT_READ | PROT_WRITE,
+ MAP_FIXED | MAP_FILE | MAP_SHARED, fd, 0);
+ if (ptr == MAP_FAILED)
+ {
+ int saved_errno = errno;
+ xmunmap (target, total_size);
+ xclose (fd);
+ errno = saved_errno;
+ return (struct support_blob_repeat) { 0 };
+ }
+ if (ptr != target)
+ FAIL_EXIT1 ("mapping of %zu bytes moved from %p to %p",
+ stride_size, target, ptr);
+
+ /* Write the repeating data. */
+ fill (target, element, element_size, stride_size / element_size);
+
+ /* Return to a PROT_NONE mapping, just to be on the safe side. */
+ ptr = mmap (target, stride_size, PROT_NONE,
+ MAP_FIXED | MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+ if (ptr == MAP_FAILED)
+ FAIL_EXIT1 ("Failed to reinstate PROT_NONE mapping: %m");
+ if (ptr != target)
+ FAIL_EXIT1 ("PROT_NONE mapping of %zu bytes moved from %p to %p",
+ stride_size, target, ptr);
+ }
+
+ /* Create the alias mappings. */
+ {
+ size_t remaining_size = total_size;
+ char *current = target;
+ int flags = MAP_FIXED | MAP_FILE | MAP_PRIVATE;
+#ifdef MAP_NORESERVE
+ flags |= MAP_NORESERVE;
+#endif
+ while (remaining_size > 0)
+ {
+ size_t to_map = stride_size;
+ if (to_map > remaining_size)
+ to_map = remaining_size;
+ void *ptr = mmap (current, to_map, PROT_READ | PROT_WRITE,
+ flags, fd, 0);
+ if (ptr == MAP_FAILED)
+ {
+ int saved_errno = errno;
+ xmunmap (target, total_size);
+ xclose (fd);
+ errno = saved_errno;
+ return (struct support_blob_repeat) { 0 };
+ }
+ if (ptr != current)
+ FAIL_EXIT1 ("MAP_PRIVATE mapping of %zu bytes moved from %p to %p",
+ to_map, target, ptr);
+ remaining_size -= to_map;
+ current += to_map;
+ }
+ }
+
+ xclose (fd);
+
+ return (struct support_blob_repeat)
+ {
+ .start = target,
+ .size = total_size,
+ .use_malloc = false
+ };
+}
+
+struct support_blob_repeat
+support_blob_repeat_allocate (const void *element, size_t element_size,
+ size_t count)
+{
+ size_t total_size;
+ if (check_mul_overflow_size_t (element_size, count, &total_size))
+ {
+ errno = EOVERFLOW;
+ return (struct support_blob_repeat) { 0 };
+ }
+ if (total_size <= maximum_small_size)
+ return allocate_malloc (total_size, element, element_size, count);
+ else
+ return allocate_big (total_size, element, element_size, count);
+}
+
+void
+support_blob_repeat_free (struct support_blob_repeat *blob)
+{
+ if (blob->size > 0)
+ {
+ int saved_errno = errno;
+ if (blob->use_malloc)
+ free (blob->start);
+ else
+ xmunmap (blob->start, blob->size);
+ errno = saved_errno;
+ }
+ *blob = (struct support_blob_repeat) { 0 };
+}
--- /dev/null
+/* Repeating a memory blob, with alias mapping optimization.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifndef SUPPORT_BLOB_REPEAT_H
+#define SUPPORT_BLOB_REPEAT_H
+
+#include <stdbool.h>
+#include <stddef.h>
+
+struct support_blob_repeat
+{
+ void *start;
+ size_t size;
+ bool use_malloc;
+};
+
+/* Return an allocation of COUNT elements, each of ELEMENT_SIZE bytes,
+ initialized with the bytes starting at ELEMENT. The memory is
+ writable (and thus counts towards the commit charge). In case of
+ on error, all members of the return struct are zero-initialized,
+ and errno is set accordingly. */
+struct support_blob_repeat support_blob_repeat_allocate (const void *element,
+ size_t element_size,
+ size_t count);
+
+/* Deallocate the blob created by support_blob_repeat_allocate. */
+void support_blob_repeat_free (struct support_blob_repeat *);
+
+#endif /* SUPPORT_BLOB_REPEAT_H */
/* Internal function used to test the failure recording framework. */
void support_record_failure_reset (void);
+/* Returns true or false depending on whether there have been test
+ failures or not. */
+int support_record_failure_is_failed (void);
+
__END_DECLS
#endif /* SUPPORT_CHECK_H */
--- /dev/null
+/* Monitoring file descriptor usage.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifndef SUPPORT_DESCRIPTORS_H
+#define SUPPORT_DESCRIPTORS_H
+
+#include <stdio.h>
+
+/* Opaque pointer, for capturing file descriptor lists. */
+struct support_descriptors;
+
+/* Record the currently open file descriptors and store them in the
+ returned list. Terminate the process if the listing operation
+ fails. */
+struct support_descriptors *support_descriptors_list (void);
+
+/* Deallocate the list of descriptors. */
+void support_descriptors_free (struct support_descriptors *);
+
+/* Write the list of descriptors to STREAM, adding PREFIX to each
+ line. */
+void support_descriptors_dump (struct support_descriptors *,
+ const char *prefix, FILE *stream);
+
+/* Check for file descriptor leaks and other file descriptor changes:
+ Compare the current list of descriptors with the passed list.
+ Record a test failure if there are additional open descriptors,
+ descriptors have been closed, or if a change in file descriptor can
+ be detected. */
+void support_descriptors_check (struct support_descriptors *);
+
+#endif /* SUPPORT_DESCRIPTORS_H */
the result). */
char *support_quote_blob (const void *blob, size_t length);
+/* Quote the contents of the string, in such a way that the result
+ string can be included in a C literal (in single/double quotes,
+ without putting the quotes into the result). */
+char *support_quote_string (const char *);
+
/* Error-checking wrapper functions which terminate the process on
error. */
int stdout_pipe[2];
xpipe (stdout_pipe);
+ TEST_VERIFY (stdout_pipe[0] > STDERR_FILENO);
+ TEST_VERIFY (stdout_pipe[1] > STDERR_FILENO);
int stderr_pipe[2];
xpipe (stderr_pipe);
+ TEST_VERIFY (stderr_pipe[0] > STDERR_FILENO);
+ TEST_VERIFY (stderr_pipe[1] > STDERR_FILENO);
TEST_VERIFY (fflush (stdout) == 0);
TEST_VERIFY (fflush (stderr) == 0);
xclose (stderr_pipe[0]);
xdup2 (stdout_pipe[1], STDOUT_FILENO);
xdup2 (stderr_pipe[1], STDERR_FILENO);
+ xclose (stdout_pipe[1]);
+ xclose (stderr_pipe[1]);
callback (closure);
_exit (0);
}
--- /dev/null
+/* Monitoring file descriptor usage.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <dirent.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/support.h>
+#include <sys/stat.h>
+#include <sys/sysmacros.h>
+#include <xunistd.h>
+
+struct procfs_descriptor
+{
+ int fd;
+ char *link_target;
+ dev_t dev;
+ ino64_t ino;
+};
+
+/* Used with qsort. */
+static int
+descriptor_compare (const void *l, const void *r)
+{
+ const struct procfs_descriptor *left = l;
+ const struct procfs_descriptor *right = r;
+ /* Cannot overflow due to limited file descriptor range. */
+ return left->fd - right->fd;
+}
+
+#define DYNARRAY_STRUCT descriptor_list
+#define DYNARRAY_ELEMENT struct procfs_descriptor
+#define DYNARRAY_PREFIX descriptor_list_
+#define DYNARRAY_ELEMENT_FREE(e) free ((e)->link_target)
+#define DYNARRAY_INITIAL_SIZE 0
+#include <malloc/dynarray-skeleton.c>
+
+struct support_descriptors
+{
+ struct descriptor_list list;
+};
+
+struct support_descriptors *
+support_descriptors_list (void)
+{
+ struct support_descriptors *result = xmalloc (sizeof (*result));
+ descriptor_list_init (&result->list);
+
+ DIR *fds = opendir ("/proc/self/fd");
+ if (fds == NULL)
+ FAIL_EXIT1 ("opendir (\"/proc/self/fd\"): %m");
+
+ while (true)
+ {
+ errno = 0;
+ struct dirent64 *e = readdir64 (fds);
+ if (e == NULL)
+ {
+ if (errno != 0)
+ FAIL_EXIT1 ("readdir: %m");
+ break;
+ }
+
+ if (e->d_name[0] == '.')
+ continue;
+
+ char *endptr;
+ long int fd = strtol (e->d_name, &endptr, 10);
+ if (*endptr != '\0' || fd < 0 || fd > INT_MAX)
+ FAIL_EXIT1 ("readdir: invalid file descriptor name: /proc/self/fd/%s",
+ e->d_name);
+
+ /* Skip the descriptor which is used to enumerate the
+ descriptors. */
+ if (fd == dirfd (fds))
+ continue;
+
+ char *target;
+ {
+ char *path = xasprintf ("/proc/self/fd/%ld", fd);
+ target = xreadlink (path);
+ free (path);
+ }
+ struct stat64 st;
+ if (fstat64 (fd, &st) != 0)
+ FAIL_EXIT1 ("readdir: fstat64 (%ld) failed: %m", fd);
+
+ struct procfs_descriptor *item = descriptor_list_emplace (&result->list);
+ if (item == NULL)
+ FAIL_EXIT1 ("descriptor_list_emplace: %m");
+ item->fd = fd;
+ item->link_target = target;
+ item->dev = st.st_dev;
+ item->ino = st.st_ino;
+ }
+
+ closedir (fds);
+
+ /* Perform a merge join between descrs and current. This assumes
+ that the arrays are sorted by file descriptor. */
+
+ qsort (descriptor_list_begin (&result->list),
+ descriptor_list_size (&result->list),
+ sizeof (struct procfs_descriptor), descriptor_compare);
+
+ return result;
+}
+
+void
+support_descriptors_free (struct support_descriptors *descrs)
+{
+ descriptor_list_free (&descrs->list);
+ free (descrs);
+}
+
+void
+support_descriptors_dump (struct support_descriptors *descrs,
+ const char *prefix, FILE *fp)
+{
+ struct procfs_descriptor *end = descriptor_list_end (&descrs->list);
+ for (struct procfs_descriptor *d = descriptor_list_begin (&descrs->list);
+ d != end; ++d)
+ {
+ char *quoted = support_quote_string (d->link_target);
+ fprintf (fp, "%s%d: target=\"%s\" major=%lld minor=%lld ino=%lld\n",
+ prefix, d->fd, quoted,
+ (long long int) major (d->dev),
+ (long long int) minor (d->dev),
+ (long long int) d->ino);
+ free (quoted);
+ }
+}
+
+static void
+dump_mismatch (bool *first,
+ struct support_descriptors *descrs,
+ struct support_descriptors *current)
+{
+ if (*first)
+ *first = false;
+ else
+ return;
+
+ puts ("error: Differences found in descriptor set");
+ puts ("Reference descriptor set:");
+ support_descriptors_dump (descrs, " ", stdout);
+ puts ("Current descriptor set:");
+ support_descriptors_dump (current, " ", stdout);
+ puts ("Differences:");
+}
+
+static void
+report_closed_descriptor (bool *first,
+ struct support_descriptors *descrs,
+ struct support_descriptors *current,
+ struct procfs_descriptor *left)
+{
+ support_record_failure ();
+ dump_mismatch (first, descrs, current);
+ printf ("error: descriptor %d was closed\n", left->fd);
+}
+
+static void
+report_opened_descriptor (bool *first,
+ struct support_descriptors *descrs,
+ struct support_descriptors *current,
+ struct procfs_descriptor *right)
+{
+ support_record_failure ();
+ dump_mismatch (first, descrs, current);
+ char *quoted = support_quote_string (right->link_target);
+ printf ("error: descriptor %d was opened (\"%s\")\n", right->fd, quoted);
+ free (quoted);
+}
+
+void
+support_descriptors_check (struct support_descriptors *descrs)
+{
+ struct support_descriptors *current = support_descriptors_list ();
+
+ /* Perform a merge join between descrs and current. This assumes
+ that the arrays are sorted by file descriptor. */
+
+ struct procfs_descriptor *left = descriptor_list_begin (&descrs->list);
+ struct procfs_descriptor *left_end = descriptor_list_end (&descrs->list);
+ struct procfs_descriptor *right = descriptor_list_begin (¤t->list);
+ struct procfs_descriptor *right_end = descriptor_list_end (¤t->list);
+
+ bool first = true;
+ while (left != left_end && right != right_end)
+ {
+ if (left->fd == right->fd)
+ {
+ if (strcmp (left->link_target, right->link_target) != 0)
+ {
+ support_record_failure ();
+ char *left_quoted = support_quote_string (left->link_target);
+ char *right_quoted = support_quote_string (right->link_target);
+ dump_mismatch (&first, descrs, current);
+ printf ("error: descriptor %d changed from \"%s\" to \"%s\"\n",
+ left->fd, left_quoted, right_quoted);
+ free (left_quoted);
+ free (right_quoted);
+ }
+ if (left->dev != right->dev)
+ {
+ support_record_failure ();
+ dump_mismatch (&first, descrs, current);
+ printf ("error: descriptor %d changed device"
+ " from %lld:%lld to %lld:%lld\n",
+ left->fd,
+ (long long int) major (left->dev),
+ (long long int) minor (left->dev),
+ (long long int) major (right->dev),
+ (long long int) minor (right->dev));
+ }
+ if (left->ino != right->ino)
+ {
+ support_record_failure ();
+ dump_mismatch (&first, descrs, current);
+ printf ("error: descriptor %d changed ino from %lld to %lld\n",
+ left->fd,
+ (long long int) left->ino, (long long int) right->ino);
+ }
+ ++left;
+ ++right;
+ }
+ else if (left->fd < right->fd)
+ {
+ /* Gap on the right. */
+ report_closed_descriptor (&first, descrs, current, left);
+ ++left;
+ }
+ else
+ {
+ /* Gap on the left. */
+ TEST_VERIFY_EXIT (left->fd > right->fd);
+ report_opened_descriptor (&first, descrs, current, right);
+ ++right;
+ }
+ }
+
+ while (left != left_end)
+ {
+ /* Closed descriptors (more descriptors on the left). */
+ report_closed_descriptor (&first, descrs, current, left);
+ ++left;
+ }
+
+ while (right != right_end)
+ {
+ /* Opened descriptors (more descriptors on the right). */
+ report_opened_descriptor (&first, descrs, current, right);
+ ++right;
+ }
+
+ support_descriptors_free (current);
+}
--- /dev/null
+/* Quote a string so that it can be used in C literals.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <string.h>
+#include <support/support.h>
+
+char *
+support_quote_string (const char *str)
+{
+ return support_quote_blob (str, strlen (str));
+}
__atomic_store_n (&state->failed, 0, __ATOMIC_RELAXED);
__atomic_add_fetch (&state->counter, 0, __ATOMIC_RELAXED);
}
+
+int
+support_record_failure_is_failed (void)
+{
+ /* Relaxed MO is sufficient because we need (blocking) external
+ synchronization for reliable test error reporting anyway. */
+ return __atomic_load_n (&state->failed, __ATOMIC_RELAXED);
+}
#include <string.h>
#include <sys/param.h>
#include <sys/resource.h>
