In a pathological scenario where we are trying to bunmapi a single
extent in which every other block is shared, it's possible that trying
to unmap the entire large extent in a single transaction can generate so
many EFIs that we overflow the transaction reservation.
Therefore, use a heuristic to guess at the number of blocks we can
safely unmap from a reflink file's data fork in an single transaction.
This should prevent problems such as the log head slamming into the tail
and ASSERTs that trigger because we've exceeded the transaction
reservation.
Note that since bunmapi can fail to unmap the entire range, we must also
teach the deferred unmap code to roll into a new transaction whenever we
get low on reservation.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
[hch: random edits, all bugs are my fault] Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reclaim during quotacheck can lead to deadlocks on the dquot flush
lock:
- Quotacheck populates a local delwri queue with the physical dquot
buffers.
- Quotacheck performs the xfs_qm_dqusage_adjust() bulkstat and
dirties all of the dquots.
- Reclaim kicks in and attempts to flush a dquot whose buffer is
already queud on the quotacheck queue. The flush succeeds but
queueing to the reclaim delwri queue fails as the backing buffer is
already queued. The flush unlock is now deferred to I/O completion
of the buffer from the quotacheck queue.
- The dqadjust bulkstat continues and dirties the recently flushed
dquot once again.
- Quotacheck proceeds to the xfs_qm_flush_one() walk which requires
the flush lock to update the backing buffers with the in-core
recalculated values. It deadlocks on the redirtied dquot as the
flush lock was already acquired by reclaim, but the buffer resides
on the local delwri queue which isn't submitted until the end of
quotacheck.
This is reproduced by running quotacheck on a filesystem with a
couple million inodes in low memory (512MB-1GB) situations. This is
a regression as of commit 43ff2122e6 ("xfs: on-stack delayed write
buffer lists"), which removed a trylock and buffer I/O submission
from the quotacheck dquot flush sequence.
Quotacheck first resets and collects the physical dquot buffers in a
delwri queue. Then, it traverses the filesystem inodes via bulkstat,
updates the in-core dquots, flushes the corrected dquots to the
backing buffers and finally submits the delwri queue for I/O. Since
the backing buffers are queued across the entire quotacheck
operation, dquot reclaim cannot possibly complete a dquot flush
before quotacheck completes.
Therefore, quotacheck must submit the buffer for I/O in order to
cycle the flush lock and flush the dirty in-core dquot to the
buffer. Add a delwri queue buffer push mechanism to submit an
individual buffer for I/O without losing the delwri queue status and
use it from quotacheck to avoid the deadlock. This restores
quotacheck behavior to as before the regression was introduced.
Reported-by: Martin Svec <martin.svec@zoner.cz> Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 0-day kernel test robot reports assertion failures on
!CONFIG_SMP kernels due to failed spin_is_locked() checks. As it
turns out, spin_is_locked() is hardcoded to return zero on
!CONFIG_SMP kernels and so this function cannot be relied on to
verify spinlock state in this configuration.
To avoid this problem, replace the associated asserts with lockdep
variants that do the right thing regardless of kernel configuration.
Drop the one assert that checks for an unlocked lock as there is no
suitable lockdep variant for that case. This moves the spinlock
checks from XFS debug code to lockdep, but generally provides the
same level of protection.
Reported-by: kbuild test robot <fengguang.wu@intel.com> Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently several places in xfs_find_get_desired_pgoff() handle the case
of a missing page. Make them all handled in one place after the loop has
terminated.
Signed-off-by: Jan Kara <jack@suse.cz> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Switching FS and GS is a mess, and the current code is still subtly
wrong: it assumes that "Loading a nonzero value into FS sets the
index and base", which is false on AMD CPUs if the value being
loaded is 1, 2, or 3.
(The current code came from commit 3e2b68d752c9 ("x86/asm,
sched/x86: Rewrite the FS and GS context switch code"), which made
it better but didn't fully fix it.)
Rewrite it to be much simpler and more obviously correct. This
should fix it fully on AMD CPUs and shouldn't adversely affect
performance.
Signed-off-by: Andy Lutomirski <luto@kernel.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Borislav Petkov <bpetkov@suse.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Chang Seok <chang.seok.bae@intel.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In ELF_COPY_CORE_REGS, we're copying from the current task, so
accessing thread.fsbase and thread.gsbase makes no sense. Just read
the values from the CPU registers.
In practice, the old code would have been correct most of the time
simply because thread.fsbase and thread.gsbase usually matched the
CPU registers.
Signed-off-by: Andy Lutomirski <luto@kernel.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Borislav Petkov <bpetkov@suse.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Chang Seok <chang.seok.bae@intel.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
execve used to leak FSBASE and GSBASE on AMD CPUs. Fix it.
The security impact of this bug is small but not quite zero -- it
could weaken ASLR when a privileged task execs a less privileged
program, but only if program changed bitness across the exec, or the
child binary was highly unusual or actively malicious. A child
program that was compromised after the exec would not have access to
the leaked base.
Signed-off-by: Andy Lutomirski <luto@kernel.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Borislav Petkov <bpetkov@suse.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Chang Seok <chang.seok.bae@intel.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we set CP_ERROR_FLAG in roll-forward error, f2fs is no longer to proceed
any IOs due to f2fs_cp_error(). But, for example, if some stale data is involved
on roll-forward process, we're able to get -ENOENT, getting fs stuck.
If we get any error, let fill_super set SBI_NEED_FSCK and try to recover back
to stable point.
Commit fb586f25300f ("sctp: delay calls to sk_data_ready() as much as
possible") minimized the number of wake ups that are triggered in case
the association receives a packet with multiple data chunks on it and/or
when io_events are enabled and then commit 0970f5b36659 ("sctp: signal
sk_data_ready earlier on data chunks reception") moved the wake up to as
soon as possible. It thus relies on the state machine running later to
clean the flag that the event was already generated.
The issue is that there are 2 call paths that calls
sctp_ulpq_tail_event() outside of the state machine, causing the flag to
linger and possibly omitting a needed wake up in the sequence.
One of the call paths is when enabling SCTP_SENDER_DRY_EVENTS via
setsockopt(SCTP_EVENTS), as noticed by Harald Welte. The other is when
partial reliability triggers removal of chunks from the send queue when
the application calls sendmsg().
This commit fixes it by not setting the flag in case the socket is not
owned by the user, as it won't be cleaned later. This works for
user-initiated calls and also for rx path processing.
Fixes: fb586f25300f ("sctp: delay calls to sk_data_ready() as much as possible") Reported-by: Harald Welte <laforge@gnumonks.org> Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
IPv6 FIB should use FIB6_TABLE_HASHSZ, not FIB_TABLE_HASHSZ.
