xen.git
20 years agoFix ia64 build. Two typos in header files.
kaf24@firebug.cl.cam.ac.uk [Thu, 24 Nov 2005 10:03:07 +0000 (11:03 +0100)]
Fix ia64 build. Two typos in header files.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMake the microcode driver builtin.
rread@ubuntu.eng.hq.xensource.com [Thu, 24 Nov 2005 05:04:13 +0000 (22:04 -0700)]
Make the microcode driver builtin.

As a module it complains about unknown symbols sys_munlock and sys_lock.

20 years agoDisable crufty modules still using cli and friends.
rread@ubuntu.eng.hq.xensource.com [Thu, 24 Nov 2005 04:55:07 +0000 (21:55 -0700)]
Disable crufty modules still using cli and friends.

20 years agoCompile IDE drivers as builtins instead of modules.
rread@ubuntu.eng.hq.xensource.com [Thu, 24 Nov 2005 04:12:38 +0000 (21:12 -0700)]
Compile IDE drivers as builtins instead of modules.

This makes the -xen config compatible with the distribution kernel configs,
but at the cost of some bloat when being used as a guest kernel.

20 years agoReverse accidental commits for changesets 8001 and 7996-7999.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:37:33 +0000 (19:37 +0000)]
Reverse accidental commits for changesets 8001 and 7996-7999.

The Mercurial Queues extensions suddenly looks a lot less useful than it first
seemed.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:31:22 +0000 (19:31 +0000)]
Merged.

20 years agoAdded comment describing usage.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:31:14 +0000 (19:31 +0000)]
Added comment describing usage.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoimported patch quiet_linux_build.patch
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:19:24 +0000 (19:19 +0000)]
imported patch quiet_linux_build.patch

20 years agoimported patch trace-thread.patch
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:19:24 +0000 (19:19 +0000)]
imported patch trace-thread.patch

20 years agoimported patch my-config
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:19:23 +0000 (19:19 +0000)]
imported patch my-config

20 years agoimported patch logging-thread.patch
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:19:23 +0000 (19:19 +0000)]
imported patch logging-thread.patch

20 years agoMerged.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:19:04 +0000 (19:19 +0000)]
Merged.

20 years agoDon't treat the failure of ifconfig vifX.Y down as a fatal error. This hasn't
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:16:54 +0000 (19:16 +0000)]
Don't treat the failure of ifconfig vifX.Y down as a fatal error.  This hasn't
emerged as a problem in the field, AFAIK, but it seems that distro scripts
could have brought the device down already by this point (Debian stable's
net.agent does ifdown vifX.Y, and it wouldn't surprise me if someone decided
to do ifconfig vifX.Y down if this failed, for example).

Logging but ignoring the failure seems sensible.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove the frontend device directories as well as the backend ones.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 19:13:13 +0000 (19:13 +0000)]
Remove the frontend device directories as well as the backend ones.

Closes bug #424.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoInside xenbus_switch_state, don't write the new state if the node has already
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 18:47:54 +0000 (18:47 +0000)]
Inside xenbus_switch_state, don't write the new state if the node has already
disappeared.  This happens when the driver is being torn down, and we don't
want to resurrect the device directory in the store because of it.

Partially addresses bug #424.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded a hotplug KOBJ_OFFLINE call, for symmetry with the ONLINE call used here.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 18:45:57 +0000 (18:45 +0000)]
Added a hotplug KOBJ_OFFLINE call, for symmetry with the ONLINE call used here.
We need a hook to be able to close down the device properly, but the REMOVE
event is no good because we don't get the script and vif environment variables
set in that case.

Closes bug #417.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoBuild the mouse device into the kernel.
rread@ubuntu.eng.hq.xensource.com [Wed, 23 Nov 2005 17:09:18 +0000 (10:09 -0700)]
Build the mouse device into the kernel.

Both RHEL and SLES do this.

20 years agoFix Xen to build with gcc-4.1.
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 13:15:35 +0000 (14:15 +0100)]
Fix Xen to build with gcc-4.1.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoAdd __iomem and __user pointer annotations to linux sparse tree.
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 13:14:01 +0000 (14:14 +0100)]
Add __iomem and __user pointer annotations to linux sparse tree.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoFix __gpfn_to_mfn() to work correctly when not executed
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 12:58:44 +0000 (13:58 +0100)]
Fix __gpfn_to_mfn() to work correctly when not executed
in the queried domain's context.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoClean up use of get_pfn_from_mfn().
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 12:54:40 +0000 (13:54 +0100)]
Clean up use of get_pfn_from_mfn().

