xen.git
20 years agothis patch fixes X86_64 Xorg Xfree86 issues.
kaf24@firebug.cl.cam.ac.uk [Sun, 7 Aug 2005 08:56:57 +0000 (08:56 +0000)]
this patch fixes X86_64 Xorg Xfree86 issues.
Signed-off-by: Xin Li <xin.b.li@intel.com>
20 years agoFix perfc_defn.h to allow multiple inclusion.
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 15:47:20 +0000 (15:47 +0000)]
Fix perfc_defn.h to allow multiple inclusion.

20 years agoxend prints out some messages when starting, which is annoying. This
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 15:45:41 +0000 (15:45 +0000)]
xend prints out some messages when starting, which is annoying. This
patch send those to xend log file instead - people want them always
know where to find them.

Signed-off-by: Nguyen Anh Quynh <aquynh@gmail.com>
20 years agoMore time-interface fixes.
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 15:24:22 +0000 (15:24 +0000)]
More time-interface fixes.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix xenbus_read() usage in shutdown handler.
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 09:54:57 +0000 (09:54 +0000)]
Fix xenbus_read() usage in shutdown handler.

20 years ago- include linux/vmalloc.h for vmalloc
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 09:52:01 +0000 (09:52 +0000)]
- include linux/vmalloc.h for vmalloc
- include grant table includes from common.h rather than blkback.c,
  since both C files use grant tables
- include linux/version.h from block.h (a macro there uses it)
- include asm-xen/hypervisor.h from block.h (virt_to_machine)

Signed-off-by: Matthew Chapman <matthewc@hp.com>
20 years agoThis patch changes the vcpu-hotplug handler from using control message
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 09:50:51 +0000 (09:50 +0000)]
This patch changes the vcpu-hotplug handler from using control message
over to using xenstore for triggers.  Dropping the control messages also
fixes the issue with not being able to use xm vcpu-hotplug on dom0.

Signed-off-by: Ryan Harper <ryanh@us.ibm.com>
20 years agoThis patch changes the probe notifier, per Rusty's suggestion. It
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 09:49:22 +0000 (09:49 +0000)]
This patch changes the probe notifier, per Rusty's suggestion.  It
also converts the balloon driver to use the notifier structure, and
removes the root-watch workaround.

Signed-off-by: Dan Smith <danms@us.ibm.com>
20 years agoThis patch changes the shutdown driver and xend to use strings instead
kaf24@firebug.cl.cam.ac.uk [Sat, 6 Aug 2005 09:48:47 +0000 (09:48 +0000)]
This patch changes the shutdown driver and xend to use strings instead
of codes for signaling a shutdown request.

Also, we fix the return code for the shutdown notifier, so other
notifiers will get to run ;)

Signed-off-by: Dan Smith <danms@us.ibm.com>
20 years agoFix mach2phys table allocation for 32-bit PAE Xen.
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 15:11:46 +0000 (15:11 +0000)]
Fix mach2phys table allocation for 32-bit PAE Xen.
Found by Gerd Knorr.

20 years agoFix xtime_lock handling to avoid deadlock in sched_clock().
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 14:57:43 +0000 (14:57 +0000)]
Fix xtime_lock handling to avoid deadlock in sched_clock().
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoNo need to define errno defined at the top of reboot.c.
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:53:04 +0000 (09:53 +0000)]
No need to define errno defined at the top of reboot.c.
err2 is only needed when CONFIG_MAGIC_SYSRQ is defined.

Signed-off-by: Li Xin <xin.b.li@intel.com>
20 years agoxenstored has had a testsuite for some time. This splits the test
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:04:01 +0000 (09:04 +0000)]
xenstored has had a testsuite for some time.  This splits the test
targets into "make -s check" (the GNU standard) which is designed to run
in several seconds or less, with only "<dirname>:" and a seried of "."
for output, and "make fullcheck" which is more verbose and complete, and
runs in minutes.

I plan to submit a patch to plumb these tests in at the top level soon,
with the idea of having a "make -s check" which you can reasonable run
every checkin (time make check << time make), and a "make fullcheck"
which you can run after significant changes.

This patch speeds up testing by making tests one single "xs_test"
script, enhancing xs_test with a simple "expect" command.  It also only
runs the "high value" tests under "make check", and leaves the rest for
"make fullcheck".