Fixes: ba1cc08d9488 ("ipv6: fix memory leak with multiple tables during netns destruction") Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fib6_net_exit only frees the main and local tables. If another table was
created with fib6_alloc_table, we leak it when the netns is destroyed.
Fix this in the same way ip_fib_net_exit cleans up tables, by walking
through the whole hashtable of fib6_table's. We can get rid of the
special cases for local and main, since they're also part of the
hashtable.
Reproducer:
ip netns add x
ip -net x -6 rule add from 6003:1::/64 table 100
ip netns del x
Reported-by: Jianlin Shi <jishi@redhat.com> Fixes: 58f09b78b730 ("[NETNS][IPV6] ip6_fib - make it per network namespace") Signed-off-by: Sabrina Dubroca <sd@queasysnail.net> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now when probessing ICMPV6_PKT_TOOBIG, ip6gre_err only subtracts the
offset of gre header from mtu info. The expected mtu of gre device
should also subtract gre header. Otherwise, the next packets still
can't be sent out.
Jianlin found this issue when using the topo:
client(ip6gre)<---->(nic1)route(nic2)<----->(ip6gre)server
and reducing nic2's mtu, then both tcp and sctp's performance with
big size data became 0.
This patch is to fix it by also subtracting grehdr (tun->tun_hlen)
from mtu info when updating gre device's mtu in ip6gre_err(). It
also needs to subtract ETH_HLEN if gre dev'type is ARPHRD_ETHER.
Reported-by: Jianlin Shi <jishi@redhat.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We check tx avail through vhost_enable_notify() in the past which is
wrong since it only checks whether or not guest has filled more
available buffer since last avail idx synchronization which was just
done by vhost_vq_avail_empty() before. What we really want is checking
pending buffers in the avail ring. Fix this by calling
vhost_vq_avail_empty() instead.
This issue could be noticed by doing netperf TCP_RR benchmark as
client from guest (but not host). With this fix, TCP_RR from guest to
localhost restores from 1375.91 trans per sec to 55235.28 trans per
sec on my laptop (Intel(R) Core(TM) i7-5600U CPU @ 2.60GHz).
Fixes: 030881372460 ("vhost_net: basic polling support") Signed-off-by: Jason Wang <jasowang@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The wrong register is checked for the Tx flow control bit,
it should have been maccfg1 not maccfg2.
This went unnoticed for so long probably because the impact is
hardly visible, not to mention the tangled code from adjust_link().
First, link flow control (i.e. handling of Rx/Tx link level pause frames)
is disabled by default (needs to be enabled via 'ethtool -A').
Secondly, maccfg2 always returns 0 for tx_flow_oldval (except for a few
old boards), which results in Tx flow control remaining always on
once activated.
Fixes: 45b679c9a3ccd9e34f28e6ec677b812a860eb8eb ("gianfar: Implement PAUSE frame generation support") Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After reverting commit 6d7b857d541e ("net: use lib/percpu_counter API
for fragmentation mem accounting") then here is no need for this
fix-up patch. As percpu_counter is no longer used, it cannot
memory leak it any-longer.
Fixes: 6d7b857d541e ("net: use lib/percpu_counter API for fragmentation mem accounting") Fixes: 1d6119baf061 ("net: fix percpu memory leaks") Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a bug in fragmentation codes use of the percpu_counter API,
that can cause issues on systems with many CPUs.
The frag_mem_limit() just reads the global counter (fbc->count),
without considering other CPUs can have upto batch size (130K) that
haven't been subtracted yet. Due to the 3MBytes lower thresh limit,
this become dangerous at >=24 CPUs (3*1024*1024/130000=24).
The correct API usage would be to use __percpu_counter_compare() which
does the right thing, and takes into account the number of (online)
CPUs and batch size, to account for this and call __percpu_counter_sum()
when needed.
We choose to revert the use of the lib/percpu_counter API for frag
memory accounting for several reasons:
1) On systems with CPUs > 24, the heavier fully locked
__percpu_counter_sum() is always invoked, which will be more
expensive than the atomic_t that is reverted to.
Given systems with more than 24 CPUs are becoming common this doesn't
seem like a good option. To mitigate this, the batch size could be
decreased and thresh be increased.
2) The add_frag_mem_limit+sub_frag_mem_limit pairs happen on the RX
CPU, before SKBs are pushed into sockets on remote CPUs. Given
NICs can only hash on L2 part of the IP-header, the NIC-RXq's will
likely be limited. Thus, a fair chance that atomic add+dec happen
on the same CPU.
Revert note that commit 1d6119baf061 ("net: fix percpu memory leaks")
removed init_frag_mem_limit() and instead use inet_frags_init_net().
After this revert, inet_frags_uninit_net() becomes empty.
Fixes: 6d7b857d541e ("net: use lib/percpu_counter API for fragmentation mem accounting") Fixes: 1d6119baf061 ("net: fix percpu memory leaks") Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Acked-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 6bc506b4fb06 ("bridge: switchdev: Add forward mark support for
stacked devices") added the 'offload_fwd_mark' bit to the skb in order
to allow drivers to indicate to the bridge driver that they already
forwarded the packet in L2.
In case the bit is set, before transmitting the packet from each port,
the port's mark is compared with the mark stored in the skb's control
block. If both marks are equal, we know the packet arrived from a switch
device that already forwarded the packet and it's not re-transmitted.
However, if the packet is transmitted from the bridge device itself
(e.g., br0), we should clear the 'offload_fwd_mark' bit as the mark
stored in the skb's control block isn't valid.
This scenario can happen in rare cases where a packet was trapped during
L3 forwarding and forwarded by the kernel to a bridge device.
Fixes: 6bc506b4fb06 ("bridge: switchdev: Add forward mark support for stacked devices") Signed-off-by: Ido Schimmel <idosch@mellanox.com> Reported-by: Yotam Gigi <yotamg@mellanox.com> Tested-by: Yotam Gigi <yotamg@mellanox.com> Reviewed-by: Jiri Pirko <jiri@mellanox.com> Acked-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The mlxsw driver relies on NETDEV_CHANGEUPPER events to configure the
device in case a port is enslaved to a master netdev such as bridge or
bond.
Since the driver ignores events unrelated to its ports and their
uppers, it's possible to engineer situations in which the device's data
path differs from the kernel's.
One example to such a situation is when a port is enslaved to a bond
that is already enslaved to a bridge. When the bond was enslaved the
driver ignored the event - as the bond wasn't one of its uppers - and
therefore a bridge port instance isn't created in the device.
Until such configurations are supported forbid them by checking that the
upper device doesn't have uppers of its own.
