xen.git
20 years agoRemove GPL from a couple of Xen public interfaces. There are
kaf24@firebug.cl.cam.ac.uk [Sat, 10 Dec 2005 15:07:03 +0000 (16:07 +0100)]
Remove GPL from a couple of Xen public interfaces. There are
a few more files to be done, after consulting their authors.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoUpdate "make uninstall" to reflect new files.
kaf24@firebug.cl.cam.ac.uk [Sat, 10 Dec 2005 14:57:11 +0000 (15:57 +0100)]
Update "make uninstall" to reflect new files.

Signed-off-by: Natasha Jarymowycz <natasha@us.ibm.com>
20 years agoIn all cases in dom0_ops.c, return EFAULT if copy_user
kaf24@firebug.cl.cam.ac.uk [Sat, 10 Dec 2005 14:54:27 +0000 (15:54 +0100)]
In all cases in dom0_ops.c, return EFAULT if copy_user
fails.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
20 years agoAllows a 'make prep-kernels' which sets up the kernel directories, allowing
vhanquez@kneesa.uk.xensource.com [Sat, 10 Dec 2005 00:11:33 +0000 (00:11 +0000)]
Allows a 'make prep-kernels' which sets up the kernel directories, allowing
certain types of test automation to be done more easily.

Signed-off-by: Paul Larson <pl@us.ibm.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 11:05:06 +0000 (11:05 +0000)]
Merged.

20 years agoAdd initial VMX support to xm-test:
stekloff@elm3b216.beaverton.ibm.com [Fri, 9 Dec 2005 11:04:55 +0000 (11:04 +0000)]
Add initial VMX support to xm-test:

1) Added script create_disk_image to make full virt image
2) Changed XenDomain.py to build separate XmTestDomain object
   for vmx guests (this structure will need to be cleaned up)
3) Changed ramdisk Makefile.am to create disk.img if vmx configured

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:51:35 +0000 (10:51 +0000)]
Merged.

20 years agoUse new do_without_error function to squelch the error when closing down the
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:51:20 +0000 (10:51 +0000)]
Use new do_without_error function to squelch the error when closing down the
interface.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoUse new do_without_error function to squelch the errors when closing down an
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:50:49 +0000 (10:50 +0000)]
Use new do_without_error function to squelch the errors when closing down an
interface.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded do_without_error function, and squelch the error from mkdir when trying
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:50:26 +0000 (10:50 +0000)]
Added do_without_error function, and squelch the error from mkdir when trying
to create the lock directory.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdd one more fork to the Xend start-up process, this time with the parent
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:49:29 +0000 (10:49 +0000)]
Add one more fork to the Xend start-up process, this time with the parent
staying alive to watch the child.  If the child crashes, then the parent
starts another one in its place.  This provides robustness in the face of
crashes like one currently tracked as bug #411.

Tidy up chunks of SrvDaemon.py on the way past.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoTwo domUs, ping from one to the other
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:40:41 +0000 (10:40 +0000)]
Two domUs, ping from one to the other

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
20 years agoSingle domU, ping to dom0.
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:40:28 +0000 (10:40 +0000)]
Single domU, ping to dom0.

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
20 years agoSingle domU, ping loopback and the local eth0.
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:40:00 +0000 (10:40 +0000)]
Single domU, ping loopback and the local eth0.

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
20 years agoSome network test utilities for the library:
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:39:31 +0000 (10:39 +0000)]
Some network test utilities for the library:

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
20 years agoMakefile plumbing for new network tests.
emellor@leeni.uk.xensource.com [Fri, 9 Dec 2005 10:39:16 +0000 (10:39 +0000)]
Makefile plumbing for new network tests.

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
20 years agoFix blkback to build on ia64.
kaf24@firebug.cl.cam.ac.uk [Fri, 9 Dec 2005 08:29:52 +0000 (09:29 +0100)]
Fix blkback to build on ia64.

Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>
20 years ago The attached patch makes a couple of changes to the ACM security tools
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 18:21:05 +0000 (18:21 +0000)]
  The attached patch makes a couple of changes to the ACM security tools
and installation. Particularly it does the following:
- the Makefile installs the ACM security-related tools
into /etc/xen/acm-security
- improves and cleans up some of the tools
- updates the documentation

Signed-off-by: Tom Lendacky <toml@us.ibm.com>
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
20 years agoThe attached patch uncouples the device id used inside XenD from the
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 18:19:24 +0000 (18:19 +0000)]
The attached patch uncouples the device id used inside XenD from the
instance number of the vTPM.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
20 years agoLooks like someone merged and moved the pincpu operation to after the
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 18:17:59 +0000 (18:17 +0000)]
Looks like someone merged and moved the pincpu operation to after the
memory reservation and didn't read the comments I left in the code:

# repin domain vcpus if a restricted cpus list is provided
# this is done prior to memory allocation to aide in memory
# distribution for NUMA systems.

This patch moves the pincpu op back to where it was.

Signed-off-by: Ryan Harper <ryanh@us.ibm.com>
20 years agoDetach Xend from terminal, courtesy of Horms <horms@verge.net.au>.
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 16:17:53 +0000 (16:17 +0000)]
Detach Xend from terminal, courtesy of Horms <horms@verge.net.au>.

* For setsid to effectively detach a process from the terminal,
  the following needs to occur:

    fork ()   Return control to the shell
    setsid () New session with no controlling terminal
    fork ()   The session leader (parent) exits and thus
              the resulting child process can never regain the terminal

  This patch adds the second fork

* The call to self.daemonize(), which now forks, is moved to
  run before self.tracing(), as now that it actually disconnects
  from the terminal, and thus the prevailing process, the trace
  loses the processes created in self.run().

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFlush the addresses from the new eth0 before trying to add the new address to
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 16:11:48 +0000 (16:11 +0000)]
Flush the addresses from the new eth0 before trying to add the new address to
it, when using ip on the kernel command line and not using the installed ifup /
ifdown.  This stops the script bailing out if there is already an address
attached to that device, which was leaving the device without routing (since
this has already been removed).

Fix the stop action under the same circumstances.  No attempt was made to
restore the kernel command line details to the interfaces if the ifdown failed
in that case.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAllocate separate vm areas for rx and tx rings in netback
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 15:58:41 +0000 (16:58 +0100)]
Allocate separate vm areas for rx and tx rings in netback
driver as part of preparation for ia64 support.

Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>
20 years agoFix ip_of to cope with interfaces with multiple addresses, courtesy of
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 15:54:35 +0000 (15:54 +0000)]
Fix ip_of to cope with interfaces with multiple addresses, courtesy of
Adam Heath <doogie@brainfood.com>.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoSupport CFQ scheduling of guest block requests by creating
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 15:53:53 +0000 (16:53 +0100)]
Support CFQ scheduling of guest block requests by creating
a kernel thread per blkif connection. General cleanup work
in blkback driver.

Signed-off-by: Gerd Knorr <kraxel@suse.de>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 15:04:41 +0000 (15:04 +0000)]
Merged.

20 years agoStrip huge piles of cruft from the connection infrastructure. We now actually
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 15:04:31 +0000 (15:04 +0000)]
Strip huge piles of cruft from the connection infrastructure.  We now actually
block inside accept rather than using select to poll and then calling accept
regardless of the outcome of the select call, and then failing because the
socket is non-blocking.

SocketClientConnection, SocketConnector, TCPClientConnection, TCPConnector,
connectTCP, UnixClientConnection, UnixConnector, connectUnix have gone.

loseConnection and stopListening and closeSocket (where they are needed) are
now called close.  startListening is now called listen.

Closes bug #379.

Relieves a weight from the shoulders of the universe.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMake balloon driver arch-neutral.
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:33:10 +0000 (15:33 +0100)]
Make balloon driver arch-neutral.

Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>
20 years agoMove xm destroy handling out of crufty destroy.py and into main.py, where it
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 14:30:15 +0000 (14:30 +0000)]
Move xm destroy handling out of crufty destroy.py and into main.py, where it
can share the better argument parsing.

Closes bug #443.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix start_time parsing on restart -- it's a float, not an int.
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 14:28:53 +0000 (14:28 +0000)]
Fix start_time parsing on restart -- it's a float, not an int.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoA small VMX code cleanup.
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:25:44 +0000 (15:25 +0100)]
A small VMX code cleanup.

Signed-off-by: Eddie Dong <eddie.dong@intel.com>
20 years agoMerge in the newer Xenbus implementation from Linux to the Mini-OS. The new
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:24:02 +0000 (15:24 +0100)]
Merge in the newer Xenbus implementation from Linux to the Mini-OS. The new
version compiles and starts up, but I'm not really sure how to test the new
xenbus implementation.

* Added unbind_evtchn
* Copied parts of the Linux spinlock implementation to make the changes to
  xenbus compared to Linux smaller. Also added a dummy rwsem implementation.
* Updated the xenbus-files

Signed-off-by: Simon Kagstrom <simon.kagstrom@bth.se>
20 years agoDon't unshadow when the shadow l2 is the current used shadow table.
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:21:36 +0000 (15:21 +0100)]
Don't unshadow when the shadow l2 is the current used shadow table.
Otherwise, when destroying the domain, if the current sl2 is unpinned
then the shadow page will be freed without accquire the shadow_lock.
This may also improve the performance due to avoiding uncessary
unshadow.

Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
20 years agoFixed a get/put_page mismatch for guest L2 pages. Sometimes, when you
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:18:13 +0000 (15:18 +0100)]
Fixed a get/put_page mismatch for guest L2 pages.  Sometimes, when you
destroy an OS when it is in real mode, there is an extra put_page in
domain_relinquish_resource.  This causes the page be freed with
non-zero type count.  With ASSERT enabled, hypervisor will crash; with
debug=n, on some machines, the system will hang in
alloc_domheap_pages, due to type_info being a union with cpumask and
not equal to 0.

Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
20 years agoImprove VMX guest TSC accuracy when guests are
kaf24@firebug.cl.cam.ac.uk [Thu, 8 Dec 2005 14:13:38 +0000 (15:13 +0100)]
Improve VMX guest TSC accuracy when guests are
competing for the same physical CPU.

Signed-off-by: Eddie Dong <eddie.dong@intel.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 12:14:33 +0000 (12:14 +0000)]
Merged.

20 years agoLoop retrying when ballooning out, even when the dom0-min-mem setting means
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 12:13:06 +0000 (12:13 +0000)]
Loop retrying when ballooning out, even when the dom0-min-mem setting means
that there is not sufficient memory available at the moment.  Memory may be
about to be freed up by the system after a domain destruction (i.e. the memory
is being scrubbed asynchronously, and will be released soon).

Closes bug #407, bug #429.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRead start_time and image from the store on restart. Closes bugs #441 and
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 12:10:22 +0000 (12:10 +0000)]
Read start_time and image from the store on restart.  Closes bugs #441 and
#442.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRefuse to save dom0, and move the check that refuses to migrate dom0 from
emellor@leeni.uk.xensource.com [Thu, 8 Dec 2005 12:08:59 +0000 (12:08 +0000)]
Refuse to save dom0, and move the check that refuses to migrate dom0 from
migrate.py into XendDomain.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoreally fix mkpatches to work correctly.
vhanquez@kneesa.uk.xensource.com [Thu, 8 Dec 2005 08:57:55 +0000 (08:57 +0000)]
really fix mkpatches to work correctly.
unfortunately, add some code duplication in mkpatches area.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agofix mkpatches for creating correctly linux-2.6-xen.patch
vhanquez@kneesa.uk.xensource.com [Thu, 8 Dec 2005 08:09:09 +0000 (08:09 +0000)]
fix mkpatches for creating correctly linux-2.6-xen.patch

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoallow a port to be specified for migration.
vhanquez@kneesa.uk.xensource.com [Thu, 8 Dec 2005 07:11:09 +0000 (07:11 +0000)]
allow a port to be specified for migration.