+#include <sys/time.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <time.h>
/* The cleanup handler passed to test_main. */
static void (*cleanup_function) (void);
+static void
+print_timestamp (const char *what, struct timeval tv)
+{
+ struct tm tm;
+ if (gmtime_r (&tv.tv_sec, &tm) == NULL)
+ printf ("%s: %lld.%06d\n",
+ what, (long long int) tv.tv_sec, (int) tv.tv_usec);
+ else
+ printf ("%s: %04d-%02d-%02dT%02d:%02d:%02d.%06d\n",
+ what, 1900 + tm.tm_year, tm.tm_mon + 1, tm.tm_mday,
+ tm.tm_hour, tm.tm_min, tm.tm_sec, (int) tv.tv_usec);
+}
+
/* Timeout handler. We kill the child and exit with an error. */
static void
__attribute__ ((noreturn))
int killed;
int status;
+ /* Do this first to avoid further interference from the
+ subprocess. */
+ struct timeval now;
+ bool now_available = gettimeofday (&now, NULL) == 0;
+ struct stat64 st;
+ bool st_available = fstat64 (STDOUT_FILENO, &st) == 0 && st.st_mtime != 0;
+
assert (test_pid > 1);
/* Kill the whole process group. */
kill (-test_pid, SIGKILL);
printf ("Timed out: killed the child process but it exited %d\n",
WEXITSTATUS (status));
+ if (now_available)
+ print_timestamp ("Termination time", now);
+ if (st_available)
+ print_timestamp ("Last write to standard output",
+ (struct timeval) { st.st_mtim.tv_sec,
+ st.st_mtim.tv_nsec / 1000 });
+
/* Exit with an error. */
exit (1);
}
--- /dev/null
+/* Tests for <support/blob_repeat.h>
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <stdio.h>
+#include <support/blob_repeat.h>
+#include <support/check.h>
+
+static int
+do_test (void)
+{
+ struct support_blob_repeat repeat
+ = support_blob_repeat_allocate ("5", 1, 5);
+ TEST_COMPARE_BLOB (repeat.start, repeat.size, "55555", 5);
+ support_blob_repeat_free (&repeat);
+
+ repeat = support_blob_repeat_allocate ("ABC", 3, 3);
+ TEST_COMPARE_BLOB (repeat.start, repeat.size, "ABCABCABC", 9);
+ support_blob_repeat_free (&repeat);
+
+ repeat = support_blob_repeat_allocate ("abc", 4, 3);
+ TEST_COMPARE_BLOB (repeat.start, repeat.size, "abc\0abc\0abc", 12);
+ support_blob_repeat_free (&repeat);
+
+ size_t gigabyte = 1U << 30;
+ repeat = support_blob_repeat_allocate ("X", 1, gigabyte + 1);
+ if (repeat.start == NULL)
+ puts ("warning: not enough memory for 1 GiB mapping");
+ else
+ {
+ TEST_COMPARE (repeat.size, gigabyte + 1);
+ {
+ unsigned char *p = repeat.start;
+ for (size_t i = 0; i < gigabyte + 1; ++i)
+ if (p[i] != 'X')
+ FAIL_EXIT1 ("invalid byte 0x%02x at %zu", p[i], i);
+
+ /* Check that there is no sharing across the mapping. */
+ p[0] = 'Y';
+ p[1U << 24] = 'Z';
+ for (size_t i = 0; i < gigabyte + 1; ++i)
+ if (i == 0)
+ TEST_COMPARE (p[i], 'Y');
+ else if (i == 1U << 24)
+ TEST_COMPARE (p[i], 'Z');
+ else if (p[i] != 'X')
+ FAIL_EXIT1 ("invalid byte 0x%02x at %zu", p[i], i);
+ }
+ }
+ support_blob_repeat_free (&repeat);
+
+ repeat = support_blob_repeat_allocate ("012345678", 9, 10 * 1000 * 1000);
+ if (repeat.start == NULL)
+ puts ("warning: not enough memory for large mapping");
+ else
+ {
+ unsigned char *p = repeat.start;
+ for (int i = 0; i < 10 * 1000 * 1000; ++i)
+ for (int j = 0; j <= 8; ++j)
+ if (p[i * 9 + j] != '0' + j)
+ {
+ printf ("error: element %d index %d\n", i, j);
+ TEST_COMPARE (p[i * 9 + j], '0' + j);
+ }
+ }
+ support_blob_repeat_free (&repeat);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Tests for monitoring file descriptor usage.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <fcntl.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/capture_subprocess.h>
+#include <support/check.h>
+#include <support/descriptors.h>
+#include <support/support.h>
+#include <support/xunistd.h>
+
+/* This is the next free descriptor that the subprocess will pick. */
+static int free_descriptor;
+
+static void
+subprocess_no_change (void *closure)
+{
+ struct support_descriptors *descrs = support_descriptors_list ();
+ int fd = xopen ("/dev/null", O_WRONLY, 0);
+ TEST_COMPARE (fd, free_descriptor);
+ xclose (fd);
+ support_descriptors_free (descrs);
+}
+
+static void
+subprocess_closed_descriptor (void *closure)
+{
+ int fd = xopen ("/dev/null", O_WRONLY, 0);
+ TEST_COMPARE (fd, free_descriptor);
+ struct support_descriptors *descrs = support_descriptors_list ();
+ xclose (fd);
+ support_descriptors_check (descrs); /* Will report failure. */
+ puts ("EOT");
+ support_descriptors_free (descrs);
+}
+
+static void
+subprocess_opened_descriptor (void *closure)
+{
+ struct support_descriptors *descrs = support_descriptors_list ();
+ int fd = xopen ("/dev/null", O_WRONLY, 0);
+ TEST_COMPARE (fd, free_descriptor);
+ support_descriptors_check (descrs); /* Will report failure. */
+ puts ("EOT");
+ support_descriptors_free (descrs);
+}
+
+static void
+subprocess_changed_descriptor (void *closure)
+{
+ int fd = xopen ("/dev/null", O_WRONLY, 0);
+ TEST_COMPARE (fd, free_descriptor);
+ struct support_descriptors *descrs = support_descriptors_list ();
+ xclose (fd);
+ TEST_COMPARE (xopen ("/dev", O_DIRECTORY | O_RDONLY, 0), fd);
+ support_descriptors_check (descrs); /* Will report failure. */
+ puts ("EOT");
+ support_descriptors_free (descrs);
+}
+
+static void
+report_subprocess_output (const char *name,
+ struct support_capture_subprocess *proc)
+{
+ printf ("info: BEGIN %s output\n"
+ "%s"
+ "info: END %s output\n",
+ name, proc->out.buffer, name);
+}
+
+/* Use an explicit flag to preserve failure status across
+ support_record_failure_reset calls. */
+static bool good = true;
+
+static void
+test_run (void)
+{
+ struct support_capture_subprocess proc = support_capture_subprocess
+ (&subprocess_no_change, NULL);
+ support_capture_subprocess_check (&proc, "subprocess_no_change",
+ 0, sc_allow_none);
+ support_capture_subprocess_free (&proc);
+
+ char *expected = xasprintf ("\nDifferences:\n"
+ "error: descriptor %d was closed\n"
+ "EOT\n",
+ free_descriptor);
+ good = good && !support_record_failure_is_failed ();
+ proc = support_capture_subprocess (&subprocess_closed_descriptor, NULL);
+ good = good && support_record_failure_is_failed ();
+ support_record_failure_reset (); /* Discard the reported error. */
+ report_subprocess_output ("subprocess_closed_descriptor", &proc);
+ TEST_VERIFY (strstr (proc.out.buffer, expected) != NULL);
+ support_capture_subprocess_check (&proc, "subprocess_closed_descriptor",
+ 0, sc_allow_stdout);
+ support_capture_subprocess_free (&proc);
+ free (expected);
+
+ expected = xasprintf ("\nDifferences:\n"
+ "error: descriptor %d was opened (\"/dev/null\")\n"
+ "EOT\n",
+ free_descriptor);
+ good = good && !support_record_failure_is_failed ();
+ proc = support_capture_subprocess (&subprocess_opened_descriptor, NULL);
+ good = good && support_record_failure_is_failed ();
+ support_record_failure_reset (); /* Discard the reported error. */
+ report_subprocess_output ("subprocess_opened_descriptor", &proc);
+ TEST_VERIFY (strstr (proc.out.buffer, expected) != NULL);
+ support_capture_subprocess_check (&proc, "subprocess_opened_descriptor",
+ 0, sc_allow_stdout);
+ support_capture_subprocess_free (&proc);
+ free (expected);
+
+ expected = xasprintf ("\nDifferences:\n"
+ "error: descriptor %d changed from \"/dev/null\""
+ " to \"/dev\"\n"
+ "error: descriptor %d changed ino ",
+ free_descriptor, free_descriptor);
+ good = good && !support_record_failure_is_failed ();
+ proc = support_capture_subprocess (&subprocess_changed_descriptor, NULL);
+ good = good && support_record_failure_is_failed ();
+ support_record_failure_reset (); /* Discard the reported error. */
+ report_subprocess_output ("subprocess_changed_descriptor", &proc);
+ TEST_VERIFY (strstr (proc.out.buffer, expected) != NULL);
+ support_capture_subprocess_check (&proc, "subprocess_changed_descriptor",
+ 0, sc_allow_stdout);
+ support_capture_subprocess_free (&proc);
+ free (expected);
+}
+
+static int
+do_test (void)
+{
+ puts ("info: initial descriptor set");
+ {
+ struct support_descriptors *descrs = support_descriptors_list ();
+ support_descriptors_dump (descrs, "info: ", stdout);
+ support_descriptors_free (descrs);
+ }
+
+ free_descriptor = xopen ("/dev/null", O_WRONLY, 0);
+ puts ("info: descriptor set with additional free descriptor");
+ {
+ struct support_descriptors *descrs = support_descriptors_list ();
+ support_descriptors_dump (descrs, "info: ", stdout);
+ support_descriptors_free (descrs);
+ }
+ TEST_VERIFY (free_descriptor >= 3);
+ xclose (free_descriptor);
+
+ /* Initial test run without a sentinel descriptor. The presence of
+ such a descriptor exercises different conditions in the list
+ comparison in support_descriptors_check. */
+ test_run ();
+
+ /* Allocate a sentinel descriptor at the end of the descriptor list,
+ after free_descriptor. */
+ int sentinel_fd;
+ {
+ int fd = xopen ("/dev/full", O_WRONLY, 0);
+ TEST_COMPARE (fd, free_descriptor);
+ sentinel_fd = dup (fd);
+ TEST_VERIFY_EXIT (sentinel_fd > fd);
+ xclose (fd);
+ }
+ puts ("info: descriptor set with sentinel descriptor");
+ {
+ struct support_descriptors *descrs = support_descriptors_list ();
+ support_descriptors_dump (descrs, "info: ", stdout);
+ support_descriptors_free (descrs);
+ }
+
+ /* Second test run with sentinel descriptor. */
+ test_run ();
+
+ xclose (sentinel_fd);
+
+ return !good;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test the support_quote_string function.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <support/check.h>
+#include <support/support.h>
+#include <string.h>
+#include <stdlib.h>
+
+static int
+do_test (void)
+{
+ char *p = support_quote_string ("");
+ TEST_COMPARE (strlen (p), 0);
+ free (p);
+ p = support_quote_string ("X");
+ TEST_COMPARE (strlen (p), 1);
+ TEST_COMPARE (p[0], 'X');
+ free (p);
+
+ /* Check escaping of backslash-escaped characters, and lack of
+ escaping for other shell meta-characters. */
+ p = support_quote_string ("$()*?`@[]{}~\'\"X");
+ TEST_COMPARE (strcmp (p, "$()*?`@[]{}~\\'\\\"X"), 0);
+ free (p);
+
+ /* Check lack of escaping for letters and digits. */
+#define LETTERS_AND_DIGTS \
+ "abcdefghijklmnopqrstuvwxyz" \
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZ" \
+ "0123456789"
+ p = support_quote_string (LETTERS_AND_DIGTS "@");
+ TEST_COMPARE (strcmp (p, LETTERS_AND_DIGTS "@"), 0);
+ free (p);
+
+ /* Check escaping of control characters and other non-printable
+ characters. */
+ p = support_quote_string ("\r\n\t\a\b\f\v\1\177\200\377@");
+ TEST_COMPARE (strcmp (p, "\\r\\n\\t\\a\\b\\f\\v\\001"
+ "\\177\\200\\377@"), 0);
+ free (p);
+
+ return 0;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* pthread_rwlock_destroy with error checking.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <support/xthread.h>
+
+void
+xpthread_rwlock_destroy (pthread_rwlock_t *rwlock)
+{
+ xpthread_check_return ("pthread_rwlock_destroy",
+ pthread_rwlock_destroy (rwlock));
+}
void xpthread_rwlock_wrlock (pthread_rwlock_t *rwlock);
void xpthread_rwlock_rdlock (pthread_rwlock_t *rwlock);
void xpthread_rwlock_unlock (pthread_rwlock_t *rwlock);
+void xpthread_rwlock_destroy (pthread_rwlock_t *rwlock);
__END_DECLS
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif
/* Check for unexpected PLT reloc type. */
if (__builtin_expect (r_type == AARCH64_R(JUMP_SLOT), 1))
{
- if (__builtin_expect (map->l_mach.plt, 0) == 0)
- *reloc_addr += l_addr;
- else
- *reloc_addr = map->l_mach.plt;
+ if (map->l_mach.plt == 0)
+ {
+ /* Prelinking. */
+ *reloc_addr += l_addr;
+ return;
+ }
+
+ if (1) /* DT_AARCH64_VARIANT_PCS is not available, so always check. */
+ {
+ /* Check the symbol table for variant PCS symbols. */
+ const Elf_Symndx symndx = ELFW (R_SYM) (reloc->r_info);
+ const ElfW (Sym) *symtab =
+ (const void *)D_PTR (map, l_info[DT_SYMTAB]);
+ const ElfW (Sym) *sym = &symtab[symndx];
+ if (__glibc_unlikely (sym->st_other & STO_AARCH64_VARIANT_PCS))
+ {
+ /* Avoid lazy resolution of variant PCS symbols. */
+ const struct r_found_version *version = NULL;
+ if (map->l_info[VERSYMIDX (DT_VERSYM)] != NULL)
+ {
+ const ElfW (Half) *vernum =
+ (const void *)D_PTR (map, l_info[VERSYMIDX (DT_VERSYM)]);
+ version = &map->l_versions[vernum[symndx] & 0x7fff];
+ }
+ elf_machine_rela (map, reloc, sym, version, reloc_addr,
+ skip_ifunc);
+ return;
+ }
+ }
+
+ *reloc_addr = map->l_mach.plt;
}
else if (__builtin_expect (r_type == AARCH64_R(TLSDESC), 1))
{
libc_ifunc (__libc_memcpy,
(IS_THUNDERX (midr)
? __memcpy_thunderx
- : (IS_FALKOR (midr) || IS_PHECDA (midr)
+ : (IS_FALKOR (midr) || IS_PHECDA (midr) || IS_ARES (midr)
? __memcpy_falkor
: (IS_THUNDERX2 (midr) || IS_THUNDERX2PA (midr)
? __memcpy_thunderx2
ldouble: 2
Function: "cos":
+double: 1
float: 1
+idouble: 1
ifloat: 1
ildouble: 1
ldouble: 1
ldouble: 1
Function: "pow":
+double: 1
float: 1
+idouble: 1
ifloat: 1
ildouble: 2
ldouble: 2
ldouble: 2
Function: "sin":
+double: 1
float: 1
+idouble: 1
ifloat: 1
ildouble: 1
ldouble: 1
ldouble: 3
Function: "sincos":
+double: 1
float: 1
+idouble: 1
ifloat: 1
ildouble: 1
ldouble: 1
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
--- /dev/null
+/* As default architectures with sizeof (off_t) < sizeof (off64_t) the mmap is
+ implemented with __SYS_mmap2 syscall and the offset is represented in
+ multiples of page size. For offset larger than
+ '1 << (page_shift + 8 * sizeof (off_t))' (that is, 1<<44 on system with
+ page size of 4096 bytes) the system call silently truncates the offset.