Fixes: 0d65fc13042f ("mlxsw: spectrum: Implement LAG port join/leave") Signed-off-by: Ido Schimmel <idosch@mellanox.com> Reported-by: Nogah Frankel <nogahf@mellanox.com> Tested-by: Nogah Frankel <nogahf@mellanox.com> Signed-off-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When tcp_disconnect() is called, inet_csk_delack_init() sets
icsk->icsk_ack.rcv_mss to 0.
This could potentially cause tcp_recvmsg() => tcp_cleanup_rbuf() =>
__tcp_select_window() call path to have division by 0 issue.
So this patch initializes rcv_mss to TCP_MIN_MSS instead of 0.
Reported-by: Andrey Konovalov <andreyknvl@google.com> Signed-off-by: Wei Wang <weiwan@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This reverts commit 7ad813f208533cebfcc32d3d7474dc1677d1b09a ("net: phy:
Correctly process PHY_HALTED in phy_stop_machine()") because it is
creating the possibility for a NULL pointer dereference.
David Daney provide the following call trace and diagram of events:
The original motivation for this change originated from Marc Gonzales
indicating that his network driver did not have its adjust_link callback
executing with phydev->link = 0 while he was expecting it.
PHYLIB has never made any such guarantees ever because phy_stop() merely just
tells the workqueue to move into PHY_HALTED state which will happen
asynchronously.
Reported-by: Geert Uytterhoeven <geert+renesas@glider.be> Reported-by: David Daney <ddaney.cavm@gmail.com> Fixes: 7ad813f20853 ("net: phy: Correctly process PHY_HALTED in phy_stop_machine()") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
syzkaller had no problem to trigger a deadlock, attaching a KCM socket
to another one (or itself). (original syzkaller report was a very
confusing lockdep splat during a sendmsg())
It seems KCM claims to only support TCP, but no enforcement is done,
so we might need to add additional checks.
Fixes: ab7ac4eb9832 ("kcm: Kernel Connection Multiplexor module") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Acked-by: Tom Herbert <tom@quantonium.net> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
... which may happen with certain values of tp_reserve and maclen.
Fixes: 58d19b19cd99 ("packet: vnet_hdr support for tpacket_rcv") Signed-off-by: Benjamin Poirier <bpoirier@suse.com> Cc: Willem de Bruijn <willemb@google.com> Acked-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Passing commands for logging to t4_record_mbox() with size
MBOX_LEN, when the actual command size is actually smaller,
causes out-of-bounds stack accesses in t4_record_mbox() while
copying command words here:
for (i = 0; i < size / 8; i++)
entry->cmd[i] = be64_to_cpu(cmd[i]);
Up to 48 bytes from the stack are then leaked to debugfs.
When we call t4_record_mbox() to log a command reply, a MBOX_LEN
size can be used though, as get_mbox_rpl() will fill cmd_rpl up
completely.
Fixes: 7f080c3f2ff0 ("cxgb4: Add support to enable logging of firmware mailbox commands") Signed-off-by: Stefano Brivio <sbrivio@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a deadlock possible when canceling the link status
delayed work queue. The removal process is run with RTNL held,
and the link status callback is acquring RTNL.
Resolve the issue by using trylock and rescheduling.
If cancel is in process, that block it from happening.
Fixes: 122a5f6410f4 ("staging: hv: use delayed_work for netvsc_send_garp()") Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc-8.0.0 (snapshot) points out that we copy a variable-length string
into a fixed length field using memcpy() with the destination length,
and that ends up copying whatever follows the string:
inlined from 'ql_core_dump' at drivers/net/ethernet/qlogic/qlge/qlge_dbg.c:1106:2:
drivers/net/ethernet/qlogic/qlge/qlge_dbg.c:708:2: error: 'memcpy' reading 15 bytes from a region of size 14 [-Werror=stringop-overflow=]
memcpy(seg_hdr->description, desc, (sizeof(seg_hdr->description)) - 1);
Changing it to use strncpy() will instead zero-pad the destination,
which seems to be the right thing to do here.
The bug is probably harmless, but it seems like a good idea to address
it in stable kernels as well, if only for the purpose of building with
gcc-8 without warnings.
Fixes: a61f80261306 ("qlge: Add ethtool register dump function.") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
inet_diag_msg_sctp{,l}addr_fill() and sctp_get_sctp_info() copy
sizeof(sockaddr_storage) bytes to fill in sockaddr structs used
to export diagnostic information to userspace.
However, the memory allocated to store sockaddr information is
smaller than that and depends on the address family, so we leak
up to 100 uninitialized bytes to userspace. Just use the size of
the source structs instead, in all the three cases this is what
userspace expects. Zero out the remaining memory.
Unused bytes (i.e. when IPv4 addresses are used) in source
structs sctp_sockaddr_entry and sctp_transport are already
cleared by sctp_add_bind_addr() and sctp_transport_new(),
respectively.
Noticed while testing KASAN-enabled kernel with 'ss':
Junote Cai reported that he was not able to get a DSA setup involving the
Freescale DPAA/FMAN driver to work and narrowed it down to
of_find_net_device_by_node(). This function requires the network device's
device reference to be correctly set which is the case here, though we have
lost any device_node association there.
The problem is that dpaa_eth_add_device() allocates a "dpaa-ethernet" platform
device, and later on dpaa_eth_probe() is called but SET_NETDEV_DEV() won't be
propagating &pdev->dev.of_node properly. Fix this by inherenting both the parent
device and the of_node when dpaa_eth_add_device() creates the platform device.
Fixes: 3933961682a3 ("fsl/fman: Add FMan MAC driver") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When peeking, if a bad csum is discovered, the skb is unlinked from
the queue with __sk_queue_drop_skb and the peek operation restarted.
__sk_queue_drop_skb only drops packets that match the queue head.
This fails if the skb was found after the head, using SO_PEEK_OFF
socket option. This causes an infinite loop.
We MUST drop this problematic skb, and we can simply check if skb was
already removed by another thread, by looking at skb->next :
This pointer is set to NULL by the __skb_unlink() operation, that might
have happened only under the spinlock protection.
Many thanks to syzkaller team (and particularly Dmitry Vyukov who
provided us nice C reproducers exhibiting the lockup) and Willem de
Bruijn who provided first version for this patch and a test program.
Fixes: 627d2d6b5500 ("udp: enable MSG_PEEK at non-zero offset") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Cc: Willem de Bruijn <willemb@google.com> Acked-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit c5cff8561d2d adds rcu grace period before freeing fib6_node. This
generates a new sparse warning on rt->rt6i_node related code:
net/ipv6/route.c:1394:30: error: incompatible types in comparison
expression (different address spaces)
./include/net/ip6_fib.h:187:14: error: incompatible types in comparison
expression (different address spaces)
This commit adds "__rcu" tag for rt6i_node and makes sure corresponding
rcu API is used for it.
After this fix, sparse no longer generates the above warning.