Signed-off-by: Vincent Hanquez <vincent@xensource.com>
20 years agoUpdated info on Xend.
smh22@firebug.cl.cam.ac.uk [Wed, 7 Dec 2005 17:59:33 +0000 (18:59 +0100)]
Updated info on Xend.

Signed-off-by: Nivedita Singhvi <niv@us.ibm.com>
Signed-off-by: Robb Romans <FMJ@us.ibm.com>
Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoDon't try and print out the info returned by the server after a restore -- there
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 16:26:07 +0000 (16:26 +0000)]
Don't try and print out the info returned by the server after a restore -- there
isn't any.  Don't try and unpause the restored domain either -- XendCheckpoint
does that.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoUse xm create's --dryrun option to get the name from the configuration file,
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 16:25:13 +0000 (16:25 +0000)]
Use xm create's --dryrun option to get the name from the configuration file,
rather than the eversoskanky approach taken previously (use grep, and replace
any occurrences of %d with [0-9]* !)

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMake the xm-test ramdisk makefile correct the permissions on the rcS script
danms@us.ibm.com [Wed, 7 Dec 2005 15:50:59 +0000 (15:50 +0000)]
Make the xm-test ramdisk makefile correct the permissions on the rcS script

Signed-off-by: Dan Smith <danms@us.ibm.com>
20 years agoFix the startup behaviour when the tdb file exists. We were attempting to
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 15:44:38 +0000 (15:44 +0000)]
Fix the startup behaviour when the tdb file exists.  We were attempting to
remove and recreate the /local directory, but the manual_node call only works
if the tdb is empty -- we must use create_node, otherwise the store is
corrupted.

It's not clear to me that the other manual_node calls are safe, but I have only
made this smaller change for the sake of stability.  These other calls should
be reviewed.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix usleep definition -- we would be sleeping 1000 times too long if the usleep
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 15:41:41 +0000 (15:41 +0000)]
Fix usleep definition -- we would be sleeping 1000 times too long if the usleep
command were not available elsewhere on the machine (on my Debian boxes, for
example).

Sleep for a configurable amount of time between xm create calls (by default,
5 seconds).  This spaces out the booting of the new domains, meaning that they
should not thrash the disk so much.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoTidy up use of environment variables in xenbus-hotplug interaction.
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:57:26 +0000 (11:57 +0000)]
Tidy up use of environment variables in xenbus-hotplug interaction.

I've changed this to only write "backend" (xenbus_backend.root) into
XENBUS_BASE_PATH.  After appending XENBUS_TYPE (also exported) you can
construct the "base path" easily in the scripts.  I've completely
removed writing XENBUS_FRONTEND_ID in favour of reading it from the
store at "$XENBUS_PATH/frontend-id".

Avoid re-reading the store over and over for the frontend domain's vm-path.

The patch also fixes a memory leak in backend_bus_id where we
leak the memory referenced by frontend if the function succeeds.

This work is by Christian Limpach.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:51:31 +0000 (11:51 +0000)]
Merged.

20 years agoChange configuration handling wrt to the vcpus entry, so that we only see a
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:50:55 +0000 (11:50 +0000)]
Change configuration handling wrt to the vcpus entry, so that we only see a
vcpus entry in the image section when using VMX, only add VMX options at all
if the builder is vmx, and issue a warning when overriding the global vcpus
setting with the one from the image.

This removes the duplicated vcpus entries in the domain's sxpr.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove timeout on select -- there is no need for it, since we don't do anything
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:47:46 +0000 (11:47 +0000)]
Remove timeout on select -- there is no need for it, since we don't do anything
in the main loop except service I/O.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:44:02 +0000 (11:44 +0000)]
Merged.