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
20 years agoThe C interface no longer has a priority arg, but the python interface
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:02:58 +0000 (09:02 +0000)]
The C interface no longer has a priority arg, but the python interface
still takes one (and just doesn't use it).  Noone uses it anyway, so
nothing should break by taking it away.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
20 years agoAttached is a randomized tester for consoled. It's a patch against
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:02:09 +0000 (09:02 +0000)]
Attached is a randomized tester for consoled.  It's a patch against
xen-unstable.hg but it doesn't add itself to any of the Makefiles but
just as a subdirectory of consoled.  There's a README in the directory
that should be read otherwise you won't be able to do anything useful
with it :-)

Right now, we can get pretty far before seeing data corruption.
Previously, we'd see it almost immediately (after ~600 bytes, which is
the max the control channel can hold before filling up).

It's actual general enough that it could be also used to test network
traffic corruption but there are certainly better tools for that.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoInstead of queuing messages when the control channels are full, xcs just
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:01:30 +0000 (09:01 +0000)]
Instead of queuing messages when the control channels are full, xcs just
does nothing (see ctrl_interface.c:ctrl_chan_write_request()).

The following patch throttles the rate in which consoled writes data to
xcs.  With the current values, you get a responsive console but avoid
data corruption in most scenarios.

I'm able to get pretty far in my regression test with this patch.  With
higher throttle values I'm able to get even further (but the console
becomes painfully slow).

I implemented proper control channel queuing in xenctld in VM-Tools and
it's pretty nasty stuff.  This should prevent corruption for most users
until we can get rid of xcs.

Regards,

Anthony Liguori

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com
20 years agoxc_console is probably too chatty about termios failures since ssh
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 09:00:50 +0000 (09:00 +0000)]
xc_console is probably too chatty about termios failures since ssh
sessions apparently don't have proper term settings.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoThe attached patch:
kaf24@firebug.cl.cam.ac.uk [Fri, 5 Aug 2005 08:59:41 +0000 (08:59 +0000)]
The attached patch:

1. Converts the shutdown driver and xend to use the store instead of
   control messages,

2. Includes Anthony's xenstore notification code, and

3. Changes xend so that sysrq's are no longer sent as "special case"
   shutdown messages.  Store keys are cheap, so making the sysrq
   delivery less obscure is good.

I think I have made all of the appropriate modifications to Xend, but
it is complex, so I may have missed something.  Comments are welcome.

Signed-off-by: Dan Smith <danms@us.ibm.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoAttached are
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 18:51:55 +0000 (18:51 +0000)]
Attached are
three patches to fix a problem with Xend starting consoled.  consoled
depends on xenstored to be running and xenstored is started on demand in
Xend.  The patches change xenstored to manage its own pidfile, and have
xend start actually start up xenstored.

20 years agoFix my previous patch to the trap-frame creation code
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 17:37:09 +0000 (17:37 +0000)]
Fix my previous patch to the trap-frame creation code
in Xen. I was clobbering an in-use register (AL): use
CH instead.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerge
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 16:53:30 +0000 (16:53 +0000)]
Merge

20 years agoFix event-upcall masking when creating a trap frame
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 16:53:11 +0000 (16:53 +0000)]
Fix event-upcall masking when creating a trap frame
on guest kernel stack. Previously an exception would
clear an already-set mask.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoBlktap updates: request batching, O_DIRECT/AIO support.
akw27@arcadians.cl.cam.ac.uk [Thu, 4 Aug 2005 16:35:35 +0000 (16:35 +0000)]
Blktap updates: request batching, O_DIRECT/AIO support.

This patch changes the blktap code mapping pages to user space to be
faster and to allow page lookups to foreign mapped pages through linux
to do direct io.  An AIO test driver on this achieves comparable
performance to the in-kernel block-backend.

Signed-off-by: andrew.warfield@cl.cam.ac.uk
20 years agoAdd missing parallax makefile.
akw27@arcadians.cl.cam.ac.uk [Thu, 4 Aug 2005 15:02:09 +0000 (15:02 +0000)]
Add missing parallax makefile.
Seems to have vanished in the directory shuffle of last month.