+ For this case, glibc mmap implementation returns EINVAL. */
+
+/* Return the maximum value expected as offset argument in mmap64 call. */
+static inline uint64_t
+mmap64_maximum_offset (long int page_shift)
+{
+ if (sizeof (off_t) < sizeof (off64_t))
+ return (UINT64_C(1) << (page_shift + (8 * sizeof (off_t)))) - 1;
+ else
+ return UINT64_MAX;
+}
struct frame_state_reg_info *old_rs = fs->regs.prev;
#ifdef _LIBC
if (old_rs == NULL)
- __libc_fatal ("invalid DWARF unwind data");
+ __libc_fatal ("Invalid DWARF unwind data.\n");
else
#endif
{
return self->thread_specifics[key];
}
strong_alias (__pthread_getspecific, pthread_getspecific);
+hidden_def (__pthread_getspecific)
return 0;
}
strong_alias (__pthread_setspecific, pthread_setspecific);
+hidden_def (__pthread_setspecific)
#if IS_IN (libpthread)
hidden_proto (__pthread_key_create)
+hidden_proto (__pthread_getspecific)
+hidden_proto (__pthread_setspecific)
hidden_proto (_pthread_mutex_init)
#endif
+++ /dev/null
-/* Linux/i386 CET initializers function.
- Copyright (C) 2018 Free Software Foundation, Inc.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <http://www.gnu.org/licenses/>. */
-
-
-#define LINKAGE static inline
-#define _dl_cet_check cet_check
-#include <sysdeps/x86/dl-cet.c>
-#undef _dl_cet_check
-
-#ifdef SHARED
-void
-_dl_cet_check (struct link_map *main_map, const char *program)
-{
- cet_check (main_map, program);
-
- if ((GL(dl_x86_feature_1)[0] & GNU_PROPERTY_X86_FEATURE_1_SHSTK))
- {
- /* Replace _dl_runtime_resolve and _dl_runtime_profile with
- _dl_runtime_resolve_shstk and _dl_runtime_profile_shstk,
- respectively if SHSTK is enabled. */
- extern void _dl_runtime_resolve (Elf32_Word) attribute_hidden;
- extern void _dl_runtime_resolve_shstk (Elf32_Word) attribute_hidden;
- extern void _dl_runtime_profile (Elf32_Word) attribute_hidden;
- extern void _dl_runtime_profile_shstk (Elf32_Word) attribute_hidden;
- unsigned int i;
- struct link_map *l;
- Elf32_Addr *got;
-
- if (main_map->l_info[DT_JMPREL])
- {
- got = (Elf32_Addr *) D_PTR (main_map, l_info[DT_PLTGOT]);
- if (got[2] == (Elf32_Addr) &_dl_runtime_resolve)
- got[2] = (Elf32_Addr) &_dl_runtime_resolve_shstk;
- else if (got[2] == (Elf32_Addr) &_dl_runtime_profile)
- got[2] = (Elf32_Addr) &_dl_runtime_profile_shstk;
- }
-
- i = main_map->l_searchlist.r_nlist;
- while (i-- > 0)
- {
- l = main_map->l_initfini[i];
- if (l->l_info[DT_JMPREL])
- {
- got = (Elf32_Addr *) D_PTR (l, l_info[DT_PLTGOT]);
- if (got[2] == (Elf32_Addr) &_dl_runtime_resolve)
- got[2] = (Elf32_Addr) &_dl_runtime_resolve_shstk;
- else if (got[2] == (Elf32_Addr) &_dl_runtime_profile)
- got[2] = (Elf32_Addr) &_dl_runtime_profile_shstk;
- }
- }
- }
-}
-#endif
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
Elf32_Addr *got;
extern void _dl_runtime_resolve (Elf32_Word) attribute_hidden;
extern void _dl_runtime_profile (Elf32_Word) attribute_hidden;
+ extern void _dl_runtime_resolve_shstk (Elf32_Word) attribute_hidden;
+ extern void _dl_runtime_profile_shstk (Elf32_Word) attribute_hidden;
+ /* Check if SHSTK is enabled by kernel. */
+ bool shstk_enabled
+ = (GL(dl_x86_feature_1)[0] & GNU_PROPERTY_X86_FEATURE_1_SHSTK) != 0;
if (l->l_info[DT_JMPREL] && lazy)
{
end in this function. */
if (__glibc_unlikely (profile))
{
- got[2] = (Elf32_Addr) &_dl_runtime_profile;
+ got[2] = (shstk_enabled
+ ? (Elf32_Addr) &_dl_runtime_profile_shstk
+ : (Elf32_Addr) &_dl_runtime_profile);
if (GLRO(dl_profile) != NULL
&& _dl_name_match_p (GLRO(dl_profile), l))
else
/* This function will get called to fix up the GOT entry indicated by
the offset on the stack, and then jump to the resolved address. */
- got[2] = (Elf32_Addr) &_dl_runtime_resolve;
+ got[2] = (shstk_enabled
+ ? (Elf32_Addr) &_dl_runtime_resolve_shstk
+ : (Elf32_Addr) &_dl_runtime_resolve);
}
return lazy;
NULL
*/
- .text
- .globl _start
- .type _start,@function
-_start:
+#include <sysdep.h>
+
+ENTRY (_start)
+ /* Clearing frame pointer is insufficient, use CFI. */
+ cfi_undefined (eip)
/* Clear the frame pointer. The ABI suggests this be done, to mark
the outermost frame obviously. */
xorl %ebp, %ebp
1: movl (%esp), %ebx
ret
#endif
+END (_start)
/* To fulfill the System V/i386 ABI we need this symbol. Yuck, it's so
meaningless since we don't support machines < 80386. */
.section .text
-GLOBAL_LIBM_ENTRY(__exp2f)
+WEAK_LIBM_ENTRY(exp2f)
{.mfi
}
;;
-GLOBAL_LIBM_END(__exp2f)
+WEAK_LIBM_END(exp2f)
libm_alias_float_other (__exp2, exp2)
#ifdef SHARED
-.symver __exp2f,exp2f@@GLIBC_2.27
+.symver exp2f,exp2f@@GLIBC_2.27
.weak __exp2f_compat
.set __exp2f_compat,__exp2f
.symver __exp2f_compat,exp2f@GLIBC_2.2
.section .text
-GLOBAL_LIBM_ENTRY(__log2f)
+WEAK_LIBM_ENTRY(log2f)
{ .mfi
alloc r32=ar.pfs,1,4,4,0
br.ret.sptk b0;;
}
-GLOBAL_LIBM_END(__log2f)
+WEAK_LIBM_END(log2f)
libm_alias_float_other (__log2, log2)
#ifdef SHARED
-.symver __log2f,log2f@@GLIBC_2.27
+.symver log2f,log2f@@GLIBC_2.27
.weak __log2f_compat
.set __log2f_compat,__log2f
.symver __log2f_compat,log2f@GLIBC_2.2
LOCAL_OBJECT_END(pow_tbl2)
.section .text
-GLOBAL_LIBM_ENTRY(__powf)
+WEAK_LIBM_ENTRY(powf)
// Get exponent of x. Will be used to calculate K.
{ .mfi
}
;;
-GLOBAL_LIBM_END(__powf)
+WEAK_LIBM_END(powf)
libm_alias_float_other (__pow, pow)
#ifdef SHARED
-.symver __powf,powf@@GLIBC_2.27
+.symver powf,powf@@GLIBC_2.27
.weak __powf_compat
.set __powf_compat,__powf
.symver __powf_compat,powf@GLIBC_2.2
#undef fdivl
#include <math-narrow.h>
+#include <libc-diag.h>
+
+/* R_f[01] are not set in cases where they are not used in packing,
+ but the compiler does not see that they are set in all cases where
+ they are used, resulting in warnings that they may be used
+ uninitialized. The location of the warning differs in different
+ versions of GCC, it may be where R is defined using a macro or it
+ may be where the macro is defined. This happens only with -O1. */
+DIAG_PUSH_NEEDS_COMMENT;
+DIAG_IGNORE_NEEDS_COMMENT (8, "-Wmaybe-uninitialized");
#include <soft-fp.h>
#include <single.h>
#include <double.h>
CHECK_NARROW_DIV (ret, x, y);
return ret;
}
+DIAG_POP_NEEDS_COMMENT;
+
libm_alias_float_double (div)
unsigned int __nusers;
#endif
/* KIND must stay at this position in the structure to maintain
- binary compatibility with static initializers. */
+ binary compatibility with static initializers.
+
+ Concurrency notes:
+ The __kind of a mutex is initialized either by the static
+ PTHREAD_MUTEX_INITIALIZER or by a call to pthread_mutex_init.
+
+ After a mutex has been initialized, the __kind of a mutex is usually not
+ changed. BUT it can be set to -1 in pthread_mutex_destroy or elision can
+ be enabled. This is done concurrently in the pthread_mutex_*lock functions
+ by using the macro FORCE_ELISION. This macro is only defined for
+ architectures which supports lock elision.
+
+ For elision, there are the flags PTHREAD_MUTEX_ELISION_NP and
+ PTHREAD_MUTEX_NO_ELISION_NP which can be set in addition to the already set
+ type of a mutex.
+ Before a mutex is initialized, only PTHREAD_MUTEX_NO_ELISION_NP can be set
+ with pthread_mutexattr_settype.
+ After a mutex has been initialized, the functions pthread_mutex_*lock can
+ enable elision - if the mutex-type and the machine supports it - by setting
+ the flag PTHREAD_MUTEX_ELISION_NP. This is done concurrently. Afterwards
+ the lock / unlock functions are using specific elision code-paths. */
int __kind;
__PTHREAD_COMPAT_PADDING_MID
#if __PTHREAD_MUTEX_NUSERS_AFTER_KIND
but our current fork implementation is not. */
bool multiple_threads = THREAD_GETMEM (THREAD_SELF, header.multiple_threads);
- __run_fork_handlers (atfork_run_prepare);
+ __run_fork_handlers (atfork_run_prepare, multiple_threads);
/* If we are not running multiple threads, we do not have to
preserve lock state. If fork runs from a signal handler, only
__rtld_lock_initialize (GL(dl_load_lock));
/* Run the handlers registered for the child. */
- __run_fork_handlers (atfork_run_child);
+ __run_fork_handlers (atfork_run_child, multiple_threads);
}
else
{
}
/* Run the handlers registered for the parent. */
- __run_fork_handlers (atfork_run_parent);
+ __run_fork_handlers (atfork_run_parent, multiple_threads);
}
return pid;
- atfork_run_child: run all the CHILD_HANDLER and unlocks the internal
lock.
- atfork_run_parent: run all the PARENT_HANDLER and unlocks the internal
- lock. */
-extern void __run_fork_handlers (enum __run_fork_handler_type who)
- attribute_hidden;
+ lock.