20 years agoFurther fix to same_vm check. This fixes intermittent failure of live
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 11:43:43 +0000 (11:43 +0000)]
Further fix to same_vm check.  This fixes intermittent failure of live
localhost migration.

return [ test ] is not valid, simply [ test ] at the end of the function is the
correct syntax.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix printf of ssize_t that broke 64-bit build.
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 10:54:57 +0000 (10:54 +0000)]
Fix printf of ssize_t that broke 64-bit build.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoSquelch error messages from calls when failure is handled gracefully.
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 10:53:47 +0000 (10:53 +0000)]
Squelch error messages from calls when failure is handled gracefully.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoRemove one of the multiple occurrences of vcpus in the sxpr. This value is
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 10:51:55 +0000 (10:51 +0000)]
Remove one of the multiple occurrences of vcpus in the sxpr.  This value is
in ROUNDTRIPPING_CONFIG_ENTRIES, so there is no need for it to be mentioned
explicitly in sxpr() too.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoxm-test report fails because IA64's /cpu/procinfo format is different from
emellor@leeni.uk.xensource.com [Wed, 7 Dec 2005 10:27:03 +0000 (10:27 +0000)]
xm-test report fails because IA64's /cpu/procinfo format is different from
x86's.  Here's a small patch to fix the problem.

Signed-off-by: Matsumoto <n_matumoto@soft.fujitsu.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Tue, 6 Dec 2005 17:11:04 +0000 (17:11 +0000)]
Merged.

20 years agoAdd a "consumed" parameter to the buffer declaration, and use this to advance
emellor@leeni.uk.xensource.com [Tue, 6 Dec 2005 17:10:59 +0000 (17:10 +0000)]
Add a "consumed" parameter to the buffer declaration, and use this to advance
through the buffer when the data are read by console clients.  This avoids a
costly memmove with every read from the console.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoPull nmi/traps changes from Linux 2.6.14.
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 16:48:57 +0000 (17:48 +0100)]
Pull nmi/traps changes from Linux 2.6.14.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Tue, 6 Dec 2005 16:40:50 +0000 (16:40 +0000)]
Merged.

20 years agoMake the master fd belonging to the pty pair non-blocking. Writes were blocking
emellor@leeni.uk.xensource.com [Tue, 6 Dec 2005 16:40:43 +0000 (16:40 +0000)]
Make the master fd belonging to the pty pair non-blocking.  Writes were blocking
if the console client disconnected before all the data had been written to the
pty, even though the select call had indicated that the fd was previously
writable.  Reads have been observed exhibiting similar behaviour too.

Closes bug #380.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoThe Mini-OS build once again broke, this time since vcpu_data became vcpu_info
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 15:05:31 +0000 (16:05 +0100)]
The Mini-OS build once again broke, this time since vcpu_data became vcpu_info
and that XEN_VER in the __xen_guest section changed slightly. I've modified
the mini-os so that it at least compiles and boots again.

Signed-off-by: Simon Kagstrom <simon.kagstrom@bth.se>
20 years agoAdd map_domain_pages/unmap_domain_pages to map contiguous
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 14:52:51 +0000 (15:52 +0100)]
Add map_domain_pages/unmap_domain_pages to map contiguous
multi-franme extents of domain memory.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix the request size for a full page-sized skbuff in netfront
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 14:52:15 +0000 (15:52 +0100)]
Fix the request size for a full page-sized skbuff in netfront
driver (must allow 16 bytes for dev_alloc_skb headroom).

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoFix netfront to accept received packets at a wider range of
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 14:02:54 +0000 (15:02 +0100)]
Fix netfront to accept received packets at a wider range of
offsets within a memory page. Also fix, skb realloc-and-copy
when copying really is required (we were forgetting whether
the packet data is known valid, and checksumming packets that
do not have the csum field filled in).

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoPre-decrement rather than post-decrement in shadow32.c.
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 11:49:05 +0000 (12:49 +0100)]
Pre-decrement rather than post-decrement in shadow32.c.
Already fixed in shadow.c.

Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
20 years agoUpdate xen cap to reflect 32 bit VMX guest support on x86_64
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 11:46:43 +0000 (12:46 +0100)]
Update xen cap to reflect 32 bit VMX guest support on x86_64

Signed-off-by: Xin Li <xin.b.li@intel.com>
20 years agoFix vtpm support in xend.
kaf24@firebug.cl.cam.ac.uk [Tue, 6 Dec 2005 11:45:30 +0000 (12:45 +0100)]
Fix vtpm support in xend.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
20 years agoOutput an explicit newline before the "REASON:" string so the report script
dan@guaranine.beaverton.ibm.com [Mon, 5 Dec 2005 15:36:11 +0000 (15:36 +0000)]
Output an explicit newline before the "REASON:" string so the report script
can find it, even when a domain outputs a non-newline-terminated line.
Signed-off-by: Dan Smith <danms@us.ibm.com>
20 years agoMention udev, clean up xm help section, mention network-nat and network-route
emellor@leeni.uk.xensource.com [Mon, 5 Dec 2005 15:30:10 +0000 (15:30 +0000)]
Mention udev, clean up xm help section, mention network-nat and network-route
scripts and the fact that network is now called network-bridge.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoAdded tag RELEASE-3.0.0 for changeset c8fdb0caa77b429cf47f9707926e83947778cb48
iap10@freefall.cl.cam.ac.uk [Mon, 5 Dec 2005 13:02:01 +0000 (14:02 +0100)]
Added tag RELEASE-3.0.0 for changeset c8fdb0caa77b429cf47f9707926e83947778cb48

20 years agoUpdated docs for Xen 3.0.
smh22@firebug.cl.cam.ac.uk [Mon, 5 Dec 2005 12:39:26 +0000 (13:39 +0100)]
Updated docs for Xen 3.0.

Signed-off-by: Steven Hand <steven@xensource.com>
20 years agoTemporarily remove NetBSD and FreeBSD sparse trees to avoid user confusion.
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 19:12:00 +0000 (20:12 +0100)]
Temporarily remove NetBSD and FreeBSD sparse trees to avoid user confusion.
The NetBSD and FreeBSD currently don't build against the final Xen 3.0 API.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoMerge
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 17:24:24 +0000 (18:24 +0100)]
Merge

20 years agoPull requests from blkfront into blkback a bit more
kaf24@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 17:20:41 +0000 (18:20 +0100)]
Pull requests from blkfront into blkback a bit more
aggressively, to keep the scheduling pipeline filled.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoUpdate version number for Xen 3.0.0 release.
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 17:17:33 +0000 (18:17 +0100)]
Update version number for Xen 3.0.0 release.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoMake version strings consistent across all interfaces.
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 17:16:33 +0000 (18:16 +0100)]
Make version strings consistent across all interfaces.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoUpdate -xen config file to include newgen megaraid and qla scsi drivers.
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 17:12:18 +0000 (18:12 +0100)]
Update -xen config file to include newgen megaraid and qla scsi drivers.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoRename set-vcpus to vcpu-set (match docs) and update short help page.
cl349@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 16:29:46 +0000 (17:29 +0100)]
Rename set-vcpus to vcpu-set (match docs) and update short help page.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoBump blktap and dom0op version numbers.
kaf24@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 11:06:38 +0000 (12:06 +0100)]
Bump blktap and dom0op version numbers.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoImage.py should send 2 new arguments to qemu-dm for setting QEMU window
kaf24@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 10:57:32 +0000 (11:57 +0100)]
Image.py should send 2 new arguments to qemu-dm for setting QEMU window
Title. 1 is const string "-domain-name", the other is the string of
domain name. At present imagy.py combines these two strings to 1. So
Qemu will fail to recognize it and fail to start.

Signed-off-by: Yongkang You <yongkang.you@intel.com>
20 years agoAdd a comment to tricky corner case in evtchn_close().
kaf24@firebug.cl.cam.ac.uk [Sun, 4 Dec 2005 10:54:04 +0000 (11:54 +0100)]
Add a comment to tricky corner case in evtchn_close().