Signed-off-by: andrew.warfield@cl.cam.ac.uk
Signed-off-by: Nguyen Anh Quynh <aquynh@gmail.com>
20 years agoFix two syntax errors in frame-and-sect decoding macros.
akw27@arcadians.cl.cam.ac.uk [Thu, 4 Aug 2005 14:32:21 +0000 (14:32 +0000)]
Fix two syntax errors in frame-and-sect decoding macros.
Both are in the non-grant-tables cases of
cset ae51fa1d32b23066c4342bbeb5f4dc401b82ea37.

Signed-off-by: andrew.warfield@cl.cam.ac.uk
20 years agoFix iopl() syscall so that we pass LTP's iopl02 test.
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 13:13:02 +0000 (13:13 +0000)]
Fix iopl() syscall so that we pass LTP's iopl02 test.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoThe attached patch removes all of the console handling code from Xend
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 09:00:16 +0000 (09:00 +0000)]
The attached patch removes all of the console handling code from Xend
and changes xm to simply exec /usr/libexec/xen/xc_console.

I've done a pretty good amount of testing and this seems to clear up all
of the Xend console issues in bugzilla.

This is a pretty big change to Xend though with a lot of deleted code so
I'd appreciate if people more familiar with the Xend codebase could look
through it critically.

In addition to the patch, the following files are no longer used:

tools/python/xen/util/console_client.py
tools/python/xen/xend/XendConsole.py
tools/python/xen/xend/server/SrvConsole.py
tools/python/xen/xend/server/SrvConsoleDir.py
tools/python/xen/xend/server/console.py

Regards,

Anthony Liguori

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoHere are another couple of places where 8 sectors/page is assumed,
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 08:58:18 +0000 (08:58 +0000)]
Here are another couple of places where 8 sectors/page is assumed,
which I missed in my first patch.

Signed-off-by: Matthew Chapman <matthewc@hp.com>
20 years agolibxenstore (used to access the daemon) should be licensed under LGPL.
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 08:57:37 +0000 (08:57 +0000)]
libxenstore (used to access the daemon) should be licensed under LGPL.
There are several people I am aware of interested in writing non-GPL
tools, and this code is trivial, so they'd just end up writing their
own.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
20 years agoMerge.
kaf24@firebug.cl.cam.ac.uk [Thu, 4 Aug 2005 08:49:45 +0000 (08:49 +0000)]
Merge.

20 years agoAdd linux 2.6.12.3 patch
iap10@freefall.cl.cam.ac.uk [Wed, 3 Aug 2005 21:17:17 +0000 (21:17 +0000)]
Add linux 2.6.12.3 patch
Signed-off-by: ian@xensource.com
20 years agoMove more drivers to bind_evtchn_to_irqhandler() interface.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 16:11:32 +0000 (16:11 +0000)]
Move more drivers to bind_evtchn_to_irqhandler() interface.
The most important ones are all done now.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoOn x86, Xen does not claim any serial port unless the user
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 14:44:19 +0000 (14:44 +0000)]
On x86, Xen does not claim any serial port unless the user
explicitly specifies a 'com1=' or 'com2=' configuration
string. This prevents users from being surprised when Xen
stops domain0 from accessing serial port registers.

On ia64 we still allow serial ports to get automatically
configured and locked down even if no 'com1' is specified
on the command line. This maintains the default preferred
by the IA64 developers.

Also, the method for specifying 'use existing baud rate'
has changed. Instead of specifying com1=0, you must now
specify e.g., com1=auto

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoVirtual device drivers use new bind_evtchn_to_irqhandler() call
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 12:51:35 +0000 (12:51 +0000)]
Virtual device drivers use new bind_evtchn_to_irqhandler() call
instead of directly interfacing with Linux IRQ subsystem. This
allows them to be used in situations where the IRQ subsystem is
not paravirtualised.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMErge.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:35:38 +0000 (09:35 +0000)]
MErge.

20 years agoMerge,
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:35:16 +0000 (09:35 +0000)]
Merge,

20 years agoThe attached trivial patch fixes a spurious gcc-4.0 warning in
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:34:37 +0000 (09:34 +0000)]
The attached trivial patch fixes a spurious gcc-4.0 warning in
xen/arch/x86/mm.c when building the current hg tip.  Due to -Werror,
this breaks builds with gcc-4.0.

Signed-off-by: Josh Triplett <josht@us.ibm.com>
20 years agogcc-3.4/ia64 compilation fixes.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:33:09 +0000 (09:33 +0000)]
gcc-3.4/ia64 compilation fixes.