Fixes: c5cff8561d2d ("ipv6: add rcu grace period before freeing fib6_node") Signed-off-by: Wei Wang <weiwan@google.com> Acked-by: Eric Dumazet <edumazet@google.com> Acked-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We currently keep rt->rt6i_node pointing to the fib6_node for the route.
And some functions make use of this pointer to dereference the fib6_node
from rt structure, e.g. rt6_check(). However, as there is neither
refcount nor rcu taken when dereferencing rt->rt6i_node, it could
potentially cause crashes as rt->rt6i_node could be set to NULL by other
CPUs when doing a route deletion.
This patch introduces an rcu grace period before freeing fib6_node and
makes sure the functions that dereference it takes rcu_read_lock().
Note: there is no "Fixes" tag because this bug was there in a very
early stage.
Signed-off-by: Wei Wang <weiwan@google.com> Acked-by: Eric Dumazet <edumazet@google.com> Acked-by: Martin KaFai Lau <kafai@fb.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A packet length of exactly IPV6_MAXPLEN is allowed, we should
refuse parsing options only if the size is 64KiB or more.
While at it, remove one extra variable and one assignment which
were also introduced by the commit that introduced the size
check. Checking the sum 'offset + len' and only later adding
'len' to 'offset' doesn't provide any advantage over directly
summing to 'offset' and checking it.
Fixes: 6399f1fae4ec ("ipv6: avoid overflow of offset in ip6_find_1stfragopt") Signed-off-by: Stefano Brivio <sbrivio@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If using a kernel with CONFIG_XFS_RT=y and we set the RHINHERIT flag on
a directory in a filesystem that does not have a realtime device and
create a new file in that directory, it gets marked as a real time file.
When data is written and a fsync is issued, the filesystem attempts to
flush a non-existent rt device during the fsync process.
This results in a crash dereferencing a null buftarg pointer in
xfs_blkdev_issue_flush():
Setting RT inode flags does not require special privileges so any
unprivileged user can cause this oops to occur. To reproduce, confirm
kernel is compiled with CONFIG_XFS_RT=y and run:
Since commit 18290650b1c8 ("NFS: Move buffered I/O locking into
nfs_file_write()") nfs_file_write() has not flushed the correct byte
range during synchronous writes. generic_write_sync() expects that
iocb->ki_pos points to the right edge of the range rather than the
left edge.
To replicate the problem, open a file with O_DSYNC, have the client
write at increasing offsets, and then print the successful offsets.
Block port 2049 partway through that sequence, and observe that the
client application indicates successful writes in advance of what the
server received.
When there's a fatal signal pending, arm's do_page_fault()
implementation returns 0. The intent is that we'll return to the
faulting userspace instruction, delivering the signal on the way.
However, if we take a fatal signal during fixing up a uaccess, this
results in a return to the faulting kernel instruction, which will be
instantly retried, resulting in the same fault being taken forever. As
the task never reaches userspace, the signal is not delivered, and the
task is left unkillable. While the task is stuck in this state, it can
inhibit the forward progress of the system.
To avoid this, we must ensure that when a fatal signal is pending, we
apply any necessary fixup for a faulting kernel instruction. Thus we
will return to an error path, and it is up to that code to make forward
progress towards delivering the fatal signal.
Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Steve Capper <steve.capper@arm.com> Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The GIC-500 integrated in the Armada-37xx SoCs is compliant with
the GICv3 architecture, and thus provides a maintenance interrupt
that is required for hypervisors to function correctly.
With the interrupt provided in the DT, KVM now works as it should.
Tested on an Espressobin system.
Fixes: adbc3695d9e4 ("arm64: dts: add the Marvell Armada 3700 family and a development board") Signed-off-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It is because cxt.lwsa and cxt.lrsa don't get freed in module_exit, so free
them in lock_torture_cleanup() and free writer_tasks if reader_tasks is
failed at memory allocation.
Signed-off-by: Yang Shi <yang.shi@linaro.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org> Cc: 石洋 <yang.s@alibaba-inc.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Seen while reading the code, in handle_mm_fault(), in the case
arch_vma_access_permitted() is failing the call to
mem_cgroup_oom_disable() is not made.
To fix that, move the call to mem_cgroup_oom_enable() after calling
arch_vma_access_permitted() as it should not have entered the memcg OOM.
Several distributions mount the "proper root" as ro during initrd and
then remount it as rw before pivot_root(2). Thus, if a rescan had been
aborted by a previous shutdown, the rescan would never be resumed.
This issue would manifest itself as several btrfs ioctl(2)s causing the
entire machine to hang when btrfs_qgroup_wait_for_completion was hit
(due to the fs_info->qgroup_rescan_running flag being set but the rescan
itself not being resumed). Notably, Docker's btrfs storage driver makes
regular use of BTRFS_QUOTA_CTL_DISABLE and BTRFS_IOC_QUOTA_RESCAN_WAIT
(causing this problem to be manifested on boot for some machines).
Cc: Jeff Mahoney <jeffm@suse.com> Fixes: b382a324b60f ("Btrfs: fix qgroup rescan resume on mount") Signed-off-by: Aleksa Sarai <asarai@suse.de> Reviewed-by: Nikolay Borisov <nborisov@suse.com> Tested-by: Nikolay Borisov <nborisov@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The default host NQN, which is generated based on the host's UUID,
does not follow the UUID-based NQN format laid out in the NVMe 1.3
specification. Remove the "NVMf:" portion of the NQN to match the spec.
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com> Reviewed-by: Max Gurtovoy <maxg@mellanox.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The READ_START_VALID should be enabled for sending PAGE_READ
command. READ_STOP_VALID should be cleared since normal NAND
page read does not require READ_STOP command.
Commit a894cf6c5a82 ("mtd: nand: mxc: switch to mtd_ooblayout_ops")
introduced a bug in the OOB layout description. Even if the driver claims
that 3 ECC bytes are reserved to protect 512 bytes of data, it's actually
5 ECC bytes to protect 512+6 bytes of data (some OOB bytes are also
protected using extra ECC bytes).
Fix the mxc_v1_ooblayout_{free,ecc}() functions to reflect this behavior.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com> Fixes: a894cf6c5a82 ("mtd: nand: mxc: switch to mtd_ooblayout_ops") Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In chasing down a previous issue with EDID probing from calling
drm_helper_hpd_irq_event() from irq context, Laurent noticed
that the DRM documentation suggests that
drm_kms_helper_hotplug_event() should be used instead.
Thus this patch replaces drm_helper_hpd_irq_event() with
drm_kms_helper_hotplug_event(), which requires we update the
connector.status entry and only call _hotplug_event() when the
status changes.