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerge docs
rread@ubuntu.eng.hq.xensource.com [Sun, 4 Dec 2005 00:52:38 +0000 (17:52 -0700)]
Merge docs

20 years agoReadme typos
iap10@freefall.cl.cam.ac.uk [Sun, 4 Dec 2005 00:34:25 +0000 (01:34 +0100)]
Readme typos
Signed-off-by: ian@xensource.com
20 years agoFix same_vm test by adding missing '$'.
cl349@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 23:51:15 +0000 (00:51 +0100)]
Fix same_vm test by adding missing '$'.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoUse test ([) instead of expr to compare integers.
cl349@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 23:49:51 +0000 (00:49 +0100)]
Use test ([) instead of expr to compare integers.
expr $int1 '<' $int2 seems to be broken on some shells.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoFix request-notification holdoff in blkback. New code is
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 20:53:01 +0000 (21:53 +0100)]
Fix request-notification holdoff in blkback. New code is
more correct and clearer in intent.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoRemove scary warning message about 'breaking irq affinity'
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 20:17:11 +0000 (21:17 +0100)]
Remove scary warning message about 'breaking irq affinity'
when hot-unplugging cpus. It is a harmless side-effect of
running the irqbalance daemon in user space.

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoMerge
iap10@freefall.cl.cam.ac.uk [Sat, 3 Dec 2005 19:16:04 +0000 (20:16 +0100)]
Merge

20 years agoUpdate xend-config.sxp so that dom0-min-mem is 196MB by default, as this is rather...
iap10@freefall.cl.cam.ac.uk [Sat, 3 Dec 2005 19:05:39 +0000 (20:05 +0100)]
Update xend-config.sxp so that dom0-min-mem is 196MB by default, as this is rather more user friendly.
Also do some minor fixes to the README
Signed-off-by: ian@xensource.com
20 years agoAlways initialise a swiotlb for domain0. Make it a small
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 18:09:19 +0000 (19:09 +0100)]
Always initialise a swiotlb for domain0. Make it a small
aperture (2MB) for small systems (< 2GB).

Signed-off-by: Keir Fraser <keir@xensource.com>
20 years agoEnable multiple NICs for vmx domain by configuration file
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 16:59:37 +0000 (17:59 +0100)]
Enable multiple NICs for vmx domain by configuration file

Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
20 years agoSet the device model windows title as the domain name
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 16:58:31 +0000 (17:58 +0100)]
Set the device model windows title as the domain name
This helps when creating multile vmx domain.

Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
20 years agoFixes unmodified 32-bit guest support on the x86_64 Xen.
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 16:41:16 +0000 (17:41 +0100)]
Fixes unmodified 32-bit guest support on the x86_64 Xen.

Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
20 years agoFix block backend teardown when the device didn't setup completely.
cl349@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 12:36:22 +0000 (13:36 +0100)]
Fix block backend teardown when the device didn't setup completely.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
20 years agoMerged.
emellor@leeni.uk.xensource.com [Sat, 3 Dec 2005 12:21:52 +0000 (12:21 +0000)]
Merged.

20 years agoSet hostname in DHCP config from the guest name, and if more than one interface
emellor@leeni.uk.xensource.com [Sat, 3 Dec 2005 12:21:27 +0000 (12:21 +0000)]
Set hostname in DHCP config from the guest name, and if more than one interface
is specified, from guest-2, guest-3 etc.  Have the DHCP server send the
hostname to the guest.

Added a lock around DHCP config file frobbing, to avoid racing creation vs
deletion.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
20 years agoFix blkback request notification holdoff. req_cons must be
kaf24@firebug.cl.cam.ac.uk [Sat, 3 Dec 2005 10:41:49 +0000 (11:41 +0100)]
Fix blkback request notification holdoff. req_cons must be
updated before make_response() is called. This fixes attaching
ramdisk-backed vbds to a blkfront.

Signed-off-by: Keir Fraser <keir@xensource.com>