The ia64 parts of this patch are already in xen-ia64-unstable.

Signed-off-by: Arun Sharma <arun.sharma@intel.com>
20 years ago1) Make more functions in io.c static
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:32:39 +0000 (09:32 +0000)]
1) Make more functions in io.c static
2) Fix memory leak in domain free
3) Fix error path in read (reported by aq)
4) Move reading of buffer limit to tty reconnect so that changes in
limit take affect more quickly.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoMerge.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:27:27 +0000 (09:27 +0000)]
Merge.

20 years agoManual merge.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:22:03 +0000 (09:22 +0000)]
Manual merge.

20 years agoUpdate -xen config file for net grant tables. Default to 'y' since this is the kernel...
iap10@freefall.cl.cam.ac.uk [Wed, 3 Aug 2005 09:11:49 +0000 (09:11 +0000)]
Update -xen config file for net grant tables. Default to 'y' since this is the kernel we want to use on the test CD and we want to have grant tables on by default.

Signed-off-by: ian@xensource.com
20 years agoManula merge.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:06:10 +0000 (09:06 +0000)]
Manula merge.

20 years agoFix typo in Linux header file. Totally harmless -- the code
kaf24@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 09:05:46 +0000 (09:05 +0000)]
Fix typo in Linux header file. Totally harmless -- the code
was never actually compiled in to a XenLinux kernel.

20 years agoModify default Linux 2.6 config files to build for PentiumPro instead of Pentium4...
iap10@freefall.cl.cam.ac.uk [Wed, 3 Aug 2005 09:05:29 +0000 (09:05 +0000)]
Modify default Linux 2.6 config files to build for PentiumPro instead of Pentium4. No idea how that ended up being the default.

Signed-off-by: ian@xensource.com
20 years agoFix automatic toggling of PAE in Linux config files
iap10@freefall.cl.cam.ac.uk [Wed, 3 Aug 2005 08:52:28 +0000 (08:52 +0000)]
Fix automatic toggling of PAE in Linux config files
Signed-off-by: ian@xensource.com
20 years agoSince we use configs from dist/install/boot in preference to the default configs...
iap10@freefall.cl.cam.ac.uk [Wed, 3 Aug 2005 08:25:58 +0000 (08:25 +0000)]
Since we use configs from dist/install/boot in preference to the default configs, update the PAE auto-switcher to work with these too.

Signed-off-by: ian@xensource.com
20 years agoDon't enable PAE by default just yet.
smh22@firebug.cl.cam.ac.uk [Wed, 3 Aug 2005 06:46:53 +0000 (06:46 +0000)]
Don't enable PAE by default just yet.

20 years agofix netdriver grant table option back to n by default
vh249@arcadians.cl.cam.ac.uk [Wed, 3 Aug 2005 00:07:48 +0000 (00:07 +0000)]
fix netdriver grant table option back to n by default

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agomerge?
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 23:13:50 +0000 (23:13 +0000)]
merge?

20 years agoFix migration for SMP guests with 1 vcpu.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 23:13:13 +0000 (23:13 +0000)]
Fix migration for SMP guests with 1 vcpu.
Free/setup timer irq on suspend/restore.
Only tested to localhost.  Also add initial code for >1 vcpu guests.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoupdate x86_64 config files for grant table in netdriver
vh249@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 22:57:51 +0000 (22:57 +0000)]
update x86_64 config files for grant table in netdriver

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoFix CONFIG_VTI build
djm@kirby.fc.hp.com [Tue, 2 Aug 2005 22:38:45 +0000 (15:38 -0700)]
Fix CONFIG_VTI build
Signed-off-by: Arun Sharma <arun.sharma@intel.com>
20 years agoRemove USB front/back from 2.4 build, totally.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:32:48 +0000 (19:32 +0000)]
Remove USB front/back from 2.4 build, totally.