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMove arch-specific portions of grant-table implementation
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 12:50:14 +0000 (13:50 +0100)]
Move arch-specific portions of grant-table implementation
out of common code. Ready for use by ia64.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerge.
sos22@douglas.cl.cam.ac.uk [Wed, 23 Nov 2005 12:38:34 +0000 (13:38 +0100)]
Merge.

Signed-off-by: Steven Smith, sos22@cam.ac.uk
20 years agoMerge.
sos22@douglas.cl.cam.ac.uk [Wed, 23 Nov 2005 12:35:50 +0000 (13:35 +0100)]
Merge.

Signed-off-by: Steven Smith, sos22@cam.ac.uk
20 years agoMake sure that accesses to the machine_to_phys table all go through
sos22@douglas.cl.cam.ac.uk [Wed, 23 Nov 2005 12:34:25 +0000 (13:34 +0100)]
Make sure that accesses to the machine_to_phys table all go through
suitable macros.

Signed-off-by: Steven Smith, sos22@cam.ac.uk
20 years agoxm-test: add tests for network-attach command
root@localhost.localdomain [Wed, 23 Nov 2005 11:32:44 +0000 (12:32 +0100)]
xm-test: add tests for network-attach command

Signed-off-by: Murillo Fernandes Bernardes <mfb@br.ibm.com>
20 years agoAdd tests for bug #331
dan@guaranine.beaverton.ibm.com [Wed, 23 Nov 2005 11:31:55 +0000 (12:31 +0100)]
Add tests for bug #331

20 years agoMake 04_memset_smallmem differentiate between console run-away and console
dan@guaranine.beaverton.ibm.com [Wed, 23 Nov 2005 11:31:30 +0000 (12:31 +0100)]
Make 04_memset_smallmem differentiate between console run-away and console
daemon death.

20 years agoFix to xm create --help from Ken Grant.
emellor@leeni.uk.xensource.com [Wed, 23 Nov 2005 11:29:58 +0000 (12:29 +0100)]
Fix to xm create --help from Ken Grant.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoBUG_ON test for xmalloc alignment should not be excluded
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 11:22:58 +0000 (12:22 +0100)]
BUG_ON test for xmalloc alignment should not be excluded
for ia64. SMP_CACHE_BYTES should be increased for uniproc
ia64 systems (how many uniproc ia64 boxes are there?!).

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoIntel systems with more than 8 CPUs require that the APIC is configured
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 11:11:05 +0000 (12:11 +0100)]
Intel systems with more than 8 CPUs require that the APIC is configured
in clustered mode with the apic=bigsmp kernel parameter. This patch does
the APIC mode selection automatically without needing to manually specify
the kernel paremeter.

(KAF note: I see the patch is also present in Linux 2.6.14.2)

Signed-off-by: Nitin Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
20 years agoAdd -Wdeclaration-after-statement to CFLAGS only if the
kaf24@firebug.cl.cam.ac.uk [Wed, 23 Nov 2005 10:15:18 +0000 (11:15 +0100)]
Add -Wdeclaration-after-statement to CFLAGS only if the
compiler supports it. Move test-gcc-flag function to
Config.mk where it can be used by any component of the
build.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoAvoid GPF when attempting to save / restore 64bit guests. More thought
smh22@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 18:10:50 +0000 (19:10 +0100)]
Avoid GPF when attempting to save / restore 64bit guests. More thought
required as to best way to get M2P mfns under 64...

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoAdd -Wdeclaration-after-statement to Xen and tools build.
kaf24@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 17:44:08 +0000 (18:44 +0100)]
Add -Wdeclaration-after-statement to Xen and tools build.
Fix the compile errors that result from this.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoTidy up debugging output.
smh22@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 17:21:22 +0000 (18:21 +0100)]
Tidy up debugging output.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoBetter handle page type info becoming incoherent during live migrate.
smh22@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 16:50:41 +0000 (17:50 +0100)]
Better handle page type info becoming incoherent during live migrate.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Tue, 22 Nov 2005 16:00:56 +0000 (17:00 +0100)]
Merged.