+
+ Perform locking only if DO_LOCKING. */
+extern void __run_fork_handlers (enum __run_fork_handler_type who,
+ _Bool do_locking) attribute_hidden;
/* C library side function to register new fork handlers. */
extern int __register_atfork (void (*__prepare) (void),
static __always_inline __attribute__ ((__noreturn__)) void
futex_fatal_error (void)
{
- __libc_fatal ("The futex facility returned an unexpected error code.");
+ __libc_fatal ("The futex facility returned an unexpected error code.\n");
}
#endif /* futex-internal.h */
malloc_name = true;
}
- if (__inet_aton (name, (struct in_addr *) at->addr) != 0)
+ if (__inet_aton_exact (name, (struct in_addr *) at->addr) != 0)
{
if (req->ai_family == AF_UNSPEC || req->ai_family == AF_INET)
at->family = AF_INET;
#ifdef __powerpc64__
TCB_AT_PLATFORM (offsetof (tcbhead_t, at_platform) - TLS_TCB_OFFSET - sizeof(tcbhead_t))
#endif
-TM_CAPABLE (offsetof (tcbhead_t, tm_capable) - TLS_TCB_OFFSET - sizeof (tcbhead_t))
#ifndef __powerpc64__
TCB_AT_PLATFORM (offsetof (tcbhead_t, at_platform) - TLS_TCB_OFFSET - sizeof(tcbhead_t))
PADDING (offsetof (tcbhead_t, padding) - TLS_TCB_OFFSET - sizeof(tcbhead_t))
uint32_t padding;
uint32_t at_platform;
#endif
- /* Indicate if HTM capable (ISA 2.07). */
- uint32_t tm_capable;
+ uint32_t __unused;
/* Reservation for AT_PLATFORM data - powerpc64. */
#ifdef __powerpc64__
uint32_t at_platform;
# define TLS_INIT_TP(tcbp) \
({ \
__thread_register = (void *) (tcbp) + TLS_TCB_OFFSET; \
- THREAD_SET_TM_CAPABLE (__tcb_hwcap & PPC_FEATURE2_HAS_HTM ? 1 : 0); \
THREAD_SET_HWCAP (__tcb_hwcap); \
THREAD_SET_AT_PLATFORM (__tcb_platform); \
NULL; \
/* Value passed to 'clone' for initialization of the thread register. */
# define TLS_DEFINE_INIT_TP(tp, pd) \
void *tp = (void *) (pd) + TLS_TCB_OFFSET + TLS_PRE_TCB_SIZE; \
- (((tcbhead_t *) ((char *) tp - TLS_TCB_OFFSET))[-1].tm_capable) = \
- THREAD_GET_TM_CAPABLE (); \
(((tcbhead_t *) ((char *) tp - TLS_TCB_OFFSET))[-1].hwcap) = \
THREAD_GET_HWCAP (); \
(((tcbhead_t *) ((char *) tp - TLS_TCB_OFFSET))[-1].at_platform) = \
+ TLS_PRE_TCB_SIZE))[-1].pointer_guard \
= THREAD_GET_POINTER_GUARD())
-/* tm_capable field in TCB head. */
-# define THREAD_GET_TM_CAPABLE() \
- (((tcbhead_t *) ((char *) __thread_register \
- - TLS_TCB_OFFSET))[-1].tm_capable)
-# define THREAD_SET_TM_CAPABLE(value) \
- (THREAD_GET_TM_CAPABLE () = (value))
-
/* hwcap field in TCB head. */
# define THREAD_GET_HWCAP() \
(((tcbhead_t *) ((char *) __thread_register \
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
cfi_endproc; \
ASM_SIZE_DIRECTIVE(name)
-#if !IS_IN(rtld) && !defined(__SPE__)
-# define ABORT_TRANSACTION_IMPL \
- cmpwi 2,0; \
- beq 1f; \
- lwz 0,TM_CAPABLE(2); \
- cmpwi 0,0; \
- beq 1f; \
- li 11,_ABORT_SYSCALL; \
- tabort. 11; \
- .align 4; \
-1:
-#else
-# define ABORT_TRANSACTION_IMPL
-#endif
-#define ABORT_TRANSACTION ABORT_TRANSACTION_IMPL
-
#define DO_CALL(syscall) \
- ABORT_TRANSACTION \
li 0,syscall; \
sc
#define N r5
#define VL r6
+#define R27SAVE (-40)
+#define R28SAVE (-32)
+#define R29SAVE (-24)
+#define R30SAVE (-16)
+#define R31SAVE (-8)
+
ENTRY_TOCLESS (FUNC, 5)
- std r31, -8(r1)
+ std r31, R31SAVE(r1)
rldicl. r0, N, 0, 62
- std r30, -16(r1)
+ std r30, R30SAVE(r1)
cmpdi VL, r0, 2
- std r29, -24(r1)
+ std r29, R29SAVE(r1)
addi N, N, 3
- std r28, -32(r1)
+ std r28, R28SAVE(r1)
srdi N, N, 2
- std r27, -40(r1)
+ std r27, R27SAVE(r1)
+ cfi_offset(r31, R31SAVE)
+ cfi_offset(r30, R30SAVE)
+ cfi_offset(r29, R29SAVE)
+ cfi_offset(r28, R28SAVE)
+ cfi_offset(r27, R27SAVE)
mtctr N
beq cr0, L(b00)
blt cr6, L(b01)
addic r11, r11, 1
#endif
addze RP, r8
- ld r31, -8(r1)
- ld r30, -16(r1)
- ld r29, -24(r1)
- ld r28, -32(r1)
- ld r27, -40(r1)
+ ld r31, R31SAVE(r1)
+ ld r30, R30SAVE(r1)
+ ld r29, R29SAVE(r1)
+ ld r28, R28SAVE(r1)
+ ld r27, R27SAVE(r1)
blr
END(FUNC)
#endif
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
#define TNC r0
#define U0 r30
#define U1 r31
+#define U0SAVE (-16)
+#define U1SAVE (-8)
#define RETVAL r5
ENTRY_TOCLESS (__mpn_lshift, 5)
- std U1, -8(r1)
- std U0, -16(r1)
+ std U1, U1SAVE(r1)
+ std U0, U0SAVE(r1)
+ cfi_offset(U1, U1SAVE)
+ cfi_offset(U0, U0SAVE)
subfic TNC, CNT, 64
sldi r7, N, RP
add UP, UP, r7
L(cj2): std r10, -32(RP)
std r8, -40(RP)
-L(ret): ld U1, -8(r1)
- ld U0, -16(r1)
+L(ret): ld U1, U1SAVE(r1)
+ ld U0, U0SAVE(r1)
mr RP, RETVAL
blr
END(__mpn_lshift)
#define N r5
#define VL r6
+#define R26SAVE (-48)
+#define R27SAVE (-40)
+
ENTRY_TOCLESS (__mpn_mul_1, 5)
- std r27, -40(r1)
- std r26, -48(r1)
+ std r27, R27SAVE(r1)
+ std r26, R26SAVE(r1)
+ cfi_offset(r27, R27SAVE)
+ cfi_offset(r26, R26SAVE)
li r12, 0
ld r26, 0(UP)
std r0, 0(RP)
std r7, 8(RP)
L(ret): addze RP, r8
- ld r27, -40(r1)
- ld r26, -48(r1)
+ ld r27, R27SAVE(r1)
+ ld r26, R26SAVE(r1)
blr
END(__mpn_mul_1)
TRACEBACK_MASK(name,mask); \
END_2(name)
-#if !IS_IN(rtld)
-# define ABORT_TRANSACTION_IMPL \
- cmpdi 13,0; \
- beq 1f; \
- lwz 0,TM_CAPABLE(13); \
- cmpwi 0,0; \
- beq 1f; \
- li 11,_ABORT_SYSCALL; \
- tabort. 11; \
- .p2align 4; \
-1:
-#else
-# define ABORT_TRANSACTION_IMPL
-#endif
-#define ABORT_TRANSACTION ABORT_TRANSACTION_IMPL
-
#define DO_CALL(syscall) \
- ABORT_TRANSACTION \
li 0,syscall; \
sc
*/
#define _SYSDEPS_SYSDEP_H 1
#include <bits/hwcap.h>
-#include <tls.h>
-#include <htm.h>
#define PPC_FEATURE_970 (PPC_FEATURE_POWER4 + PPC_FEATURE_HAS_ALTIVEC)
#define ALIGNARG(log2) log2
#define ASM_SIZE_DIRECTIVE(name) .size name,.-name
-#else
-
-/* Linux kernel powerpc documentation [1] states issuing a syscall inside a
- transaction is not recommended and may lead to undefined behavior. It
- also states syscalls do not abort transactions. To avoid such traps,
- we abort transaction just before syscalls.
-
- [1] Documentation/powerpc/transactional_memory.txt [Syscalls] */
-#if !IS_IN(rtld) && !defined(__SPE__)
-# define ABORT_TRANSACTION \
- ({ \
- if (THREAD_GET_TM_CAPABLE ()) \
- __libc_tabort (_ABORT_SYSCALL); \
- })
-#else
-# define ABORT_TRANSACTION
-#endif
-
#endif /* __ASSEMBLER__ */
ldouble: 2
Function: "cos":
+double: 1
+idouble: 1
ildouble: 1
ldouble: 1
ldouble: 4
Function: Imaginary part of "ctan_towardzero":
-double: 1
+double: 2
float: 2
-idouble: 1
+idouble: 2
ifloat: 2
ildouble: 5
ldouble: 5
Function: "log_upward":
double: 1
idouble: 1
-ildouble: 1
-ldouble: 1
+ildouble: 2
+ldouble: 2
Function: "pow":
+double: 1
+idouble: 1
ildouble: 2
ldouble: 2
ldouble: 2
Function: "sin":
+double: 1
+idouble: 1
ildouble: 1
ldouble: 1
ldouble: 3
Function: "sincos":
+double: 1
+idouble: 1
ildouble: 1
ldouble: 1
static __always_inline void
libc_feholdexcept_setround_riscv (fenv_t *envp, int round)
{
- libc_fesetround_riscv (round);
libc_feholdexcept_riscv (envp);
+ libc_fesetround_riscv (round);
}
#define libc_feholdexcept_setround libc_feholdexcept_setround_riscv
__libc_start_main wants this in a5. */
ENTRY (ENTRY_POINT)
+ /* Terminate call stack by noting ra is undefined. Use a dummy
+ .cfi_label to force starting the FDE. */
+ .cfi_label .Ldummy
+ cfi_undefined (ra)
call .Lload_gp
mv a5, a0 /* rtld_fini. */
/* main may be in a shared library. */
lla a4, __libc_csu_fini
mv a6, sp /* stack_end. */
- tail __libc_start_main@plt
+ call __libc_start_main@plt
+ ebreak
END (ENTRY_POINT)
/* Dynamic links need the global pointer to be initialized prior to calling
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
};
#define HWCAP_IMPORTANT (HWCAP_S390_ZARCH | HWCAP_S390_LDISP \
- | HWCAP_S390_EIMM | HWCAP_S390_DFP)
+ | HWCAP_S390_EIMM | HWCAP_S390_DFP \
+ | HWCAP_S390_VX | HWCAP_S390_VXE)
/* We cannot provide a general printing function. */
#define _dl_procinfo(type, word) -1
else if (r_type == R_SPARC_NONE)
;
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
else if (r_type == R_SPARC_NONE)
;
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
ifeq ($(subdir),dirent)
sysdep_routines += getdirentries getdirentries64
+tests-internal += tst-readdir64-compat
endif
ifeq ($(subdir),nis)
{"thunderx2t99", 0x431F0AF0},
{"thunderx2t99p1", 0x420F5160},
{"phecda", 0x680F0000},
+ {"ares", 0x411FD0C0},
{"generic", 0x0}
};
#define IS_PHECDA(midr) (MIDR_IMPLEMENTOR(midr) == 'h' \
&& MIDR_PARTNUM(midr) == 0x000)
+#define IS_ARES(midr) (MIDR_IMPLEMENTOR(midr) == 'A' \
+ && MIDR_PARTNUM(midr) == 0xd0c)
struct cpu_features
{
/* Support for copy_file_range, statx was added in kernel 4.13. */
#if __LINUX_KERNEL_VERSION < 0x040D00
# undef __ASSUME_MLOCK2
-# undef __ASSUME_COPY_FILE_RANGE
# undef __ASSUME_STATX
#endif
declarations of A_OLDVAL et al because when NEWVAL or OLDVAL is of the
form *PTR and PTR has a 'volatile ... *' type, then __typeof (*PTR) has
a 'volatile ...' type and this triggers -Wvolatile-register-var to
- complain about 'register volatile ... asm ("reg")'. */
+ complain about 'register volatile ... asm ("reg")'.
+
+ We use the same union trick in the declaration of A_PTR because when
+ MEM is of the from *PTR and PTR has a 'const ... *' type, then __typeof
+ (*PTR) has a 'const ...' type and this enables the compiler to substitute
+ the variable with its initializer in asm statements, which may cause the
+ corresponding operand to appear in a different register. */
#ifdef __thumb2__
/* Thumb-2 has ldrex/strex. However it does not have barrier instructions,
so we still need to use the kernel helper. */
# define __arm_assisted_compare_and_exchange_val_32_acq(mem, newval, oldval) \
- ({ union { __typeof (oldval) a; uint32_t v; } oldval_arg = { .a = (oldval) };\
+ ({ union { __typeof (mem) a; uint32_t v; } mem_arg = { .a = (mem) }; \
+ union { __typeof (oldval) a; uint32_t v; } oldval_arg = { .a = (oldval) };\
union { __typeof (newval) a; uint32_t v; } newval_arg = { .a = (newval) };\
register uint32_t a_oldval asm ("r0"); \
register uint32_t a_newval asm ("r1") = newval_arg.v; \
- register __typeof (mem) a_ptr asm ("r2") = (mem); \
+ register uint32_t a_ptr asm ("r2") = mem_arg.v; \
register uint32_t a_tmp asm ("r3"); \
register uint32_t a_oldval2 asm ("r4") = oldval_arg.v; \
__asm__ __volatile__ \
(__typeof (oldval)) a_tmp; })
#else
# define __arm_assisted_compare_and_exchange_val_32_acq(mem, newval, oldval) \
- ({ union { __typeof (oldval) a; uint32_t v; } oldval_arg = { .a = (oldval) };\
+ ({ union { __typeof (mem) a; uint32_t v; } mem_arg = { .a = (mem) }; \
+ union { __typeof (oldval) a; uint32_t v; } oldval_arg = { .a = (oldval) };\
union { __typeof (newval) a; uint32_t v; } newval_arg = { .a = (newval) };\
register uint32_t a_oldval asm ("r0"); \
register uint32_t a_newval asm ("r1") = newval_arg.v; \
- register __typeof (mem) a_ptr asm ("r2") = (mem); \
+ register uint32_t a_ptr asm ("r2") = mem_arg.v; \
register uint32_t a_tmp asm ("r3"); \
register uint32_t a_oldval2 asm ("r4") = oldval_arg.v; \
__asm__ __volatile__ \
/* SIGPOLL. */
struct
{
- long int si_band; /* Band event for SIGPOLL. */
+ __SI_BAND_TYPE si_band; /* Band event for SIGPOLL. */
int si_fd;
} _sigpoll;
#include <sysdep-cancel.h>
#include <unistd.h>
-/* Include the fallback implementation. */
-#ifndef __ASSUME_COPY_FILE_RANGE
-#define COPY_FILE_RANGE_DECL static
-#define COPY_FILE_RANGE copy_file_range_compat
-#include <io/copy_file_range-compat.c>
-#endif
-
ssize_t
copy_file_range (int infd, __off64_t *pinoff,
int outfd, __off64_t *poutoff,
size_t length, unsigned int flags)
{
#ifdef __NR_copy_file_range
- ssize_t ret = SYSCALL_CANCEL (copy_file_range, infd, pinoff, outfd, poutoff,
- length, flags);
-# ifndef __ASSUME_COPY_FILE_RANGE
- if (ret == -1 && errno == ENOSYS)
- ret = copy_file_range_compat (infd, pinoff, outfd, poutoff, length, flags);
-# endif
- return ret;
-#else /* !__NR_copy_file_range */
- return copy_file_range_compat (infd, pinoff, outfd, poutoff, length, flags);
+ return SYSCALL_CANCEL (copy_file_range, infd, pinoff, outfd, poutoff,
+ length, flags);
+#else
+ __set_errno (ENOSYS);
+ return -1;