I was recently seeing issues with EDID probing, where
the logic to wait for the EDID read bit to be set by the
IRQ wasn't happening and the code would time out and fail.
Digging deeper, I found this was due to the fact that
IRQs were disabled as we were running in IRQ context from
the HPD signal.
Thus this patch changes the logic to handle the HPD signal
via a work_struct so we can be out of irq context.
With this patch, the EDID probing on hotplug does not time
out.
Commit 1bc0eb044615 ("scsi: sg: protect accesses to 'reserved' page
array") adds needed concurrency protection for the "reserve" buffer.
Some checks that are initially made outside the lock are replicated once
the lock is taken to ensure the checks and resulting decisions are made
using consistent state.
The check that a request with flag SG_FLAG_MMAP_IO set fits in the
reserve buffer also needs to be performed again under the lock to ensure
the reserve buffer length compared against matches the value in effect
when the request is linked to the reserve buffer. An -ENOMEM should be
returned in this case, instead of switching over to an indirect buffer
as for non-MMAP_IO requests.
Signed-off-by: Todd Poynor <toddpoynor@google.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Take f_mutex around mmap() processing to protect against races with the
SG_SET_RESERVED_SIZE ioctl. Ensure the reserve buffer length remains
consistent during the mapping operation, and set the "mmap called" flag
to prevent further changes to the reserved buffer size as an atomic
operation with the mapping.
[mkp: fixed whitespace]
Signed-off-by: Todd Poynor <toddpoynor@google.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Several legacy devices such as Geode-based Cisco ASA appliances
and DB800 development board do possess CS5536 IDE controller
with different PCI id than existing one. Using pata_generic is
not always feasible as at least DB800 requires MSR quirk from
pata_cs5536 to be used with vendor firmware.
Commit 0a94efb5acbb ("workqueue: implicit ordered attribute should be
overridable") introduced a __WQ_ORDERED_EXPLICIT flag but gave it the
same value as __WQ_LEGACY. I don't believe these were intended to
mean the same thing, so renumber __WQ_ORDERED_EXPLICIT.
Fixes: 0a94efb5acbb ("workqueue: implicit ordered attribute should be ...") Signed-off-by: Ben Hutchings <ben@decadent.org.uk> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It appears that MSI does not work on either G5 PPC nor on a E5500-based
platform, where other hardware is reported to work fine with MSI.
Both tests were conducted with NV4x hardware, so perhaps other (or even
this) hardware can be made to work. It's still possible to force-enable
with config=NvMSI=1 on load.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Signed-off-by: Ben Skeggs <bskeggs@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Right now there is a potential hang situation for postcopy migrations,
if the guest is enabling storage keys on the target system during the
postcopy process.
For storage key virtualization, we have to forbid the empty zero page as
the storage key is a property of the physical page frame. As we enable
storage key handling lazily we then drop all mappings for empty zero
pages for lazy refaulting later on.
This does not work with the postcopy migration, which relies on the
empty zero page never triggering a fault again in the future. The reason
is that postcopy migration will simply read a page on the target system
if that page is a known zero page to fault in an empty zero page. At
the same time postcopy remembers that this page was already transferred
- so any future userfault on that page will NOT be retransmitted again
to avoid races.
If now the guest enters the storage key mode while in postcopy, we will
break this assumption of postcopy.
The solution is to disable the empty zero page for KVM guests early on
and not during storage key enablement. With this change, the postcopy
migration process is guaranteed to start after no zero pages are left.
As guest pages are very likely not empty zero pages anyway the memory
overhead is also pretty small.
While at it this also adds proper page table locking to the zero page
removal.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Janosch Frank <frankja@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds the resources and DMI ID's for the MEN SC31,
which uses a different address region to map the LPC bus than
the one used for the existing SC24.
Signed-off-by: Michael Moese <michael.moese@men.de>
[jth add stable tag] Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mwifiex records information about various channels as it receives scan
information. It does this by appending to a buffer that was sized
to the max number of supported channels on any band, but there are
numerous problems:
(a) scans can return info from more than one band (e.g., both 2.4 and 5
GHz), so the determined "max" is not large enough
(b) some firmware appears to return multiple results for a given
channel, so the max *really* isn't large enough
(c) there is no bounds checking when stashing these stats, so problems
(a) and (b) can easily lead to buffer overflows
Let's patch this by setting a slightly-more-correct max (that accounts
for a combination of both 2.4G and 5G bands) and adding a bounds check
when writing to our statistics buffer.
Due to problem (b), we still might not properly report all known survey
information (e.g., with "iw <dev> survey dump"), since duplicate results
(or otherwise "larger than expected" results) will cause some
truncation. But that's a problem for a future bugfix.
(And because of this known deficiency, only log the excess at the WARN
level, since that isn't visible by default in this driver and would
otherwise be a bit too noisy.)
Fixes: bf35443314ac ("mwifiex: channel statistics support for mwifiex") Cc: Avinash Patil <patila@marvell.com> Cc: Xinming Hu <huxm@marvell.com> Signed-off-by: Brian Norris <briannorris@chromium.org> Reviewed-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Reviewed-by: Ganapathi Bhat <gbhat@marvell.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Can be reproduced when running dlm_controld (tested on 4.4.x, 4.12.4):
# seq 1 100 | xargs -P0 -n1 dlm_tool join
# seq 1 100 | xargs -P0 -n1 dlm_tool leave
misc_register fails due to duplicate sysfs entry, which causes
dlm_device_register to free ls->ls_device.name.
In dlm_device_deregister the name was freed again, causing memory
corruption.
According to the comment in dlm_device_deregister the name should've been
set to NULL when registration fails,
so this patch does that.
Trackpoint buttons detection fails on ThinkPad 570 and 470 series,
this makes the middle button of the trackpoint to not being recogized.
As I don't believe there is any trackpoint with less than 3 buttons this
patch just assumes three buttons when the extended button information
read fails.
Signed-off-by: Oscar Campos <oscar.campos@member.fsf.org> Acked-by: Peter Hutterer <peter.hutterer@who-t.net> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Aaron Ma <aaron.ma@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The rx ring buffers are added to a hash table if
firmware support full rx reorder. If the full rx
reorder support flag is not set before allocating
the rx ring buffers, none of the buffers are added
to the hash table.
There is a race condition between rx ring refill and
rx buffer replenish from napi poll. The interrupts are
enabled in hif start, before the rx ring is refilled during init.
We replenish buffers from napi poll due to the interrupts which
get enabled after hif start. Hence before the entire rx ring is
refilled during the init, the napi poll replenishes a few buffers
in steps of 100 buffers per attempt. During this rx ring replenish
from napi poll, the rx reorder flag has not been set due to which
the replenished buffers are not added to the hash table
Set the rx full reorder support flag before we allocate
the rx ring buffer to avoid the memory leak.