20 years agoMake it possible to run valgrind on code linked with the libxc libraries.
emellor@leeni.uk.xensource.com [Tue, 22 Nov 2005 15:31:16 +0000 (16:31 +0100)]
Make it possible to run valgrind on code linked with the libxc libraries.
Valgrind cannot see when a hypercall has initialised entries in a data
structure, so appropriate memsets have been placed before using dom0_op_t,
privcmd_hypercall_t, and a couple of miscellaneous blocks passed into this
layer.  This initialises the block so that valgrind considers it to be valid,
but the data therein will be immediately overwritten by the hypercall, all
being well.

These changes are semantically neutral if -DVALGRIND is not set.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoWhen a foreign page is mapped via a grant reference, it
kaf24@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 14:53:22 +0000 (15:53 +0100)]
When a foreign page is mapped via a grant reference, it
must also be unmapped explicitly via the grant-table
interface (GNTTABOP_unmap_grant_ref). If not, the guest
ends up with a dangling grant reference that is not cleared
up until the guest dies.

Because this can obviously lead to deferred hard-to-debug
problems, debug builds of Xen use a 'spare' PTE flag to
track granted mappings and to crash a domain if it attempts
to free such a PTE without using GNTTABOP_unmap_grant_ref.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoIgnore live flag on PAE/64-bit migrations, rather than
kaf24@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 11:04:03 +0000 (12:04 +0100)]
Ignore live flag on PAE/64-bit migrations, rather than
failing the migration entirely.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoSHADOW_CONTROL_OP_OFF should be checkign whether shadow mode
kaf24@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 10:53:45 +0000 (11:53 +0100)]
SHADOW_CONTROL_OP_OFF should be checkign whether shadow mode
is actually currently enabled.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix the zombie domain when destroying 2.6 linux vmx guest.
kaf24@firebug.cl.cam.ac.uk [Tue, 22 Nov 2005 06:08:57 +0000 (07:08 +0100)]
Fix the zombie domain when destroying 2.6 linux vmx guest.
The zombie domain is caused by unreleased shadow page
because of lack of put_shadow_ref.
When unshadow a L1 page, set l2e to zero and put ref.

Signed-off-by: <xiaofeng.ling@intel.com>
20 years agoBuild the USB boot keyboard and mouse drivers as modules, so the real
rread@ubuntu.eng.hq.xensource.com [Mon, 21 Nov 2005 22:19:26 +0000 (15:19 -0700)]
Build the USB boot keyboard and mouse drivers as modules, so the real
USB modules will be loaded after boot.

20 years agoFix syntax error.
smh22@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 18:28:50 +0000 (19:28 +0100)]
Fix syntax error.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 18:08:44 +0000 (19:08 +0100)]
Merged.

20 years agoMisc fixes for breakage when moving to new Python/C interface style.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 18:08:27 +0000 (19:08 +0100)]
Misc fixes for breakage when moving to new Python/C interface style.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agouse help in create.py for "xm help create"
kgrant@localhost.localdomain [Mon, 21 Nov 2005 17:53:31 +0000 (18:53 +0100)]
use help in create.py for "xm help create"

20 years agoCatch parsing errors in xm create.
xenrtd@shispar-rt.testdev.hq.xensource.com [Mon, 21 Nov 2005 17:52:40 +0000 (18:52 +0100)]
Catch parsing errors in xm create.

20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 17:47:31 +0000 (18:47 +0100)]
Merged.

20 years agoAs with xs.c, change the module interface to match the current Python/C
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 17:46:09 +0000 (18:46 +0100)]
As with xs.c, change the module interface to match the current Python/C
tutorial, simplifying the interface here, and making it more conventional.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove unnecessary assert.
smh22@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 17:45:09 +0000 (18:45 +0100)]
Remove unnecessary assert.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoTidy use of module / package names.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 17:34:25 +0000 (18:34 +0100)]
Tidy use of module / package names.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 17:33:27 +0000 (18:33 +0100)]
Merged.

20 years agoReturn the changed path as an argument to the watch callback. This is useful
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 17:33:17 +0000 (18:33 +0100)]
Return the changed path as an argument to the watch callback.  This is useful
for the test CD, and simplifies DevController everso slightly.  It ought to be
generally useful, too.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoChange semantics of gnttab_transfer to take the page away
kaf24@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 17:10:35 +0000 (18:10 +0100)]
Change semantics of gnttab_transfer to take the page away
from the calling domain, even on error (except for special
error GNTST_bad_page). This has a knock-on advantage of
simplifying Linux's netback driver.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoRemove many unnecessary uses of keyword arguments, to reduce complexity.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 16:48:22 +0000 (17:48 +0100)]
Remove many unnecessary uses of keyword arguments, to reduce complexity.