#endif
}
# include <shlib-compat.h>
# if SHLIB_COMPAT(libc, GLIBC_2_1, GLIBC_2_2)
-# include <olddirent.h>
+# include <olddirent.h>
+# include <unistd.h>
-/* kernel definition of as of 3.2. */
-struct compat_linux_dirent
+static ssize_t
+handle_overflow (int fd, __off64_t offset, ssize_t count)
{
- /* Both d_ino and d_off are compat_ulong_t which are defined in all
- architectures as 'u32'. */
- uint32_t d_ino;
- uint32_t d_off;
- unsigned short d_reclen;
- char d_name[1];
-};
+ /* If this is the first entry in the buffer, we can report the
+ error. */
+ if (offset == 0)
+ {
+ __set_errno (EOVERFLOW);
+ return -1;
+ }
+
+ /* Otherwise, seek to the overflowing entry, so that the next call
+ will report the error, and return the data read so far. */
+ if (__lseek64 (fd, offset, SEEK_SET) != 0)
+ return -1;
+ return count;
+}
ssize_t
__old_getdents64 (int fd, char *buf, size_t nbytes)
{
- ssize_t retval = INLINE_SYSCALL_CALL (getdents, fd, buf, nbytes);
+ /* We do not move the individual directory entries. This is only
+ possible if the target type (struct __old_dirent64) is smaller
+ than the source type. */
+ _Static_assert (offsetof (struct __old_dirent64, d_name)
+ <= offsetof (struct dirent64, d_name),
+ "__old_dirent64 is larger than dirent64");
+ _Static_assert (__alignof__ (struct __old_dirent64)
+ <= __alignof__ (struct dirent64),
+ "alignment of __old_dirent64 is larger than dirent64");
- /* The kernel added the d_type value after the name. Change this now. */
- if (retval != -1)
+ ssize_t retval = INLINE_SYSCALL_CALL (getdents64, fd, buf, nbytes);
+ if (retval > 0)
{
- union
- {
- struct compat_linux_dirent k;
- struct dirent u;
- } *kbuf = (void *) buf;
+ /* This is the marker for the first entry. Offset 0 is reserved
+ for the first entry (see rewinddir). Here, we use it as a
+ marker for the first entry in the buffer. We never actually
+ seek to offset 0 because handle_overflow reports the error
+ directly, so it does not matter that the offset is incorrect
+ if entries have been read from the descriptor before (so that
+ the descriptor is not actually at offset 0). */
+ __off64_t previous_offset = 0;
- while ((char *) kbuf < buf + retval)
+ char *p = buf;
+ char *end = buf + retval;
+ while (p < end)
{
- char d_type = *((char *) kbuf + kbuf->k.d_reclen - 1);
- memmove (kbuf->u.d_name, kbuf->k.d_name,
- strlen (kbuf->k.d_name) + 1);
- kbuf->u.d_type = d_type;
+ struct dirent64 *source = (struct dirent64 *) p;
+
+ /* Copy out the fixed-size data. */
+ __ino_t ino = source->d_ino;
+ __off64_t offset = source->d_off;
+ unsigned int reclen = source->d_reclen;
+ unsigned char type = source->d_type;
+
+ /* Check for ino_t overflow. */
+ if (__glibc_unlikely (ino != source->d_ino))
+ return handle_overflow (fd, previous_offset, p - buf);
+
+ /* Convert to the target layout. Use a separate struct and
+ memcpy to side-step aliasing issues. */
+ struct __old_dirent64 result;
+ result.d_ino = ino;
+ result.d_off = offset;
+ result.d_reclen = reclen;
+ result.d_type = type;
+
+ /* Write the fixed-sized part of the result to the
+ buffer. */
+ size_t result_name_offset = offsetof (struct __old_dirent64, d_name);
+ memcpy (p, &result, result_name_offset);
+
+ /* Adjust the position of the name if necessary. Copy
+ everything until the end of the record, including the
+ terminating NUL byte. */
+ if (result_name_offset != offsetof (struct dirent64, d_name))
+ memmove (p + result_name_offset, source->d_name,
+ reclen - offsetof (struct dirent64, d_name));
- kbuf = (void *) ((char *) kbuf + kbuf->k.d_reclen);
+ p += reclen;
+ previous_offset = offset;
}
}
return retval;
{
int ret = __gethostbyname_r (hostname, &hostbuf,
tmpbuf.data, tmpbuf.length, &hp, &herr);
- if (ret == 0)
+ if (ret == 0 && hp != NULL)
break;
else
{
/* Enlarge the buffer on ERANGE. */
- if (herr == NETDB_INTERNAL && errno == ERANGE)
+ if (ret != 0 && herr == NETDB_INTERNAL && errno == ERANGE)
{
if (!scratch_buffer_grow (&tmpbuf))
return 0;
return 0;
#else
struct ifreq ifr;
- int fd = __opensock ();
-
- if (fd < 0)
- return 0;
-
if (strlen (ifname) >= IFNAMSIZ)
{
__set_errno (ENODEV);
}
strncpy (ifr.ifr_name, ifname, sizeof (ifr.ifr_name));
+
+ int fd = __opensock ();
+
+ if (fd < 0)
+ return 0;
+
if (__ioctl (fd, SIOCGIFINDEX, &ifr) < 0)
{
int saved_errno = errno;
# define __ASSUME_MLOCK2 1
#endif
-#if __LINUX_KERNEL_VERSION >= 0x040500
-# define __ASSUME_COPY_FILE_RANGE 1
-#endif
-
/* Support for statx was added in kernel 4.11. */
#if __LINUX_KERNEL_VERSION >= 0x040B00
# define __ASSUME_STATX 1
#ifndef _KERNEL_SIGACTION_H
# define _KERNEL_SIGACTION_H
+#ifdef SA_RESTORER
+# define HAS_SA_RESTORER 1
+#endif
+
/* This is the sigaction structure from the Linux 3.2 kernel. */
struct kernel_sigaction
{
__sighandler_t k_sa_handler;
unsigned long sa_flags;
-#ifdef SA_RESTORER
+#ifdef HAS_SA_RESTORER
void (*sa_restorer) (void);
#endif
+ /* glibc sigset is larger than kernel expected one, however sigaction
+ passes the kernel expected size on rt_sigaction syscall. */
sigset_t sa_mask;
};
-#ifndef SA_RESTORER
+#ifndef SET_SA_RESTORER
# define SET_SA_RESTORER(kact, act)
+#endif
+#ifndef RESET_SA_RESTORER
# define RESET_SA_RESTORER(act, kact)
#endif
-#ifndef _KERNEL_SIGACTION_H
-# define _KERNEL_SIGACTION_H
-
-#include <signal.h>
-
-#define SA_RESTORER 0x04000000
-
-/* This is the sigaction structure from the Linux 3.2 kernel. */
-struct kernel_sigaction
-{
- __sighandler_t k_sa_handler;
- sigset_t sa_mask;
- unsigned long sa_flags;
- void (*sa_restorer) (void);
-};
-
-#define SET_SA_RESTORER(kact, act) \
- (kact)->sa_restorer = (act)->sa_restorer
-#define RESET_SA_RESTORER(act, kact) \
- (act)->sa_restorer = (kact)->sa_restorer
-
-#endif
+/* m68k does not define SA_RESTORER, but does have sa_restorer member
+ on kernel sigaction struct. */
+#define HAS_SA_RESTORER 1
+#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
# undef __ASSUME_EXECVEAT
#endif
-/* Support for the copy_file_range syscall was added in 4.10. */
-#if __LINUX_KERNEL_VERSION < 0x040A00
-# undef __ASSUME_COPY_FILE_RANGE
#endif
/* Support for statx was added in kernel 4.12. */
--- /dev/null
+/* mips64n32 uses __NR_mmap for mmap64 while still having sizeof (off_t)
+ smaller than sizeof (off64_t). So it allows mapping large offsets
+ using mmap64 than 32-bit archs which uses __NR_mmap2. */
+
+static inline uint64_t
+mmap64_maximum_offset (long int page_shift)
+{
+#if _MIPS_SIM == _ABIN32 || _MIPS_SIM == _ABI64
+ return UINT64_MAX;
+#else
+ return (UINT64_C(1) << (page_shift + (8 * sizeof (off_t)))) - 1;
+#endif
+}
#include <sysdep.h>
#include <mmap_internal.h>
+#ifdef __NR_mmap2
/* To avoid silent truncation of offset when using mmap2, do not accept
offset larger than 1 << (page_shift + off_t bits). For archictures with
32 bits off_t and page size of 4096 it would be 1^44. */
-#define MMAP_OFF_HIGH_MASK \
+# define MMAP_OFF_HIGH_MASK \
((-(MMAP2_PAGE_UNIT << 1) << (8 * sizeof (off_t) - 1)))
+#else
+/* Some ABIs might use __NR_mmap while having sizeof (off_t) smaller than
+ sizeof (off64_t) (currently only MIPS64n32). For this case just set
+ zero the higher bits so mmap with large offset does not fail. */
+# define MMAP_OFF_HIGH_MASK 0x0
+#endif
#define MMAP_OFF_MASK (MMAP_OFF_HIGH_MASK | MMAP_OFF_LOW_MASK)
char message[200];
if (family < 0)
__snprintf (message, sizeof (message),
- "Unexpected error %d on netlink descriptor %d",
+ "Unexpected error %d on netlink descriptor %d.\n",
error_code, fd);
else
__snprintf (message, sizeof (message),
"Unexpected error %d on netlink descriptor %d"
- " (address family %d)",
+ " (address family %d).\n",
error_code, fd, family);
__libc_fatal (message);
}
/* NIOS2 uses the generic Linux UAPI but defines SA_RESTORER. */
#define SA_RESTORER 0x04000000
-#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
#define SET_SA_RESTORER(kact, act) \
(kact)->sa_restorer = (act)->sa_restorer
#define RESET_SA_RESTORER(act, kact) \
(act)->sa_restorer = (kact)->sa_restorer
+
+#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
TUNABLE_CALLBACK (set_elision_skip_trylock_internal_abort));
#endif
+ /* Linux from 3.9 through 4.2 do not abort HTM transaction on syscalls,
+ instead it suspends the transaction and resumes it when returning to
+ usercode. The side-effects of the syscall will always remain visible,
+ even if the transaction is aborted. This is an issue when a transaction
+ is used along with futex syscall, on pthread_cond_wait for instance,
+ where futex might succeed but the transaction is rolled back leading
+ the condition variable object in an inconsistent state.
+
+ Glibc used to prevent it by always aborting a transaction before issuing
+ a syscall. Linux 4.2 also decided to abort active transaction in
+ syscalls which makes the glibc workaround superflours. Worse, glibc
+ transaction abortions leads to a performance issues on recent kernels.
+
+ So Lock Elision is just enabled when it has been explict set (either
+ by tunables of by a configure switch) and if kernel aborts HTM
+ transactions on syscalls (PPC_FEATURE2_HTM_NOSC) */
+
+ __pthread_force_elision = (__pthread_force_elision
+ && GLRO (dl_hwcap2) & PPC_FEATURE2_HTM_NOSC);
+
if (!__pthread_force_elision)
__elision_aconf.try_tbegin = 0; /* Disable elision on rwlocks. */
}
/* Automatically enable elision for existing user lock kinds. */
#define FORCE_ELISION(m, s) \
- if (__pthread_force_elision \
- && (m->__data.__kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ if (__pthread_force_elision) \
{ \
- mutex->__data.__kind |= PTHREAD_MUTEX_ELISION_NP; \
- s; \
+ /* See concurrency notes regarding __kind in \
+ struct __pthread_mutex_s in \
+ sysdeps/nptl/bits/thread-shared-types.h. \
+ \
+ There are the following cases for the kind of a mutex \
+ (The mask PTHREAD_MUTEX_ELISION_FLAGS_NP covers the flags \
+ PTHREAD_MUTEX_ELISION_NP and PTHREAD_MUTEX_NO_ELISION_NP where \
+ only one of both flags can be set): \
+ - both flags are not set: \
+ This is the first lock operation for this mutex. Enable \
+ elision as it is not enabled so far. \
+ Note: It can happen that multiple threads are calling e.g. \
+ pthread_mutex_lock at the same time as the first lock \
+ operation for this mutex. Then elision is enabled for this \
+ mutex by multiple threads. Storing with relaxed MO is enough \
+ as all threads will store the same new value for the kind of \
+ the mutex. But we have to ensure that we always use the \
+ elision path regardless if this thread has enabled elision or \
+ another one. \
+ \
+ - PTHREAD_MUTEX_ELISION_NP flag is set: \
+ Elision was already enabled for this mutex by a previous lock \
+ operation. See case above. Just use the elision path. \
+ \
+ - PTHREAD_MUTEX_NO_ELISION_NP flag is set: \
+ Elision was explicitly disabled by pthread_mutexattr_settype. \
+ Do not use the elision path. \
+ Note: The flag PTHREAD_MUTEX_NO_ELISION_NP will never be \
+ changed after mutex initialization. */ \
+ int mutex_kind = atomic_load_relaxed (&((m)->__data.__kind)); \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ { \
+ mutex_kind |= PTHREAD_MUTEX_ELISION_NP; \
+ atomic_store_relaxed (&((m)->__data.__kind), mutex_kind); \
+ } \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_NP) != 0) \
+ { \
+ s; \
+ } \
}
/* powerpc kernel sigaction is similar to generic Linux UAPI one,
but the architecture also defines SA_RESTORER. */
#define SA_RESTORER 0x04000000
-#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
#define SET_SA_RESTORER(kact, act) \
(kact)->sa_restorer = (act)->sa_restorer
#define RESET_SA_RESTORER(act, kact) \
(act)->sa_restorer = (kact)->sa_restorer
+
+#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
+++ /dev/null
-/* Syscall wrapper that do not set errno. Linux powerpc version.
- Copyright (C) 2018 Free Software Foundation, Inc.