20 years agoUpdate xenstore python bindings for "Implement watching of nodes which don't exist."
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:19:16 +0000 (19:19 +0000)]
Update xenstore python bindings for "Implement watching of nodes which don't exist."
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoFix SMP save/restore for 1 vcpu by adding suspend/resume functions for SMP.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:08:28 +0000 (19:08 +0000)]
Fix SMP save/restore for 1 vcpu by adding suspend/resume functions for SMP.
We have to free_irq the irq vectors for the SMP ipi vectors.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoFix unbind_ipi_from_irq - evtchn_to_irq[evtchn] not irq_to_evtchn[evtchn].
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:06:10 +0000 (19:06 +0000)]
Fix unbind_ipi_from_irq - evtchn_to_irq[evtchn] not irq_to_evtchn[evtchn].
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoAdd prototype for bind_ipi_on_cpu_to_irq.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:05:00 +0000 (19:05 +0000)]
Add prototype for bind_ipi_on_cpu_to_irq.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoAdd prototype for balloon_init_watcher.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 19:04:13 +0000 (19:04 +0000)]
Add prototype for balloon_init_watcher.
Also cleanup balloon_init_watcher whitespace.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoImplement watching of nodes which don't exist.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 18:04:28 +0000 (18:04 +0000)]
Implement watching of nodes which don't exist.
Requires permission check every time event is generated.
Requires generalization of permissions: ask arbitrary number of
parents whether it's OK to tell about node (eg. watching
/dir/subdir/x when /dir is deleted: root permissions will now
determine whether we fire event).
Add test that we don't leak information on whether a file exists
or not.
Signed-off-by: Rusty Russel <rusty@rustcorp.com.au>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoFix typo.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 18:04:00 +0000 (18:04 +0000)]
Fix typo.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoFix up non-valgrind case to send stderr to file like valgrind case does.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 17:59:59 +0000 (17:59 +0000)]
Fix up non-valgrind case to send stderr to file like valgrind case does.
(Having trouble with signal delivery and valgrind, causing spurious test
timeouts)
Signed-off-by: Rusty Russel <rusty@rustcorp.com.au>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoDon't abort on untranslatable errors: EINVAL and print warning
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 17:59:20 +0000 (17:59 +0000)]
Don't abort on untranslatable errors: EINVAL and print warning
Signed-off-by: Rusty Russel <rusty@rustcorp.com.au>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoMerging with xen-unstable
djm@kirby.fc.hp.com [Tue, 2 Aug 2005 17:20:46 +0000 (10:20 -0700)]
Merging with xen-unstable

20 years agomerge?
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 17:13:11 +0000 (17:13 +0000)]
merge?

20 years agoFix checkpointing - add save/restore support to the xenbus driver.
cl349@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 17:12:36 +0000 (17:12 +0000)]
Fix checkpointing - add save/restore support to the xenbus driver.
- Add xenbus driver suspend/resume functions
- Change xenbus irq to not be shared
- Translate store mfn in suspend record to pfn and back
- Make tools re-introduce the domain to the store
Signed-off-by: Rusty Russel <rusty@rustcorp.com.au>
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoBetter XenLinux page fault debug output for PAE.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 16:11:31 +0000 (16:11 +0000)]
Better XenLinux page fault debug output for PAE.
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFixes.
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 15:42:06 +0000 (15:42 +0000)]
Fixes.

20 years agoA few consoled tweaks. It's not worth building by default
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 14:51:06 +0000 (14:51 +0000)]
A few consoled tweaks. It's not worth building by default
right now as it isn;t actually used.

20 years agoRemove some binary files.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 12:37:37 +0000 (12:37 +0000)]
Remove some binary files.

20 years agoThis is the first patch to move the console code out of Xend and into a
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 12:18:47 +0000 (12:18 +0000)]
This is the first patch to move the console code out of Xend and into a
separate daemon.  In this patch I've included the daemon and a
front-end.  I've also have the console code removed from Xend locally
but I wanted to clarify a few things first before submitting that.

Here's how it works:

1) Daemon sits on control channel (via xcs) waiting for console messages
2) Buffers all console data received
3) Publishes the location of a tty that can be used for reading in
/console/<domid>/tty
4) Reads a value from the store to determine what the limit of console
data should be (/console/<domid>/limit) to prevent DoS attacks

Using it is quite simple, just run consoled and use the xc_console
program to connect to the console.  It uses syslog for logging.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoThis is the first patch to move the console code out of Xend and into a
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 12:18:15 +0000 (12:18 +0000)]
This is the first patch to move the console code out of Xend and into a
separate daemon.  In this patch I've included the daemon and a
front-end.  I've also have the console code removed from Xend locally
but I wanted to clarify a few things first before submitting that.