Signed-off-by: Rakesh Pillai <pillair@qti.qualcomm.com> Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com> Cc: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This driver assumes that the device is operating in the continuous
conversion mode which performs the conversion continuously. So this driver
inserts a wait time before reading the conversion register if the
configuration is changed from a previous request.
Currently, the wait time is only the period required for a single
conversion that is calculated as the reciprocal of the sampling frequency.
However we also need to wait for the the previous conversion to complete.
Otherwise we probably get the conversion result for the previous
configuration when the sampling frequency is lower.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
pm_runtime_get_sync() and pm_runtime_put_autosuspend() return 0 on
success, 1 if the device's runtime PM status was already requested status
or error code on failure. So a positive return value doesn't indicate an
error condition.
However, any non-zero return values from buffer preenable and postdisable
callbacks are recognized as an error and this driver reuses the return
value from pm_runtime_get_sync() and pm_runtime_put_autosuspend() in
these callbacks. This change fixes the false error detections.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This driver assumes that the device is operating in the continuous
conversion mode which performs the conversion continuously. So this driver
doesn't insert a wait time before reading the conversion register if the
configuration is not changed from a previous request.
This assumption is broken if the device is runtime suspended and entered
a power-down state. The forthcoming request causes reading a stale result
from the conversion register as the device is runtime resumed just before.
Fix it by adding a flag to detect that condition and insert a necessary
wait time.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ADS1015 device have two operating modes, continuous conversion mode
and single-shot mode. This driver assumes that the continuous conversion
mode is selected by runtime resume callback when the ADC result is
requested.
If CONFIG_PM is disabled, the device is always in the default single-shot
mode and no one begins a single conversion. So the conversion register
doesn't contain valid ADC result. Fix it by changing the continuous mode
in probe function.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ti-ads1015 driver supports ADS1015 and ADS1115 devices. The same
scale information is used for both devices in this driver, however they
have actually different values and the ADS1115's one is not correct.
These devices have the same full-scale input voltage range for each PGA
selection. So instead of adding another hardcoded scale information,
compute a correct scale on demand from each device's resolution.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ti-ads1015 driver has eight iio voltage channels and each iio channel
can hold own sampling frequency information.
The ADS1015 device only have a single config register which contains an
input multiplexer selection, PGA and data rate settings. So the driver
should load the correct settings when the input multiplexer selection is
changed.
However, regardless of which channlel is currently selected, changing any
iio channel's sampling frequency information immediately overwrites the
current data rate setting in the config register.
It breaks the current data rate setting if the different channel's sampling
frequency information is changed because the data rate setting is not
reloaded when the input multiplexer is switched.
This removes the unexpected config register update and correctly load the
data rate setting before getting adc result.
Cc: Daniel Baluta <daniel.baluta@gmail.com> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The mask of sns_key_info1 suggests the upper nybble is being extracted
however the following shift of 8 bits is too large and always results in
0. Fix this by shifting only by 4 bits to correctly get the upper nybble.
Detected by CoverityScan, CID#142891 ("Operands don't affect result")
Fixes: fa590c222fba ("staging: rts5208: add support for rts5208 and rts5288") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While running reboot tests w/ a specific set of USB devices (and
slub_debug enabled), I found that once every few hours my device would
be crashed with a stack that looked like this:
Investigation using kgdb() found that the wait queue that was passed
into wake_up() had been freed (it was filled with slub_debug poison).
I analyzed and instrumented the code and reproduced. My current
belief is that this is happening:
1. async_completed() is called (from IRQ). Moves "as" onto the
completed list.
2. On another CPU, proc_reapurbnonblock_compat() calls
async_getcompleted(). Blocks on spinlock.
3. async_completed() releases the lock; keeps running; gets blocked
midway through wake_up().
4. proc_reapurbnonblock_compat() => async_getcompleted() gets the
lock; removes "as" from completed list and frees it.
5. usbdev_release() is called. Frees "ps".
6. async_completed() finally continues running wake_up(). ...but
wake_up() has a pointer to the freed "ps".
The instrumentation that led me to believe this was based on adding
some trace_printk() calls in a select few functions and then using
kdb's "ftdump" at crash time. The trace follows (NOTE: in the trace
below I cheated a little bit and added a udelay(1000) in
async_completed() after releasing the spinlock because I wanted it to
trigger quicker):
<...>-2104 0d.h2 13759034us!: async_completed at start: as=ffffffc0cc638200
mtpd-2055 3.... 13759356us : async_getcompleted before spin_lock_irqsave
mtpd-2055 3d..1 13759362us : async_getcompleted after list_del_init: as=ffffffc0cc638200
mtpd-2055 3.... 13759371us+: proc_reapurbnonblock_compat: free_async(ffffffc0cc638200)
mtpd-2055 3.... 13759422us+: async_getcompleted before spin_lock_irqsave
mtpd-2055 3.... 13759479us : usbdev_release at start: ps=ffffffc0cc042080
mtpd-2055 3.... 13759487us : async_getcompleted before spin_lock_irqsave
mtpd-2055 3.... 13759497us!: usbdev_release after kfree(ps): ps=ffffffc0cc042080
<...>-2104 0d.h2 13760294us : async_completed before wake_up(): as=ffffffc0cc638200
To fix this problem we can just move the wake_up() under the ps->lock.
There should be no issues there that I'm aware of.
Signed-off-by: Douglas Anderson <dianders@chromium.org> Acked-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Make sure that the controller is runtime resumed when system suspending
to avoid an external abort when accessing the interrupt registers:
Unhandled fault: external abort on non-linefetch (0x1008) at 0xd025840a
...
[<c05481a4>] (musb_default_readb) from [<c0545abc>] (musb_disable_interrupts+0x84/0xa8)
[<c0545abc>] (musb_disable_interrupts) from [<c0546b08>] (musb_suspend+0x38/0xb8)
[<c0546b08>] (musb_suspend) from [<c04a57f8>] (platform_pm_suspend+0x3c/0x64)
This is easily reproduced on a BBB by enabling the peripheral port only
(as the host port may enable the shared clock) and keeping it
disconnected so that the controller is runtime suspended. (Well, you
would also need to the not-yet-merged am33xx-suspend patches by Dave
Gerlach to be able to suspend the BBB.)
This is a regression that was introduced by commit 1c4d0b4e1806 ("usb:
musb: Remove pm_runtime_set_irq_safe") which allowed the parent glue
device to runtime suspend and thereby exposed a couple of older issues:
Register accesses without explicitly making sure the controller is
runtime resumed during suspend was first introduced by commit c338412b5ded
("usb: musb: unconditionally save and restore the context on suspend")
in 3.14.