- This file is part of the GNU C Library.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <http://www.gnu.org/licenses/>. */
-
-/* __access_noerrno is used during process initialization in elf/dl-tunables.c
- before the TCB is initialized, prohibiting the usage of
- ABORT_TRANSACTION. */
-#undef ABORT_TRANSACTION
-#define ABORT_TRANSACTION
-
-#include "sysdeps/unix/sysv/linux/not-errno.h"
-
-/* Recover ABORT_TRANSACTION's previous value, in order to not affect
- other syscalls. */
-#undef ABORT_TRANSACTION
-#define ABORT_TRANSACTION ABORT_TRANSACTION_IMPL
register long int r11 __asm__ ("r11"); \
register long int r12 __asm__ ("r12"); \
LOADARGS_##nr(name, args); \
- ABORT_TRANSACTION; \
__asm__ __volatile__ \
("sc \n\t" \
"mfcr %0" \
register long int r7 __asm__ ("r7"); \
register long int r8 __asm__ ("r8"); \
LOADARGS_##nr (name, ##args); \
- ABORT_TRANSACTION; \
__asm__ __volatile__ \
("sc\n\t" \
"mfcr %0\n\t" \
#include <sysdep.h>
ENTRY (syscall)
- ABORT_TRANSACTION
mr r0,r3
mr r3,r4
mr r4,r5
# ifdef __NR_preadv2
ssize_t result = SYSCALL_CANCEL (preadv2, fd, vector, count,
LO_HI_LONG (offset), flags);
- if (result >= 0)
+ if (result >= 0 || errno != ENOSYS)
return result;
# endif
/* Trying to emulate the preadv2 syscall flags is troublesome:
#ifdef __NR_preadv64v2
ssize_t result = SYSCALL_CANCEL (preadv64v2, fd, vector, count,
LO_HI_LONG (offset), flags);
- if (result >= 0)
+ if (result >= 0 || errno != ENOSYS)
return result;
#endif
/* Trying to emulate the preadv2 syscall flags is troublesome:
# ifdef __NR_pwritev2
ssize_t result = SYSCALL_CANCEL (pwritev2, fd, vector, count,
LO_HI_LONG (offset), flags);
- if (result >= 0)
+ if (result >= 0 || errno != ENOSYS)
return result;
# endif
/* Trying to emulate the pwritev2 syscall flags is troublesome:
#ifdef __NR_pwritev64v2
ssize_t result = SYSCALL_CANCEL (pwritev64v2, fd, vector, count,
LO_HI_LONG (offset), flags);
- if (result >= 0)
+ if (result >= 0 || errno != ENOSYS)
return result;
#endif
/* Trying to emulate the pwritev2 syscall flags is troublesome:
ENTRY (__thread_start)
L (thread_start):
+ /* Terminate call stack by noting ra is undefined. Use a dummy
+ .cfi_label to force starting the FDE. */
+ .cfi_label .Ldummy
+ cfi_undefined (ra)
+
/* Restore the arg for user's function. */
REG_L a1,0(sp) /* Function pointer. */
REG_L a0,SZREG(sp) /* Argument pointer. */
#include <stdlib.h>
#include <atomic.h>
#include <sys/cachectl.h>
-#include <asm/syscalls.h>
+#if __has_include__ (<asm/syscalls.h>)
+# include <asm/syscalls.h>
+#else
+# include <asm/unistd.h>
+#endif
typedef int (*func_type) (void *, void *, unsigned long int);
#undef __ASSUME_CLONE_DEFAULT
#define __ASSUME_CLONE_BACKWARDS 1
+
+/* No support for PI mutexes or robust futexes before 4.20. */
+#if __LINUX_KERNEL_VERSION < 0x041400
+# undef __ASSUME_SET_ROBUST_LIST
+#endif
/* Automatically enable elision for existing user lock kinds. */
#define FORCE_ELISION(m, s) \
- if (__pthread_force_elision \
- && (m->__data.__kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ if (__pthread_force_elision) \
{ \
- mutex->__data.__kind |= PTHREAD_MUTEX_ELISION_NP; \
- s; \
+ /* See concurrency notes regarding __kind in \
+ struct __pthread_mutex_s in \
+ sysdeps/nptl/bits/thread-shared-types.h. \
+ \
+ There are the following cases for the kind of a mutex \
+ (The mask PTHREAD_MUTEX_ELISION_FLAGS_NP covers the flags \
+ PTHREAD_MUTEX_ELISION_NP and PTHREAD_MUTEX_NO_ELISION_NP where \
+ only one of both flags can be set): \
+ - both flags are not set: \
+ This is the first lock operation for this mutex. Enable \
+ elision as it is not enabled so far. \
+ Note: It can happen that multiple threads are calling e.g. \
+ pthread_mutex_lock at the same time as the first lock \
+ operation for this mutex. Then elision is enabled for this \
+ mutex by multiple threads. Storing with relaxed MO is enough \
+ as all threads will store the same new value for the kind of \
+ the mutex. But we have to ensure that we always use the \
+ elision path regardless if this thread has enabled elision or \
+ another one. \
+ \
+ - PTHREAD_MUTEX_ELISION_NP flag is set: \
+ Elision was already enabled for this mutex by a previous lock \
+ operation. See case above. Just use the elision path. \
+ \
+ - PTHREAD_MUTEX_NO_ELISION_NP flag is set: \
+ Elision was explicitly disabled by pthread_mutexattr_settype. \
+ Do not use the elision path. \
+ Note: The flag PTHREAD_MUTEX_NO_ELISION_NP will never be \
+ changed after mutex initialization. */ \
+ int mutex_kind = atomic_load_relaxed (&((m)->__data.__kind)); \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ { \
+ mutex_kind |= PTHREAD_MUTEX_ELISION_NP; \
+ atomic_store_relaxed (&((m)->__data.__kind), mutex_kind); \
+ } \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_NP) != 0) \
+ { \
+ s; \
+ } \
}
/* SH uses the generic Linux UAPI but defines SA_RESTORER. */
#define SA_RESTORER 0x04000000
-#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
#define SET_SA_RESTORER(kact, act) \
(kact)->sa_restorer = (act)->sa_restorer
#define RESET_SA_RESTORER(act, kact) \
(act)->sa_restorer = (kact)->sa_restorer
+
+#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
#ifndef _BITS_SIGINFO_ARCH_H
#define _BITS_SIGINFO_ARCH_H 1
-#define __SI_BAND_TYPE int
+/* The kernel uses int instead of long int (as in POSIX). In 32-bit
+ mode, we can still use long int, but in 64-bit mode, we need to
+ deviate from POSIX. */
+#if __WORDSIZE == 64
+# define __SI_BAND_TYPE int
+#endif
#define __SI_SIGFAULT_ADDL \
int _si_trapno;
/* SPARC 'struct __new_sigaction' is similar to generic Linux UAPI with
a sa_restorer field, even though function is passed as an argument
to rt_sigaction syscall. */
-#define SA_RESTORER 0x04000000
+#define HAS_SA_RESTORER 1
#include <sysdeps/unix/sysv/linux/kernel_sigaction.h>
-
-#define SET_SA_RESTORER(kact, act) \
- (kact)->sa_restorer = NULL
-#define RESET_SA_RESTORER(act, kact) \
- (act)->sa_restorer = (kact)->sa_restorer
ifeq ($(subdir),stdlib)
sysdep_routines += __start_context
endif
+
+ifeq ($(subdir),conform)
+# For bug 23821 (incorrect type of si_band).
+conformtest-xfail-conds += sparc64-linux
+endif
ptrdiff_t argc = args->argc;
/* Construct an argument list for the shell. */
- char *new_argv[argc + 1];
+ char *new_argv[argc + 2];
new_argv[0] = (char *) _PATH_BSHELL;
new_argv[1] = (char *) args->file;
if (argc > 1)
# names are only used if the installed kernel headers also provide
# them.
-# The list of system calls is current as of Linux 4.17.
-kernel 4.17
+# The list of system calls is current as of Linux 4.19.
+kernel 4.19
FAST_atomic_update
FAST_cmpxchg
io_cancel
io_destroy
io_getevents
+io_pgetevents
io_setup
io_submit
ioctl
request_key
restart_syscall
rmdir
+rseq
rt_sigaction
rt_sigpending
rt_sigprocmask
--- /dev/null
+/* Test readdir64 compatibility symbol.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <dirent.h>
+#include <dlfcn.h>
+#include <errno.h>
+#include <shlib-compat.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <string.h>
+#include <support/check.h>
+
+/* Copied from <olddirent.h>. */
+struct __old_dirent64
+ {
+ __ino_t d_ino;
+ __off64_t d_off;
+ unsigned short int d_reclen;
+ unsigned char d_type;
+ char d_name[256];
+ };
+
+typedef struct __old_dirent64 *(*compat_readdir64_type) (DIR *);
+
+#if TEST_COMPAT (libc, GLIBC_2_1, GLIBC_2_2)
+struct __old_dirent64 *compat_readdir64 (DIR *);
+compat_symbol_reference (libc, compat_readdir64, readdir64, GLIBC_2_1);
+#endif
+
+static int
+do_test (void)
+{
+#if TEST_COMPAT (libc, GLIBC_2_1, GLIBC_2_2)
+
+ /* Directory stream using the non-compat readdir64 symbol. The test
+ checks against this. */
+ DIR *dir_reference = opendir (".");
+ TEST_VERIFY_EXIT (dir_reference != NULL);
+ DIR *dir_test = opendir (".");
+ TEST_VERIFY_EXIT (dir_test != NULL);
+
+ /* This loop assumes that the enumeration order is consistent for
+ two different handles. Nothing should write to the current
+ directory (in the source tree) while this test runs, so there
+ should not be any difference due to races. */
+ size_t count = 0;
+ while (true)
+ {
+ errno = 0;
+ struct dirent64 *entry_reference = readdir64 (dir_reference);
+ if (entry_reference == NULL && errno != 0)
+ FAIL_EXIT1 ("readdir64 entry %zu: %m\n", count);
+ struct __old_dirent64 *entry_test = compat_readdir64 (dir_test);
+ if (entry_reference == NULL)
+ {
+ if (errno == EOVERFLOW)
+ {
+ TEST_VERIFY (entry_reference->d_ino
+ != (__ino_t) entry_reference->d_ino);
+ printf ("info: inode number overflow at entry %zu\n", count);
+ break;
+ }
+ if (errno != 0)
+ FAIL_EXIT1 ("compat readdir64 entry %zu: %m\n", count);
+ }
+
+ /* Check that both streams end at the same time. */
+ if (entry_reference == NULL)
+ {
+ TEST_VERIFY (entry_test == NULL);
+ break;
+ }
+ else
+ TEST_VERIFY_EXIT (entry_test != NULL);
+
+ /* d_off is never zero because it is the offset of the next
+ entry (not the current entry). */
+ TEST_VERIFY (entry_reference->d_off > 0);
+
+ /* Check that the entries are the same. */
+ TEST_COMPARE_BLOB (entry_reference->d_name,
+ strlen (entry_reference->d_name),
+ entry_test->d_name, strlen (entry_test->d_name));
+ TEST_COMPARE (entry_reference->d_ino, entry_test->d_ino);
+ TEST_COMPARE (entry_reference->d_off, entry_test->d_off);
+ TEST_COMPARE (entry_reference->d_type, entry_test->d_type);
+ TEST_COMPARE (entry_reference->d_reclen, entry_test->d_reclen);
+
+ ++count;
+ }
+ printf ("info: %zu directory entries found\n", count);
+ TEST_VERIFY (count >= 3); /* ".", "..", and some source files. */
+
+ TEST_COMPARE (closedir (dir_test), 0);
+ TEST_COMPARE (closedir (dir_reference), 0);
+#endif
+ return 0;
+}
+
+#include <support/test-driver.c>
/* Automatically enable elision for existing user lock kinds. */
#define FORCE_ELISION(m, s) \
- if (__pthread_force_elision \
- && (m->__data.__kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ if (__pthread_force_elision) \
{ \
- mutex->__data.__kind |= PTHREAD_MUTEX_ELISION_NP; \
- s; \
+ /* See concurrency notes regarding __kind in \
+ struct __pthread_mutex_s in \
+ sysdeps/nptl/bits/thread-shared-types.h. \
+ \
+ There are the following cases for the kind of a mutex \
+ (The mask PTHREAD_MUTEX_ELISION_FLAGS_NP covers the flags \
+ PTHREAD_MUTEX_ELISION_NP and PTHREAD_MUTEX_NO_ELISION_NP where \
+ only one of both flags can be set): \
+ - both flags are not set: \
+ This is the first lock operation for this mutex. Enable \
+ elision as it is not enabled so far. \
+ Note: It can happen that multiple threads are calling e.g. \
+ pthread_mutex_lock at the same time as the first lock \
+ operation for this mutex. Then elision is enabled for this \
+ mutex by multiple threads. Storing with relaxed MO is enough \
+ as all threads will store the same new value for the kind of \
+ the mutex. But we have to ensure that we always use the \
+ elision path regardless if this thread has enabled elision or \
+ another one. \
+ \
+ - PTHREAD_MUTEX_ELISION_NP flag is set: \
+ Elision was already enabled for this mutex by a previous lock \
+ operation. See case above. Just use the elision path. \
+ \
+ - PTHREAD_MUTEX_NO_ELISION_NP flag is set: \
+ Elision was explicitly disabled by pthread_mutexattr_settype. \
+ Do not use the elision path. \
+ Note: The flag PTHREAD_MUTEX_NO_ELISION_NP will never be \
+ changed after mutex initialization. */ \
+ int mutex_kind = atomic_load_relaxed (&((m)->__data.__kind)); \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_FLAGS_NP) == 0) \
+ { \
+ mutex_kind |= PTHREAD_MUTEX_ELISION_NP; \
+ atomic_store_relaxed (&((m)->__data.__kind), mutex_kind); \
+ } \
+ if ((mutex_kind & PTHREAD_MUTEX_ELISION_NP) != 0) \
+ { \
+ s; \
+ } \
}
#include <signal.h>
#define SA_RESTORER 0x04000000
-#include <kernel_sigaction.h>
extern void restore_rt (void) asm ("__restore_rt") attribute_hidden;
#define RESET_SA_RESTORER(act, kact) \
(act)->sa_restorer = (kact)->sa_restorer
+#include <kernel_sigaction.h>
+
#include <sysdeps/unix/sysv/linux/sigaction.c>
/* NOTE: Please think twice before making any changes to the bits of
| bit_arch_Fast_Unaligned_Copy
| bit_arch_Prefer_PMINUB_for_stringop);
break;
+ }
+ /* Disable TSX on some Haswell processors to avoid TSX on kernels that
+ weren't updated with the latest microcode package (which disables
+ broken feature by default). */
+ switch (model)
+ {
case 0x3f:
/* Xeon E7 v3 with stepping >= 4 has working TSX. */
if (stepping >= 4)
const ElfW(Addr) align)
{
#if CET_ENABLED
+ /* Skip if we have seen a NT_GNU_PROPERTY_TYPE_0 note before. */
+ if (l->l_cet != lc_unknown)
+ return;
+
/* The NT_GNU_PROPERTY_TYPE_0 note must be aliged to 4 bytes in
32-bit objects and to 8 bytes in 64-bit objects. Skip notes
with incorrect alignment. */
const ElfW(Addr) start = (ElfW(Addr)) note;
+ unsigned int feature_1 = 0;
+ unsigned int last_type = 0;
+
while ((ElfW(Addr)) (note + 1) - start < size)
{
/* Find the NT_GNU_PROPERTY_TYPE_0 note. */
&& note->n_type == NT_GNU_PROPERTY_TYPE_0
&& memcmp (note + 1, "GNU", 4) == 0)
{
+ /* Stop if we see more than one GNU property note which may
+ be generated by the older linker. */
+ if (l->l_cet != lc_unknown)
+ return;
+
+ /* Check CET status now. */
+ l->l_cet = lc_none;
+
/* Check for invalid property. */
if (note->n_descsz < 8
|| (note->n_descsz % sizeof (ElfW(Addr))) != 0)
- break;
+ return;
/* Start and end of property array. */
unsigned char *ptr = (unsigned char *) (note + 1) + 4;
unsigned int type = *(unsigned int *) ptr;
unsigned int datasz = *(unsigned int *) (ptr + 4);
+ /* Property type must be in ascending order. */
+ if (type < last_type)
+ return;
+
ptr += 8;
if ((ptr + datasz) > ptr_end)
- break;
+ return;