Here's how it works:

1) Daemon sits on control channel (via xcs) waiting for console messages
2) Buffers all console data received
3) Publishes the location of a tty that can be used for reading in
/console/<domid>/tty
4) Reads a value from the store to determine what the limit of console
data should be (/console/<domid>/limit) to prevent DoS attacks

Using it is quite simple, just run consoled and use the xc_console
program to connect to the console.  It uses syslog for logging.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoManual merge.
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:35:45 +0000 (11:35 +0000)]
Manual merge.

20 years agoDisable build of pygrub for now - does not build with e2fsprogs-1.35
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:34:38 +0000 (11:34 +0000)]
Disable build of pygrub for now - does not build with e2fsprogs-1.35

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoFix typo in pygrub code
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:27:58 +0000 (11:27 +0000)]
Fix typo in pygrub code

20 years agoThe attached patch converts the balloon driver and xend to use
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:21:11 +0000 (11:21 +0000)]
The attached patch converts the balloon driver and xend to use
xenstore instead of control messages.

Note: Because there is no way to set a watch on a non-existent key,
this patch includes a workaround to account for the fact that dom0's
store keys are not initialized by the tools before it boots.

Signed-off-by: Dan Smith <danms@us.ibm.com>
20 years ago> to "CONFIG_PAGING_LEVELS >= 3", maybe I forgot to include some
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:20:12 +0000 (11:20 +0000)]
> to "CONFIG_PAGING_LEVELS >= 3", maybe I forgot to include some
> into the patch ...

Yep, I did, shadow_public.c bits are missing, sorry for trouble.

  Gerd

20 years agoMissing ia64_stubs file for libxc.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 11:08:47 +0000 (11:08 +0000)]
Missing ia64_stubs file for libxc.

20 years agoFix calling point for do_softirq, which should only be checked
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:47:41 +0000 (02:47 -0800)]
Fix calling point for do_softirq, which should only be checked
at the very point back to guest domain. Or else unexpected domain
switch may happen in nested interrupt in Xen.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoRemainder of ACM patch (hgrrrr).
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 10:15:17 +0000 (10:15 +0000)]
Remainder of ACM patch (hgrrrr).

Signed-off-by: Reiner Sailer <sailer@watson.ibm.com>
Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoFix compilation issue, and then VTI dom0 is up again.
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:13:02 +0000 (02:13 -0800)]
Fix compilation issue, and then VTI dom0 is up again.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoFinal set of HV change to support multiple domains on VTI.
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:11:41 +0000 (02:11 -0800)]
Final set of HV change to support multiple domains on VTI.

Signed-off-by Anthony Xu <anthony.xu@intel.com>
Signed-off-by Eddie Dong <eddie.dong@intel.com>
Signed-off-by Fred Yang <fred.yang@intel.com>
Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoAdd hypercall continuation support, or else dom_mem_op may fail.
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:09:24 +0000 (02:09 -0800)]
Add hypercall continuation support, or else dom_mem_op may fail.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoDisallow reprogram_ac_timer to change itm, since there's only one
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:08:00 +0000 (02:08 -0800)]
Disallow reprogram_ac_timer to change itm, since there's only one
hardware timer source on IA64.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoAdd copyright info.
fred@xuni-t01.sc.intel.com [Tue, 2 Aug 2005 10:05:50 +0000 (02:05 -0800)]
Add copyright info.

Signed-off-by Ke Yu <ke.yu@intel.com>
Signed-off-by Kevin Tian <kevin.tian@intel.com>

20 years agoThe patches below clean-up sHype-related file names in the tools subtree
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 09:37:00 +0000 (09:37 +0000)]
The patches below clean-up sHype-related file names  in the tools subtree
and security policy-versioning and data structures in the xen subtree
(ACM):

 *  the very generic tools directory name "policy"  is changed to
    "security" and the included "policy_tool" to "secpol_tool"

 * the single constant POLICY_INTERFACE_VERSION is replaced
   by ACM_INTERFACE_VERSION; the security policy headers are
   versioned independently since the policy can be supplied
   at boot time instead of through the  hypervisor interface

Signed-off-by: Reiner Sailer <sailer@watson.ibm.com>
Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoThe attached patch adds the support that was only stubbed in to be able
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 09:31:47 +0000 (09:31 +0000)]
The attached patch adds the support that was only stubbed in to be able
to support having reading the boot loader config when you're using a
whole disk (eg, file:/root/disk.img,xvda,w) as opposed to just a
partition.