Commit a1fc1920aaaa ("usb: musb: core: make sure musb is in RPM_ACTIVE on
resume") later started setting the RPM status to active during resume,
and this was also implicitly relying on the parent always being active.
Since commit 71723f95463d ("PM / runtime: print error when activating a
child to unactive parent") this now also results in the following
warning:
musb-hdrc musb-hdrc.0: runtime PM trying to activate child device
musb-hdrc.0 but parent (47401400.usb) is not active
This patch has been verified on 4.13-rc2, 4.12 and 4.9 using a BBB
(the dsps glue would always be active also in 4.8).
Fixes: c338412b5ded ("usb: musb: unconditionally save and restore the context on suspend") Fixes: a1fc1920aaaa ("usb: musb: core: make sure musb is in RPM_ACTIVE on resume") Fixes: 1c4d0b4e1806 ("usb: musb: Remove pm_runtime_set_irq_safe") Cc: Alan Stern <stern@rowland.harvard.edu> Cc: Daniel Mack <zonque@gmail.com> Cc: Dave Gerlach <d-gerlach@ti.com> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Tony Lindgren <tony@atomide.com> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Bin Liu <b-liu@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The following commit cause a regression on ATI chipsets.
'commit e788787ef4f9 ("usb:xhci:Add quirk for Certain
failing HP keyboard on reset after resume")'
This causes pinfo->smbus_dev to be wrongly set to NULL on
systems with the ATI chipset that this function checks for first.
Added conditional check for AMD chipsets to avoid the overwriting
pinfo->smbus_dev.
Reported-by: Ben Hutchings <ben@decadent.org.uk> Fixes: e788787ef4f9 ("usb:xhci:Add quirk for Certain
failing HP keyboard on reset after resume")
cc: Nehal Shah <Nehal-bakulchandra.Shah@amd.com> Signed-off-by: Sandeep Singh <Sandeep.Singh@amd.com> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit e0429362ab15
("usb: Add device quirk for Logitech HD Pro Webcams C920 and C930e")
introduced quirk to workaround an issue with some Logitech webcams.
Apparently model C920-C has the same issue so applying
the same quirk as well.
See aforementioned commit message for detailed explanation of the problem.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Corsair Strafe RGB keyboard has trouble to initialize:
[ 1.679455] usb 3-6: new full-speed USB device number 4 using xhci_hcd
[ 6.871136] usb 3-6: unable to read config index 0 descriptor/all
[ 6.871138] usb 3-6: can't read configurations, error -110
[ 6.991019] usb 3-6: new full-speed USB device number 5 using xhci_hcd
[ 12.246642] usb 3-6: unable to read config index 0 descriptor/all
[ 12.246644] usb 3-6: can't read configurations, error -110
[ 12.366555] usb 3-6: new full-speed USB device number 6 using xhci_hcd
[ 17.622145] usb 3-6: unable to read config index 0 descriptor/all
[ 17.622147] usb 3-6: can't read configurations, error -110
[ 17.742093] usb 3-6: new full-speed USB device number 7 using xhci_hcd
[ 22.997715] usb 3-6: unable to read config index 0 descriptor/all
[ 22.997716] usb 3-6: can't read configurations, error -110
Although it may work after several times unpluging/pluging:
[ 68.195240] usb 3-6: new full-speed USB device number 11 using xhci_hcd
[ 68.337459] usb 3-6: New USB device found, idVendor=1b1c, idProduct=1b20
[ 68.337463] usb 3-6: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[ 68.337466] usb 3-6: Product: Corsair STRAFE RGB Gaming Keyboard
[ 68.337468] usb 3-6: Manufacturer: Corsair
[ 68.337470] usb 3-6: SerialNumber: 0F013021AEB8046755A93ED3F5001941
Tried three quirks: USB_QUIRK_DELAY_INIT, USB_QUIRK_NO_LPM and
USB_QUIRK_DEVICE_QUALIFIER, user confirmed that USB_QUIRK_DELAY_INIT alone
can workaround this issue. Hence add the quirk for Corsair Strafe RGB.
The race was introduced by me in commit 971316f0503a ("epoll:
ep_unregister_pollwait() can use the freed pwq->whead"). I did not
realize that nothing can protect eventpoll after ep_poll_callback() sets
->whead = NULL, only whead->lock can save us from the race with
ep_free() or ep_remove().
Move ->whead = NULL to the end of ep_poll_callback() and add the
necessary barriers.
TODO: cleanup the ewake/EPOLLEXCLUSIVE logic, it was confusing even
before this patch.
Hopefully this explains use-after-free reported by syzcaller:
BUG: KASAN: use-after-free in debug_spin_lock_before
...
_raw_spin_lock_irqsave+0x4a/0x60 kernel/locking/spinlock.c:159
ep_poll_callback+0x29f/0xff0 fs/eventpoll.c:1148
this is spin_lock(eventpoll->lock),
...
Freed by task 17774:
...
kfree+0xe8/0x2c0 mm/slub.c:3883
ep_free+0x22c/0x2a0 fs/eventpoll.c:865
Fixes: 971316f0503a ("epoll: ep_unregister_pollwait() can use the freed pwq->whead") Reported-by: 范龙飞 <long7573@126.com> Signed-off-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When fail to get needed page for pool, need to put allocated pages
into pool. But current code has a miscalculation of allocated pages,
correct it.
Signed-off-by: Xiangliang.Yu <Xiangliang.Yu@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Monk Liu <monk.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 'dir' parameter in xfrm_migrate() is a user-controlled byte which is used
as an array index. This can lead to an out-of-bound access, kernel lockup and
DoS. Add a check for the 'dir' value.
Using sg_miter_start and sg_miter_next, the buffer of an SG is kmap'ed
to *buff. The current code calls sg_miter_stop (and thus kunmap) on the
SG entry before the last access of *buff.
The patch moves the sg_miter_stop call after the last access to *buff to
ensure that the memory pointed to by *buff is still mapped.