+
+ last_type = type;
if (type == GNU_PROPERTY_X86_FEATURE_1_AND)
{
we stop the search regardless if its size is correct
or not. There is no point to continue if this note
is ill-formed. */
- if (datasz == 4)
- {
- unsigned int feature_1 = *(unsigned int *) ptr;
- if ((feature_1 & GNU_PROPERTY_X86_FEATURE_1_IBT))
- l->l_cet |= lc_ibt;
- if ((feature_1 & GNU_PROPERTY_X86_FEATURE_1_SHSTK))
- l->l_cet |= lc_shstk;
- }
+ if (datasz != 4)
+ return;
+
+ feature_1 = *(unsigned int *) ptr;
+
+ /* Keep searching for the next GNU property note
+ generated by the older linker. */
+ break;
+ }
+ else if (type > GNU_PROPERTY_X86_FEATURE_1_AND)
+ {
+ /* Stop since property type is in ascending order. */
return;
}
+ ELF_NOTE_NEXT_OFFSET (note->n_namesz, note->n_descsz,
align));
}
+
+ /* We get here only if there is one or no GNU property note. */
+ if ((feature_1 & GNU_PROPERTY_X86_FEATURE_1_IBT))
+ l->l_cet |= lc_ibt;
+ if ((feature_1 & GNU_PROPERTY_X86_FEATURE_1_SHSTK))
+ l->l_cet |= lc_shstk;
#endif
}
/* If this object is enabled with CET. */
enum
{
- lc_none = 0, /* Not enabled with CET. */
- lc_ibt = 1 << 0, /* Enabled with IBT. */
- lc_shstk = 1 << 1, /* Enabled with STSHK. */
+ lc_unknown = 0, /* Unknown CET status. */
+ lc_none = 1 << 0, /* Not enabled with CET. */
+ lc_ibt = 1 << 1, /* Enabled with IBT. */
+ lc_shstk = 1 << 2, /* Enabled with STSHK. */
lc_ibt_and_shstk = lc_ibt | lc_shstk /* Enabled with both. */
- } l_cet:2;
+ } l_cet:3;
*reloc_addr = value;
}
else
- __libc_fatal ("unexpected reloc type in static binary");
+ __libc_fatal ("Unexpected reloc type in static binary.\n");
}
#endif /* dl-irel.h */
mov %edi, %ecx
#ifdef USE_AS_WMEMCHR
- test %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jz L(return_null)
- shl $2, %rdx
+ shl $2, %RDX_LP
#else
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
+# endif
punpcklbw %xmm1, %xmm1
- test %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jz L(return_null)
punpcklbw %xmm1, %xmm1
#endif
.text
ENTRY (memcmp)
- test %rdx, %rdx
+#ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
+#endif
+ test %RDX_LP, %RDX_LP
jz L(finz)
cmpq $1, %rdx
- jle L(finr1b)
+ jbe L(finr1b)
subq %rdi, %rsi
movq %rdx, %r10
cmpq $32, %r10
- jge L(gt32)
+ jae L(gt32)
/* Handle small chunks and last block of less than 32 bytes. */
L(small):
testq $1, %r10
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
- jge L(mt16)
+ jae L(mt16)
/* Pre-unroll to be ready for unrolled 64B loop. */
testq $32, %rdi
jz L(A64)
movq %r11, %r10
andq $-64, %r10
cmpq %r10, %rdi
- jge L(mt32)
+ jae L(mt32)
L(A64main):
movdqu (%rdi,%rsi), %xmm0
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
- jge L(mt16)
+ jae L(mt16)
L(A32main):
movdqu (%rdi,%rsi), %xmm0
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
- jge L(mt16)
+ jae L(mt16)
testq $16, %rdi
jz L(ATR32)
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
- jge L(mt16)
+ jae L(mt16)
L(ATR32res):
movdqa (%rdi,%rsi), %xmm0
ENTRY (__memrchr)
movd %esi, %xmm1
- sub $16, %rdx
+ sub $16, %RDX_LP
jbe L(length_less16)
punpcklbw %xmm1, %xmm1
punpcklbw %xmm1, %xmm1
- add %rdx, %rdi
+ add %RDX_LP, %RDI_LP
pshufd $0, %xmm1, %xmm1
movdqu (%rdi), %xmm0
ENTRY (MEMCHR)
# ifndef USE_AS_RAWMEMCHR
/* Check for zero length. */
- testq %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jz L(null)
# endif
movl %edi, %ecx
/* Broadcast CHAR to YMM0. */
vmovd %esi, %xmm0
# ifdef USE_AS_WMEMCHR
- shl $2, %rdx
+ shl $2, %RDX_LP
vpbroadcastd %xmm0, %ymm0
# else
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
+# endif
vpbroadcastb %xmm0, %ymm0
# endif
/* Check if we may cross page boundary with one vector load. */
.section .text.avx,"ax",@progbits
ENTRY (MEMCMP)
# ifdef USE_AS_WMEMCMP
- shl $2, %rdx
+ shl $2, %RDX_LP
+# elif defined __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
# endif
- cmpq $VEC_SIZE, %rdx
+ cmp $VEC_SIZE, %RDX_LP
jb L(less_vec)
/* From VEC to 2 * VEC. No branch when size == VEC_SIZE. */
.section .text.sse4.1,"ax",@progbits
ENTRY (MEMCMP)
# ifdef USE_AS_WMEMCMP
- shl $2, %rdx
+ shl $2, %RDX_LP
+# elif defined __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
# endif
pxor %xmm0, %xmm0
- cmp $79, %rdx
+ cmp $79, %RDX_LP
ja L(79bytesormore)
# ifndef USE_AS_WMEMCMP
- cmp $1, %rdx
+ cmp $1, %RDX_LP
je L(firstbyte)
# endif
add %rdx, %rsi
atom_text_section
ENTRY (MEMCMP)
# ifdef USE_AS_WMEMCMP
- shl $2, %rdx
- test %rdx, %rdx
+ shl $2, %RDX_LP
+ test %RDX_LP, %RDX_LP
jz L(equal)
+# elif defined __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
# endif
mov %rdx, %rcx
mov %rdi, %rdx
.section .text.ssse3,"ax",@progbits
#if !defined USE_AS_MEMPCPY && !defined USE_AS_MEMMOVE
ENTRY (MEMPCPY_CHK)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMPCPY_CHK)
ENTRY (MEMPCPY)
- movq %rdi, %rax
- addq %rdx, %rax
+ mov %RDI_LP, %RAX_LP
+ add %RDX_LP, %RAX_LP
jmp L(start)
END (MEMPCPY)
#endif
#if !defined USE_AS_BCOPY
ENTRY (MEMCPY_CHK)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMCPY_CHK)
#endif
ENTRY (MEMCPY)
- mov %rdi, %rax
+ mov %RDI_LP, %RAX_LP
#ifdef USE_AS_MEMPCPY
- add %rdx, %rax
+ add %RDX_LP, %RAX_LP
+#endif
+
+#ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
#endif
#ifdef USE_AS_MEMMOVE
.section .text.ssse3,"ax",@progbits
#if !defined USE_AS_MEMPCPY && !defined USE_AS_MEMMOVE
ENTRY (MEMPCPY_CHK)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMPCPY_CHK)
ENTRY (MEMPCPY)
- movq %rdi, %rax
- addq %rdx, %rax
+ mov %RDI_LP, %RAX_LP
+ add %RDX_LP, %RAX_LP
jmp L(start)
END (MEMPCPY)
#endif
#if !defined USE_AS_BCOPY
ENTRY (MEMCPY_CHK)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMCPY_CHK)
#endif
ENTRY (MEMCPY)
- mov %rdi, %rax
+ mov %RDI_LP, %RAX_LP
#ifdef USE_AS_MEMPCPY
- add %rdx, %rax
+ add %RDX_LP, %RAX_LP
+#endif
+
+#ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
#endif
#ifdef USE_AS_MEMMOVE
.section .text.avx512,"ax",@progbits
ENTRY (__mempcpy_chk_avx512_no_vzeroupper)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (__mempcpy_chk_avx512_no_vzeroupper)
ENTRY (__mempcpy_avx512_no_vzeroupper)
- movq %rdi, %rax
- addq %rdx, %rax
+ mov %RDI_LP, %RAX_LP
+ add %RDX_LP, %RAX_LP
jmp L(start)
END (__mempcpy_avx512_no_vzeroupper)
ENTRY (__memmove_chk_avx512_no_vzeroupper)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (__memmove_chk_avx512_no_vzeroupper)
ENTRY (__memmove_avx512_no_vzeroupper)
- mov %rdi, %rax
+ mov %RDI_LP, %RAX_LP
# ifdef USE_AS_MEMPCPY
- add %rdx, %rax
+ add %RDX_LP, %RAX_LP
# endif
L(start):
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
+# endif
lea (%rsi, %rdx), %rcx
lea (%rdi, %rdx), %r9
cmp $512, %rdx
.section SECTION(.text),"ax",@progbits
#if defined SHARED && IS_IN (libc)
ENTRY (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned))
#endif
ENTRY (MEMPCPY_SYMBOL (__mempcpy, unaligned))
- movq %rdi, %rax
- addq %rdx, %rax
+ mov %RDI_LP, %RAX_LP
+ add %RDX_LP, %RAX_LP
jmp L(start)
END (MEMPCPY_SYMBOL (__mempcpy, unaligned))
#if defined SHARED && IS_IN (libc)
ENTRY (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned))
#endif
ENTRY (MEMMOVE_SYMBOL (__memmove, unaligned))
movq %rdi, %rax
L(start):
- cmpq $VEC_SIZE, %rdx
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
+# endif
+ cmp $VEC_SIZE, %RDX_LP
jb L(less_vec)
- cmpq $(VEC_SIZE * 2), %rdx
+ cmp $(VEC_SIZE * 2), %RDX_LP
ja L(more_2x_vec)
#if !defined USE_MULTIARCH || !IS_IN (libc)
L(last_2x_vec):
# if VEC_SIZE == 16
ENTRY (__mempcpy_chk_erms)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (__mempcpy_chk_erms)
/* Only used to measure performance of REP MOVSB. */
ENTRY (__mempcpy_erms)
- movq %rdi, %rax
+ mov %RDI_LP, %RAX_LP
/* Skip zero length. */
- testq %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jz 2f
- addq %rdx, %rax
+ add %RDX_LP, %RAX_LP
jmp L(start_movsb)
END (__mempcpy_erms)
ENTRY (__memmove_chk_erms)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (__memmove_chk_erms)
ENTRY (__memmove_erms)
movq %rdi, %rax
/* Skip zero length. */
- testq %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jz 2f
L(start_movsb):
- movq %rdx, %rcx
- cmpq %rsi, %rdi
+ mov %RDX_LP, %RCX_LP
+ cmp %RSI_LP, %RDI_LP
jb 1f
/* Source == destination is less common. */
je 2f
- leaq (%rsi,%rcx), %rdx
- cmpq %rdx, %rdi
+ lea (%rsi,%rcx), %RDX_LP
+ cmp %RDX_LP, %RDI_LP
jb L(movsb_backward)
1:
rep movsb
# ifdef SHARED
ENTRY (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned_erms))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMMOVE_CHK_SYMBOL (__mempcpy_chk, unaligned_erms))
# endif
ENTRY (MEMMOVE_SYMBOL (__mempcpy, unaligned_erms))
- movq %rdi, %rax
- addq %rdx, %rax
+ mov %RDI_LP, %RAX_LP
+ add %RDX_LP, %RAX_LP
jmp L(start_erms)
END (MEMMOVE_SYMBOL (__mempcpy, unaligned_erms))
# ifdef SHARED
ENTRY (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned_erms))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMMOVE_CHK_SYMBOL (__memmove_chk, unaligned_erms))
# endif
ENTRY (MEMMOVE_SYMBOL (__memmove, unaligned_erms))
movq %rdi, %rax
L(start_erms):
- cmpq $VEC_SIZE, %rdx
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %edx, %edx
+# endif
+ cmp $VEC_SIZE, %RDX_LP
jb L(less_vec)
- cmpq $(VEC_SIZE * 2), %rdx
+ cmp $(VEC_SIZE * 2), %RDX_LP
ja L(movsb_more_2x_vec)
L(last_2x_vec):
/* From VEC and to 2 * VEC. No branch when size == VEC_SIZE. */
# endif
jb L(more_8x_vec_backward)
1:
- movq %rdx, %rcx
+ mov %RDX_LP, %RCX_LP
rep movsb
L(nop):
ret
vmovd %esi, %xmm0
vpbroadcastb %xmm0, %ymm0
- subq $VEC_SIZE, %rdx
+ sub $VEC_SIZE, %RDX_LP
jbe L(last_vec_or_less)
- addq %rdx, %rdi
+ add %RDX_LP, %RDI_LP
/* Check the last VEC_SIZE bytes. */
vpcmpeqb (%rdi), %ymm0, %ymm1
.section .text.avx512,"ax",@progbits
#if defined PIC
ENTRY (MEMSET_CHK)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (MEMSET_CHK)
#endif
ENTRY (MEMSET)
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
+# endif
vpxor %xmm0, %xmm0, %xmm0
vmovd %esi, %xmm1
lea (%rdi, %rdx), %rsi
.section SECTION(.text),"ax",@progbits
#if VEC_SIZE == 16 && IS_IN (libc)
ENTRY (__bzero)
- movq %rdi, %rax /* Set return value. */
- movq %rsi, %rdx /* Set n. */
+ mov %RDI_LP, %RAX_LP /* Set return value. */
+ mov %RSI_LP, %RDX_LP /* Set n. */
pxor %xmm0, %xmm0
jmp L(entry_from_bzero)
END (__bzero)
#if IS_IN (libc)
# if defined SHARED
ENTRY_CHK (WMEMSET_CHK_SYMBOL (__wmemset_chk, unaligned))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END_CHK (WMEMSET_CHK_SYMBOL (__wmemset_chk, unaligned))
# endif
ENTRY (WMEMSET_SYMBOL (__wmemset, unaligned))
- shlq $2, %rdx
+ shl $2, %RDX_LP
WMEMSET_VDUP_TO_VEC0_AND_SET_RETURN (%esi, %rdi)
jmp L(entry_from_bzero)
END (WMEMSET_SYMBOL (__wmemset, unaligned))
#if defined SHARED && IS_IN (libc)
ENTRY_CHK (MEMSET_CHK_SYMBOL (__memset_chk, unaligned))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END_CHK (MEMSET_CHK_SYMBOL (__memset_chk, unaligned))
#endif
ENTRY (MEMSET_SYMBOL (__memset, unaligned))
MEMSET_VDUP_TO_VEC0_AND_SET_RETURN (%esi, %rdi)
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
+# endif
L(entry_from_bzero):
cmpq $VEC_SIZE, %rdx
jb L(less_vec)
# if VEC_SIZE == 16
ENTRY (__memset_chk_erms)
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END (__memset_chk_erms)
/* Only used to measure performance of REP STOSB. */
ENTRY (__memset_erms)
/* Skip zero length. */
- testq %rdx, %rdx
+ test %RDX_LP, %RDX_LP
jnz L(stosb)
movq %rdi, %rax
ret
L(stosb):
/* Issue vzeroupper before rep stosb. */
VZEROUPPER
- movq %rdx, %rcx
+ mov %RDX_LP, %RCX_LP
movzbl %sil, %eax
- movq %rdi, %rdx
+ mov %RDI_LP, %RDX_LP
rep stosb
- movq %rdx, %rax
+ mov %RDX_LP, %RAX_LP
ret
# if VEC_SIZE == 16
END (__memset_erms)
# if defined SHARED && IS_IN (libc)
ENTRY_CHK (MEMSET_CHK_SYMBOL (__memset_chk, unaligned_erms))
- cmpq %rdx, %rcx
+ cmp %RDX_LP, %RCX_LP
jb HIDDEN_JUMPTARGET (__chk_fail)
END_CHK (MEMSET_CHK_SYMBOL (__memset_chk, unaligned_erms))
# endif
ENTRY (MEMSET_SYMBOL (__memset, unaligned_erms))
MEMSET_VDUP_TO_VEC0_AND_SET_RETURN (%esi, %rdi)
- cmpq $VEC_SIZE, %rdx
+# ifdef __ILP32__
+ /* Clear the upper 32 bits. */
+ mov %edx, %edx
+# endif
+ cmp $VEC_SIZE, %RDX_LP
jb L(less_vec)
- cmpq $(VEC_SIZE * 2), %rdx
+ cmp $(VEC_SIZE * 2), %RDX_LP
ja L(stosb_more_2x_vec)
/* From VEC and to 2 * VEC. No branch when size == VEC_SIZE. */
VMOVU %VEC(0), -VEC_SIZE(%rdi,%rdx)
ENTRY (STRCMP)
# ifdef USE_AS_STRNCMP
/* Check for simple cases (0 or 1) in offset. */
- cmp $1, %rdx
+ cmp $1, %RDX_LP
je L(char0)
jb L(zero)
# ifdef USE_AS_WCSCMP
/* Convert units: from wide to byte char. */
- shl $2, %rdx
+ shl $2, %RDX_LP
# endif
/* Register %r11 tracks the maximum offset. */
- movq %rdx, %r11
+ mov %RDX_LP, %R11_LP
# endif
movl %edi, %eax
xorl %edx, %edx
#endif
#if defined USE_AS_STRNCMP || defined USE_AS_STRNCASECMP_L
- test %rdx, %rdx
+ test %RDX_LP, %RDX_LP
je LABEL(strcmp_exitz)
- cmp $1, %rdx
+ cmp $1, %RDX_LP
je LABEL(Byte0)
- mov %rdx, %r11
+ mov %RDX_LP, %R11_LP
#endif
mov %esi, %ecx
mov %edi, %eax
.text
ENTRY (STRCPY)
# ifdef USE_AS_STRNCPY
- mov %rdx, %r8
- test %r8, %r8
+ mov %RDX_LP, %R8_LP
+ test %R8_LP, %R8_LP
jz L(ExitZero)
# endif
mov %rsi, %rcx
mov %rsi, %rcx
# ifdef USE_AS_STRNCPY
- mov %rdx, %r8
+ mov %RDX_LP, %R8_LP
# endif
mov %rdi, %rdx
# ifdef USE_AS_STRNCPY
- test %r8, %r8
+ test %R8_LP, %R8_LP
jz L(Exit0)
- cmp $8, %r8
+ cmp $8, %R8_LP
jbe L(StrncpyExit8Bytes)
# endif
cmpb $0, (%rcx)
ENTRY (STRLEN)
# ifdef USE_AS_STRNLEN
/* Check for zero length. */
- testq %rsi, %rsi
+ test %RSI_LP, %RSI_LP
jz L(zero)
# ifdef USE_AS_WCSLEN
- shl $2, %rsi
+ shl $2, %RSI_LP
+# elif defined __ILP32__
+ /* Clear the upper 32 bits. */
+ movl %esi, %esi
# endif
- movq %rsi, %r8
+ mov %RSI_LP, %R8_LP
# endif
movl %edi, %ecx
movq %rdi, %rdx
* This implementation uses SSE to compare up to 16 bytes at a time.