Have the self parameter to all the methods be an XsHandle, not a PyObject, to
avoid casting.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 16:30:43 +0000 (17:30 +0100)]
Merged.

20 years agoFix removal of use of scheduler, broken earlier today by mismerge.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 16:30:22 +0000 (17:30 +0100)]
Fix removal of use of scheduler, broken earlier today by mismerge.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 16:06:42 +0000 (17:06 +0100)]
Merged.

20 years agoChanges check for "udev", so that it works on debian sid (that does not
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 16:03:21 +0000 (17:03 +0100)]
Changes check for "udev", so that it works on debian sid (that does not
have /sbin/udev anymore).

Signed-off-by: Murillo F. Bernardes <mfb@br.ibm.com>
20 years agoUpdate the memory_op() hypercall. Add two new subcommands, to
kaf24@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 15:56:39 +0000 (16:56 +0100)]
Update the memory_op() hypercall. Add two new subcommands, to
query a domain's current and maximum memory reservation. Also,
XENMEM_maximum_ram_page now returns the max_page directly,
rather than writing through a passed-in pointer.

Also, disable PAE in the default config (accidentally checked
in two changesets ago).

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix 'belt + braces' check in xc_linux_save and improve error handling
smh22@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 15:40:16 +0000 (16:40 +0100)]
Fix 'belt + braces' check in xc_linux_save and improve error handling
for live migrate (this includes making shadow enable idempotent).

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoLinux swiotlb now uses correct method for determining if any RAM
kaf24@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 15:10:19 +0000 (16:10 +0100)]
Linux swiotlb now uses correct method for determining if any RAM
is mapped above 2GB.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMute error message on SMP resume if a secondary CPU is
kaf24@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 14:06:49 +0000 (15:06 +0100)]
Mute error message on SMP resume if a secondary CPU is
already back online.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 13:48:31 +0000 (14:48 +0100)]
Merged.

20 years agoReplace initialiser for readonly parameter in xshandle_init that should never
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 13:48:10 +0000 (14:48 +0100)]
Replace initialiser for readonly parameter in xshandle_init that should never
have been deleted.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix grant-table transfer implementation. Also fix transfer
kaf24@firebug.cl.cam.ac.uk [Mon, 21 Nov 2005 13:17:50 +0000 (14:17 +0100)]
Fix grant-table transfer implementation. Also fix transfer
error paths in network frontend and backend drivers.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:17:29 +0000 (13:17 +0100)]
Merged.

20 years agoRemoved, now unused.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:15:08 +0000 (13:15 +0100)]
Removed, now unused.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAppend the new device config to our internal structure, in device_create. This
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:14:38 +0000 (13:14 +0100)]
Append the new device config to our internal structure, in device_create.  This
will preserve the new device config across migration.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoCatch RuntimeError inside initDomain, for better handling of out-of-memory, etc.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:13:57 +0000 (13:13 +0100)]
Catch RuntimeError inside initDomain, for better handling of out-of-memory, etc.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove last remaining use of scheduler.py, so we can remove the whole module.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:12:09 +0000 (13:12 +0100)]
Remove last remaining use of scheduler.py, so we can remove the whole module.
It's not worth a whole module to save us 8 characters of typing to start a
timer.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoDowngrade debug message to level "trace".
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:09:45 +0000 (13:09 +0100)]
Downgrade debug message to level "trace".

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded another test. Fix the existing tests to close the written config file
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:03:45 +0000 (13:03 +0100)]
Added another test.  Fix the existing tests to close the written config file
before running the test upon it, to ensure that the file is flushed.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded reset method to Opts so that create.parseCommandLine can be used more than
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 12:00:36 +0000 (13:00 +0100)]
Added reset method to Opts so that create.parseCommandLine can be used more than
once.  This makes it possible to unit test this code.  Remove the odd aliasing
of gopts to opts inside parseCommandLine.

Rename local variables globals and locals inside opts.load -- these are built-in
functions.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove whole piles of cruft from the connection layer. The only thing that was
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:55:51 +0000 (12:55 +0100)]
Remove whole piles of cruft from the connection layer.  The only thing that was
using connectionMade, connectionLost, startedConnecting, clientConnectionLost,
and clientConnectionFailed was the test classes (TestClientFactory,
TestClientProtocol, TestServerFactory, TestServerProtocol), so they were
actually testing lots of code that no-one else was using.  All these classes
have gone, freeing up lots of cruft around.