This reads the partition table in the MBR to find the active partition
and then passes that offset down into the filesystem reading code.

Signed-off-by: Jeremy Katz <katzj@redhat.com>
20 years ago* support proper scsi disk name
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 09:29:56 +0000 (09:29 +0000)]
* support proper scsi disk name
* xvd now has an official lanana assigned major, use it

Signed-off-by: Jeremy Katz <katzj@redhat.com>
20 years agoRevert unintended enabling of grant tables for netback/netfront
smh22@firebug.cl.cam.ac.uk [Tue, 2 Aug 2005 09:26:30 +0000 (09:26 +0000)]
Revert unintended enabling of grant tables for netback/netfront

20 years agoMerge with latest xen-unstable
djm@kirby.fc.hp.com [Mon, 1 Aug 2005 19:19:36 +0000 (12:19 -0700)]
Merge with latest xen-unstable

20 years agoFix the build on gcc-3.4.x (ia64 files)
djm@kirby.fc.hp.com [Mon, 1 Aug 2005 19:11:00 +0000 (12:11 -0700)]
Fix the build on gcc-3.4.x (ia64 files)
Signed-off-by: Arun Sharma <arun.sharma@intel.com>
20 years agoWorkaround new call to do_softirq in scrub_heap_pages in common code
djm@kirby.fc.hp.com [Mon, 1 Aug 2005 18:00:10 +0000 (11:00 -0700)]
Workaround new call to do_softirq in scrub_heap_pages in common code

20 years agocorrect wrong variables pass to the set_pte_at macro.
vh249@arcadians.cl.cam.ac.uk [Mon, 1 Aug 2005 16:00:17 +0000 (16:00 +0000)]
correct wrong variables pass to the set_pte_at macro.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoManual merge.
smh22@firebug.cl.cam.ac.uk [Mon, 1 Aug 2005 10:48:24 +0000 (10:48 +0000)]
Manual merge.

20 years agoMinor clean ups of time.c and partial build fixes for 2.4
smh22@firebug.cl.cam.ac.uk [Mon, 1 Aug 2005 10:45:07 +0000 (10:45 +0000)]
Minor clean ups of time.c and partial build fixes for 2.4

20 years agoPDB: convert ptrace.c and traps.c changes to local patches
ach61@arcadians.cl.cam.ac.uk [Mon, 1 Aug 2005 10:43:16 +0000 (10:43 +0000)]
PDB: convert ptrace.c and traps.c changes to local patches

20 years agoThis patch adds two macros for construction of the
kaf24@firebug.cl.cam.ac.uk [Mon, 1 Aug 2005 09:26:51 +0000 (09:26 +0000)]
This patch adds two macros for construction of the
frame_and_sectors field in blkif messages, to eliminate the
"magic shifts" in the blkif code.

It also increases the number of bits representing the sector
number within a page, from 3 to 5, to allow page sizes up to
16K (needed for IA64).

Tested to still work on x86.

Signed-off-by: Matthew Chapman <matthewc@hp.com>
20 years agoWhile merging my development tree with the latest hg tree I noticed that
kaf24@firebug.cl.cam.ac.uk [Mon, 1 Aug 2005 09:16:25 +0000 (09:16 +0000)]
While merging my development tree with the latest hg tree I noticed that
VMX support was broken. When you boot an unmodified Linux kernel it gets
stuck in the "Calibrating delay" loop.  The reason for this is that the
vmx code is repeatedly delivering timer interrupts.

Correcting the MILLISECS() and friends macros to return s_time_t instead
of ULL fixes this problem. The other changes are just to get rid of
redundant code and variables.

Signed-Off-By: Leendert van Doorn <leendert@watson.ibm.com>
20 years agoPDB: roll out changes to ptrace.c and traps.c
ach61@arcadians.cl.cam.ac.uk [Sun, 31 Jul 2005 21:45:52 +0000 (21:45 +0000)]
PDB: roll out changes to ptrace.c and traps.c

20 years agoAccomodate change to common alloc_domheap_page
djm@kirby.fc.hp.com [Fri, 29 Jul 2005 20:33:04 +0000 (13:33 -0700)]
Accomodate change to common alloc_domheap_page