Fixes: 4816c9406430 ("lib/mpi: Fix SG miter leak") Signed-off-by: Stephan Mueller <smueller@chronox.de> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ 5668.771453] BUG: spinlock bad magic on CPU#0, kworker/u2:3/9745
[ 5668.771850] lock: 0xce63ef20, .magic: 00000000, .owner: <none>/-1,
.owner_cpu: 0
[ 5668.772277] CPU: 0 PID: 9745 Comm: kworker/u2:3 Tainted: G W 4.12.0-03002-gec979a4-dirty #40
[ 5668.772796] Hardware name: Nokia RX-51 board
[ 5668.773071] Workqueue: phy1 wl1251_irq_work
[ 5668.773345] [<c010c9e4>] (unwind_backtrace) from [<c010a274>]
(show_stack+0x10/0x14)
[ 5668.773803] [<c010a274>] (show_stack) from [<c01545a4>]
(do_raw_spin_lock+0x6c/0xa0)
[ 5668.774230] [<c01545a4>] (do_raw_spin_lock) from [<c06ca578>]
(_raw_spin_lock_irqsave+0x10/0x18)
[ 5668.774658] [<c06ca578>] (_raw_spin_lock_irqsave) from [<c048c010>]
(wl1251_op_tx+0x38/0x5c)
[ 5668.775115] [<c048c010>] (wl1251_op_tx) from [<c06a12e8>]
(ieee80211_tx_frags+0x188/0x1c0)
[ 5668.775543] [<c06a12e8>] (ieee80211_tx_frags) from [<c06a138c>]
(__ieee80211_tx+0x6c/0x130)
[ 5668.775970] [<c06a138c>] (__ieee80211_tx) from [<c06a3dbc>]
(ieee80211_tx+0xdc/0x104)
[ 5668.776367] [<c06a3dbc>] (ieee80211_tx) from [<c06a4af0>]
(__ieee80211_subif_start_xmit+0x454/0x8c8)
[ 5668.776824] [<c06a4af0>] (__ieee80211_subif_start_xmit) from
[<c06a4f94>] (ieee80211_subif_start_xmit+0x30/0x2fc)
[ 5668.777343] [<c06a4f94>] (ieee80211_subif_start_xmit) from
[<c0578848>] (dev_hard_start_xmit+0x80/0x118)
...
by adding the missing spin_lock_init().
Reported-by: Pavel Machek <pavel@ucw.cz> Cc: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Acked-by: Pavel Machek <pavel@ucw.cz> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Pavel Machek <pavel@ucw.cz> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Recent patch had an endian warning ie
cifs: return ENAMETOOLONG for overlong names in cifs_open()/cifs_lookup()
Signed-off-by: Steve French <smfrench@gmail.com> CC: Ronnie Sahlberg <lsahlber@redhat.com> Acked-by: Pavel Shilovsky <pshilov@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the maximum size of SMB2/3 header is set incorrectly which
leads to hanging of directory listing operations on encrypted SMB3
connections. Fix this by setting the maximum size to 170 bytes that
is calculated as RFC1002 length field size (4) + transform header
size (52) + SMB2 header size (64) + create response size (56).
Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com> Signed-off-by: Steve French <smfrench@gmail.com> Acked-by: Sachin Prabhu <sprabhu@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tests/attr.c: In function 'store_event':
tests/attr.c:66:27: error: format '%llu' expects argument of type 'long long unsigned int', but argument 6 has type '__u64 {aka long unsigned int}' [-Werror=format=]
snprintf(path, PATH_MAX, "%s/event-%d-%llu-%d", dir,
^
Signed-off-by: Ben Hutchings <ben@decadent.org.uk> Tested-by: Michael Cree <mcree@orcon.net.nz> Signed-off-by: Matt Turner <mattst88@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The memory_pressure control file was incorrectly set up without
a private value (0, by default). As a result, this control
file was treated like memory_migrate on read. By adding back the
FILE_MEMORY_PRESSURE private value, the correct memory pressure value
will be returned.
Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org> Fixes: 7dbdb199d3bf ("cgroup: replace cftype->mode with CFTYPE_WORLD_WRITABLE") Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When !NUMA, cpumask_of_node(@node) equals cpu_online_mask regardless of
@node. The assumption seems that if !NUMA, there shouldn't be more than
one node and thus reporting cpu_online_mask regardless of @node is
correct. However, that assumption was broken years ago to support
DISCONTIGMEM and whether a system has multiple nodes or not is
separately controlled by NEED_MULTIPLE_NODES.
This means that, on a system with !NUMA && NEED_MULTIPLE_NODES,
cpumask_of_node() will report cpu_online_mask for all possible nodes,
indicating that the CPUs are associated with multiple nodes which is an
impossible configuration.
This bug has been around forever but doesn't look like it has caused any
noticeable symptoms. However, it triggers a WARN recently added to
workqueue to verify NUMA affinity configuration.
Fix it by reporting empty cpumask on non-zero nodes if !NUMA.
ceph_readpage() unlocks page prematurely prematurely in the case
that page is reading from fscache. Caller of readpage expects that
page is uptodate when it get unlocked. So page shoule get locked
by completion callback of fscache_read_or_alloc_pages()
Wendy Wang reported off-list that a RAS HWPOISON-SOFT test case failed
and bisected it to the commit 479f854a207c ("mm, page_alloc: defer
debugging checks of pages allocated from the PCP").
The problem is that a page that was poisoned with madvise() is reused.
The commit removed a check that would trigger if DEBUG_VM was enabled
but re-enabling the check only fixes the problem as a side-effect by
printing a bad_page warning and recovering.
The root of the problem is that an madvise() can leave a poisoned page
on the per-cpu list. This patch drains all per-cpu lists after pages
are poisoned so that they will not be reused. Wendy reports that the
test case in question passes with this patch applied. While this could
be done in a targeted fashion, it is over-complicated for such a rare
operation.
Commit 7c051267931a ("mm, fork: make dup_mmap wait for mmap_sem for
write killable") made it possible to kill a forking task while it is
waiting to acquire its ->mmap_sem for write, in dup_mmap().
However, it was overlooked that this introduced an new error path before
the new mm_struct's ->uprobes_state.xol_area has been set to NULL after
being copied from the old mm_struct by the memcpy in dup_mm(). For a
task that has previously hit a uprobe tracepoint, this resulted in the
'struct xol_area' being freed multiple times if the task was killed at
just the right time while forking.
Fix it by setting ->uprobes_state.xol_area to NULL in mm_init() rather
than in uprobe_dup_mmap().
With CONFIG_UPROBE_EVENTS=y, the bug can be reproduced by the same C
program given by commit 2b7e8665b4ff ("fork: fix incorrect fput of
->exe_file causing use-after-free"), provided that a uprobe tracepoint
has been set on the fork_thread() function. For example:
For asynchronous operation, SGs are allocated without a page mapped to
them or with a page that is not used (ref-counted). If the SGL is freed,
the code must only call put_page for an SG if there was a page assigned
and ref-counted in the first place.
This fixes a kernel crash when using io_submit with more than one iocb
using the sendmsg and sendpage (vmsplice/splice) interface.
Signed-off-by: Stephan Mueller <smueller@chronox.de> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Compare the number of bytes actually seen on the wire to the byte
count field returned by the slave device.
Previously we just overwrote the byte count returned by the slave
with the real byte count and let the caller figure out if the
message was sane.
Signed-off-by: Stephen Douthit <stephend@adiengineering.com> Tested-by: Dan Priamo <danp@adiengineering.com> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
projects / linux-4.9.git / log