*/
#if defined USE_AS_STRNCMP || defined USE_AS_STRNCASECMP_L
- test %rdx, %rdx
+ test %RDX_LP, %RDX_LP
je LABEL(strcmp_exitz)
- cmp $1, %rdx
+ cmp $1, %RDX_LP
je LABEL(Byte0)
- mov %rdx, %r11
+ mov %RDX_LP, %R11_LP
#endif
mov %esi, %ecx
mov %edi, %eax
#ifdef AS_STRNLEN
/* Do not read anything when n==0. */
- test %rsi, %rsi
+ test %RSI_LP, %RSI_LP
jne L(n_nonzero)
xor %rax, %rax
ret
L(n_nonzero):
# ifdef AS_WCSLEN
- shlq $2, %rsi
+ shl $2, %RSI_LP
# endif
/* Initialize long lived registers. */
- add %rdi, %rsi
- mov %rsi, %r10
- and $-64, %r10
- mov %rsi, %r11
+ add %RDI_LP, %RSI_LP
+ mov %RSI_LP, %R10_LP
+ and $-64, %R10_LP
+ mov %RSI_LP, %R11_LP
#endif
pxor %xmm0, %xmm0
# 64-bit llround. Add -fno-builtin-lround to silence the compiler.
CFLAGS-s_llround.c += -fno-builtin-lround
endif
+
+ifeq ($(subdir),string)
+tests += tst-size_t-memchr tst-size_t-memcmp tst-size_t-memcpy \
+ tst-size_t-memrchr tst-size_t-memset tst-size_t-strncasecmp \
+ tst-size_t-strncmp tst-size_t-strncpy tst-size_t-strnlen \
+ tst-size_t-memcmp-2
+endif
+
+ifeq ($(subdir),wcsmbs)
+tests += tst-size_t-wmemchr tst-size_t-wmemcmp tst-size_t-wmemset \
+ tst-size_t-wcsncmp tst-size_t-wcsnlen
+endif
--- /dev/null
+/* Test string/memory functions with size_t in the lower 32 bits of
+ 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_MAIN
+#include <string/test-string.h>
+
+/* On x32, parameter_t may be passed in a 64-bit register with the LEN
+ field in the lower 32 bits. When the LEN field of 64-bit register
+ is passed to string/memory function as the size_t parameter, only
+ the lower 32 bits can be used. */
+typedef struct
+{
+ union
+ {
+ size_t len;
+ void (*fn) (void);
+ };
+ void *p;
+} parameter_t;
--- /dev/null
+/* Test memchr with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifndef WIDE
+# define TEST_NAME "memchr"
+#else
+# define TEST_NAME "wmemchr"
+#endif /* WIDE */
+#include "test-size_t.h"
+
+#ifndef WIDE
+# define MEMCHR memchr
+# define CHAR char
+# define UCHAR unsigned char
+#else
+# include <wchar.h>
+# define MEMCHR wmemchr
+# define CHAR wchar_t
+# define UCHAR wchar_t
+#endif /* WIDE */
+
+IMPL (MEMCHR, 1)
+
+typedef CHAR * (*proto_t) (const CHAR*, int, size_t);
+
+static CHAR *
+__attribute__ ((noinline, noclone))
+do_memchr (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, (uintptr_t) b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t src = { { page_size / sizeof (CHAR) }, buf2 };
+ parameter_t c = { { 0 }, (void *) (uintptr_t) 0x12 };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ c.fn = impl->fn;
+ CHAR *res = do_memchr (src, c);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %p != NULL",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test memcmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_MAIN
+#ifdef WIDE
+# define TEST_NAME "wmemcmp"
+#else
+# define TEST_NAME "memcmp"
+#endif
+
+#include "test-size_t.h"
+
+#ifdef WIDE
+# include <inttypes.h>
+# include <wchar.h>
+
+# define MEMCMP wmemcmp
+# define CHAR wchar_t
+#else
+# define MEMCMP memcmp
+# define CHAR char
+#endif
+
+IMPL (MEMCMP, 1)
+
+typedef int (*proto_t) (const CHAR *, const CHAR *, size_t);
+
+static int
+__attribute__ ((noinline, noclone))
+do_memcmp (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t dest = { { page_size / sizeof (CHAR) }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ memcpy (buf1, buf2, page_size);
+
+ CHAR *p = (CHAR *) buf1;
+ p[page_size / sizeof (CHAR) - 1] = (CHAR) 1;
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ int res = do_memcmp (dest, src);
+ if (res >= 0)
+ {
+ error (0, 0, "Wrong result in function %s: %i >= 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test memcmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_MAIN
+#ifdef WIDE
+# define TEST_NAME "wmemcmp"
+#else
+# define TEST_NAME "memcmp"
+#endif
+
+#include "test-size_t.h"
+
+#ifdef WIDE
+# include <inttypes.h>
+# include <wchar.h>
+
+# define MEMCMP wmemcmp
+# define CHAR wchar_t
+#else
+# define MEMCMP memcmp
+# define CHAR char
+#endif
+
+IMPL (MEMCMP, 1)
+
+typedef int (*proto_t) (const CHAR *, const CHAR *, size_t);
+
+static int
+__attribute__ ((noinline, noclone))
+do_memcmp (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t dest = { { page_size / sizeof (CHAR) }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ memcpy (buf1, buf2, page_size);
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ int res = do_memcmp (dest, src);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %i != 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test memcpy with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_NAME "memcpy"
+#include "test-size_t.h"
+
+IMPL (memcpy, 1)
+
+typedef void *(*proto_t) (void *, const void *, size_t);
+
+static void *
+__attribute__ ((noinline, noclone))
+do_memcpy (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t dest = { { page_size }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ do_memcpy (dest, src);
+ int res = memcmp (dest.p, src.p, dest.len);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %i != 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test memrchr with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_NAME "memrchr"
+#include "test-size_t.h"
+
+IMPL (memchr, 1)
+
+typedef void * (*proto_t) (const void *, int, size_t);
+
+static void *
+__attribute__ ((noinline, noclone))
+do_memrchr (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, (uintptr_t) b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t src = { { page_size }, buf2 };
+ parameter_t c = { { 0 }, (void *) (uintptr_t) 0x12 };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ c.fn = impl->fn;
+ void * res = do_memrchr (src, c);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %p != NULL",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test memset with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifdef WIDE
+# define TEST_NAME "wmemset"
+#else
+# define TEST_NAME "memset"
+#endif /* WIDE */
+
+#include "test-size_t.h"
+
+#ifdef WIDE
+# include <wchar.h>
+# define MEMSET wmemset
+# define CHAR wchar_t
+#else
+# define MEMSET memset
+# define CHAR char
+#endif /* WIDE */
+
+IMPL (MEMSET, 1)
+
+typedef CHAR *(*proto_t) (CHAR *, int, size_t);
+
+static void *
+__attribute__ ((noinline, noclone))
+do_memset (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, (uintptr_t) b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ CHAR ch = 0x23;
+ parameter_t src = { { page_size / sizeof (CHAR) }, buf2 };
+ parameter_t c = { { 0 }, (void *) (uintptr_t) ch };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ c.fn = impl->fn;
+ CHAR *p = (CHAR *) do_memset (src, c);
+ size_t i;
+ for (i = 0; i < src.len; i++)
+ if (p[i] != ch)
+ {
+ error (0, 0, "Wrong result in function %s", impl->name);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test strncaecmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_NAME "strncasecmp"
+#include "test-size_t.h"
+
+IMPL (strncasecmp, 1)
+
+typedef int (*proto_t) (const char *, const char *, size_t);
+
+static int
+__attribute__ ((noinline, noclone))
+do_strncasecmp (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t dest = { { page_size }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ strncpy ((char *) buf1, (const char *) buf2, page_size);
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ int res = do_strncasecmp (dest, src);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %i != 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test strncmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifdef WIDE
+# define TEST_NAME "wcsncmp"
+#else
+# define TEST_NAME "strncmp"
+#endif
+
+#include "test-size_t.h"
+
+#ifdef WIDE
+# include <wchar.h>
+
+# define STRNCMP wcsncmp
+# define STRNCPY wcsncpy
+# define CHAR wchar_t
+#else
+# define STRNCMP strncmp
+# define STRNCPY strncpy
+# define CHAR char
+#endif
+
+IMPL (STRNCMP, 1)
+
+typedef int (*proto_t) (const CHAR *, const CHAR *, size_t);
+
+
+static int
+__attribute__ ((noinline, noclone))
+do_strncmp (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ size_t size = page_size / sizeof (CHAR);
+ parameter_t dest = { { size }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ STRNCPY ((CHAR *) buf1, (const CHAR *) buf2, size);
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ int res = do_strncmp (dest, src);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %i != 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test strncpy with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define TEST_NAME "strncpy"
+#include "test-size_t.h"
+
+IMPL (strncpy, 1)
+
+typedef char *(*proto_t) (char *, const char*, size_t);
+
+static void *
+__attribute__ ((noinline, noclone))
+do_strncpy (parameter_t a, parameter_t b)
+{
+ return CALL (&b, a.p, b.p, a.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ parameter_t dest = { { page_size }, buf1 };
+ parameter_t src = { { 0 }, buf2 };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ do_strncpy (dest, src);
+ int res = strncmp (dest.p, src.p, dest.len);
+ if (res)
+ {
+ error (0, 0, "Wrong result in function %s: %i != 0",
+ impl->name, res);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test strnlen with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#ifdef WIDE
+# define TEST_NAME "wcsnlen"
+#else
+# define TEST_NAME "strnlen"
+#endif /* WIDE */
+
+#include "test-size_t.h"
+
+#ifdef WIDE
+# include <wchar.h>
+# define STRNLEN wcsnlen
+# define CHAR wchar_t
+#else
+# define STRNLEN strnlen
+# define CHAR char
+#endif /* WIDE */
+
+IMPL (STRNLEN, 1)
+
+typedef size_t (*proto_t) (const CHAR *, size_t);
+
+static size_t
+__attribute__ ((noinline, noclone))
+do_strnlen (parameter_t a, parameter_t b)
+{
+ return CALL (&a, a.p, b.len);
+}
+
+static int
+test_main (void)
+{
+ test_init ();
+
+ size_t size = page_size / sizeof (CHAR);
+ parameter_t src = { { 0 }, buf2 };
+ parameter_t c = { { size }, (void *) (uintptr_t) 'a' };
+
+ int ret = 0;
+ FOR_EACH_IMPL (impl, 0)
+ {
+ src.fn = impl->fn;
+ size_t res = do_strnlen (src, c);
+ if (res != size)
+ {
+ error (0, 0, "Wrong result in function %s: 0x%x != 0x%x",
+ impl->name, res, size);
+ ret = 1;
+ }
+ }
+
+ return ret ? EXIT_FAILURE : EXIT_SUCCESS;
+}
+
+#include <support/test-driver.c>
--- /dev/null
+/* Test wcsncmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define WIDE 1
+#include "tst-size_t-strncmp.c"
--- /dev/null
+/* Test wcsnlen with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define WIDE 1
+#include "tst-size_t-strnlen.c"
--- /dev/null
+/* Test wmemchr with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define WIDE 1
+#include "tst-size_t-memchr.c"
--- /dev/null
+/* Test wmemcmp with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define WIDE 1
+#include "tst-size_t-memcmp.c"
--- /dev/null
+/* Test wmemset with size_t in the lower 32 bits of 64-bit register.
+ Copyright (C) 2019 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#define WIDE 1
+#include "tst-size_t-memset.c"
/* First "register" all timezone names. */
for (i = 0; i < num_types; ++i)
- (void) __tzstring (&zone_names[types[i].idx]);
+ if (__tzstring (&zone_names[types[i].idx]) == NULL)
+ goto ret_free_transitions;
/* Find the standard and daylight time offsets used by the rule file.
We choose the offsets in the types of each flavor that are
projects / glibc.git / commitdiff