The useless classes Factory, ServerFactory, ClientFactory, RelocationFactory,
UnixServerConnection, and TCPServerConnection have gone, in favour of passing
the Protocol class directly to SocketListener, and using SocketServerConnection
directly.

connectTransport has gone, in favour of overriding connect directly.

Piles of closedown cruft has gone, as this was only supporting the correct
output of connectionLost events, none of which we need.

Unused SocketServerConnection.{getHost,getPeer} have gone.

Mark some parameters in relocate as unused.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded method so that SrvDaemon can be run in the foreground without all the
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:44:43 +0000 (12:44 +0100)]
Added method so that SrvDaemon can be run in the foreground without all the
forking implied by the start method.  This makes it easier to run valgrind
against Xend, or run it with a pydebug-compiled Python.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix typo, and failure to unpack singleton tuple correctly.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:42:47 +0000 (12:42 +0100)]
Fix typo, and failure to unpack singleton tuple correctly.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded unit test for the DevController subclasses.
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:41:39 +0000 (12:41 +0100)]
Added unit test for the DevController subclasses.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoChange the module registration declarations and change the object creation and
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:40:02 +0000 (12:40 +0100)]
Change the module registration declarations and change the object creation and
deallocation to match those in the Python/C tutorial.  This has been done to
make the interface here more conventional, in the hunt for bug #411.  One
double deallocation has been fixed, but this would only occur if we failed to
create the connection to Xenstore, and so unfortunately is not the cause of
that bug.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove all unnecessary uses of keyword arguments, to reduce complexity. In
emellor@leeni.uk.xensource.com [Mon, 21 Nov 2005 11:34:37 +0000 (12:34 +0100)]
Remove all unnecessary uses of keyword arguments, to reduce complexity.  In
many places we were supporting keywords for calling the methods here, and never
using that feature (especially where all the arguments were compulsory anyway)
and in some places we were even accepting positional and keyword arguments, and
not actually taking any arguments to the method at all!

Fix the permission parsing string in xspy_set_permissions -- this would have
blown up horribly if anyone were actually calling this code.

Fix the documentation comments to include the transaction handle throughout.

Have the self parameter to all the methods be an XsHandle, not a PyObject, to
avoid casting.

Remove unused pyvalue_int and pyvalue_str.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded document to collate information regarding distro-specific
emellor@leeni.uk.xensource.com [Sun, 20 Nov 2005 14:40:53 +0000 (15:40 +0100)]
Added document to collate information regarding distro-specific
incompatibilities in the commands that we are using in the scripts.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years ago1) "brctl show" has no parameters.
dykman@us.ibm.com [Sun, 20 Nov 2005 14:37:08 +0000 (15:37 +0100)]
1) "brctl show" has no parameters.
2) Add a -n flag to iptables -L to skip DNS lookups

20 years agoFirst stab at a xend-config.sxp man page.
dan@localhost.localdomain [Sun, 20 Nov 2005 14:01:15 +0000 (15:01 +0100)]
First stab at a xend-config.sxp man page.

20 years agoMerge with xen-ia64-unstable.hg.
kaf24@firebug.cl.cam.ac.uk [Sun, 20 Nov 2005 09:19:38 +0000 (10:19 +0100)]
Merge with xen-ia64-unstable.hg.

20 years agoSmall build fix to Python xenstore interface library.
kaf24@firebug.cl.cam.ac.uk [Sun, 20 Nov 2005 09:18:29 +0000 (10:18 +0100)]
Small build fix to Python xenstore interface library.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
20 years agoRecode usage of FPSR_DEFAULT due to strange double declaration
djm@kirby.fc.hp.com [Sat, 19 Nov 2005 21:41:08 +0000 (15:41 -0600)]
Recode usage of FPSR_DEFAULT due to strange double declaration

20 years agoThe patch allows one to configure NR_CPUS >8 (16 by default, same as the
kaf24@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 09:37:14 +0000 (10:37 +0100)]
The patch allows one to configure NR_CPUS >8 (16 by default, same as the
native) for x86_64 xenlinux. I tested 16-way xenolinux (using changeset
7892) on a 16-way SMP machine with >8GB memory. It worked fine as long
as I did sanity tests like kernel build (make -j64).

Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
20 years agoEnable the Xen hypervisor ACM security policy to be determined at boot
kaf24@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 09:35:38 +0000 (10:35 +0100)]
Enable the Xen hypervisor ACM security policy to be determined at boot
time (instead of at compile time). Allows a single xen.gz binary to
run under different security policy types, which simplifies
distributions considerably. To this end, we replace the current single
compile switch in Config.mk with the following two:

1. ACM_SECURITY (y/n) if "n", then the ACM is not compiled and no security
policies can be configured at boot time. If "y", then the ACM is compiled
and can be configured into any of the available policy types (null,
chwall, ste, chwall_ste) by specifying a valid boot policy .bin file in
the grub boot configuration.

2. ACM_DEFAULT_SECURITY_POLICY (null / chwall / ste / chwall_ste)
specifies the policy into which Xen defaults in the case that security is
enabled but no boot policy file is specified, or the specified policy file
is invalid (wrong version, magic, corrupted, ...)

While the type of the enabled policy (null, ste, ...) is decided once
during boot for the whole boot cycle, the policy instantiation can be
changed during operation as usual with the security tools.

Signed-off by: Tomas Lendacky <toml@us.ibm.com>
Signed-off by: Reiner Sailer <sailer@us.ibm.com>

20 years agoThis patch is to fix a few performance "bugs" in the xenmon.py script.
kaf24@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 09:29:28 +0000 (10:29 +0100)]
This patch is to fix a few performance "bugs" in the xenmon.py script.
No functionality is added or removed. Tested on x86/32 smp and x86/64.

Signed-off-by: Rob Gardner <rob.gardner@hp.com>
20 years agoFix shadow log-dirty mode for x86_64 xenlinux. Tested
kaf24@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 09:28:24 +0000 (10:28 +0100)]
Fix shadow log-dirty mode for x86_64 xenlinux. Tested
PAE xenlinux as well.

Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
20 years agoMerge
djm@kirby.fc.hp.com [Sat, 19 Nov 2005 05:16:29 +0000 (23:16 -0600)]
Merge

20 years agoFall back to zx1-based configs, defconfigs don't work yet to boot domU
djm@kirby.fc.hp.com [Sat, 19 Nov 2005 05:06:09 +0000 (23:06 -0600)]
Fall back to zx1-based configs, defconfigs don't work yet to boot domU

20 years agoRevert 04a7decd0645b23f1f398340e927eb191f0e754d since it's not needed.
cl349@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 03:35:37 +0000 (04:35 +0100)]
Revert 04a7decd0645b23f1f398340e927eb191f0e754d since it's not needed.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoMove tpmfe.h file next to other xen specific header files.
cl349@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 01:27:20 +0000 (02:27 +0100)]
Move tpmfe.h file next to other xen specific header files.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
--HG--
rename : linux-2.6-xen-sparse/include/linux/tpmfe.h => linux-2.6-xen-sparse/include/asm-xen/tpmfe.h

20 years agoFix support for sxp domain configuration files.
cl349@firebug.cl.cam.ac.uk [Sat, 19 Nov 2005 00:59:43 +0000 (01:59 +0100)]
Fix support for sxp domain configuration files.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoOnly resume xenbus /after/ we've prepared our VCPUs; otherwise the
smh22@firebug.cl.cam.ac.uk [Fri, 18 Nov 2005 18:56:03 +0000 (19:56 +0100)]
Only resume xenbus /after/ we've prepared our VCPUs; otherwise the
VCPU hotplug callback can race with our vcpu_prepare.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoDo not allow an uninitialised VCPU to be brought up. Also
kaf24@firebug.cl.cam.ac.uk [Fri, 18 Nov 2005 18:25:17 +0000 (19:25 +0100)]
Do not allow an uninitialised VCPU to be brought up. Also
check VCPUOP_up return code in the guest and BUG() on
failure.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix read of guest psr to properly set dfh and other bits (by Matt Chapman)
djm@kirby.fc.hp.com [Fri, 18 Nov 2005 17:04:58 +0000 (11:04 -0600)]
Fix read of guest psr to properly set dfh and other bits (by Matt Chapman)

20 years agoProperly pass iha/itir if data miss occurs when fetching privop (by Matt Chapman)
djm@kirby.fc.hp.com [Fri, 18 Nov 2005 16:58:03 +0000 (10:58 -0600)]
Properly pass iha/itir if data miss occurs when fetching privop (by Matt Chapman)