kaf24@firebug.cl.cam.ac.uk [Thu, 17 Nov 2005 10:47:49 +0000 (11:47 +0100)]
Fix IOCTL_PRIVCMD_GET_MACH2PHYS_MFNS to work properly for
4MB superpages (32-bit non-pae).
Signed-off-by: Keir Fraser <keir@xensource.com>
smh22@firebug.cl.cam.ac.uk [Thu, 17 Nov 2005 09:10:52 +0000 (10:10 +0100)]
Fix pinning logic on restore.
Signed-off-by: Steven Hand <steven@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 19:33:23 +0000 (20:33 +0100)]
Merged.
dan@guaranine.beaverton.ibm.com [Wed, 16 Nov 2005 19:33:12 +0000 (20:33 +0100)]
Test for creating a domain with too little memory.
dan@guaranine.beaverton.ibm.com [Wed, 16 Nov 2005 19:32:36 +0000 (20:32 +0100)]
Make 01_reboot test more robust.
We need to attach a console before rebooting to make sure the domain did
actually boot in the first place. Also, use a better method for pulling out
the uptime string.
smh22@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 17:41:43 +0000 (18:41 +0100)]
One more try...
Signed-off-by: Steven Hand <steven@xensource.com>
smh22@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 17:28:09 +0000 (18:28 +0100)]
Fix annoying error introduced during cleanup. Oops.
Signed-off-by: Steven Hand <steven@xensource.com>
smh22@tempest.cl.cam.ac.uk [Wed, 16 Nov 2005 17:03:48 +0000 (18:03 +0100)]
Trivial fixes for 64bit and xend.
Signed-off-by: Steven Hand <steven@xensource.com>
smh22@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 16:45:03 +0000 (17:45 +0100)]
Many fixes for save/restore and related areas for PAE in particular. Now
should be able to save/restore successfully on machines with up to 16GB
and any size of guest.
Signed-off-by: Steven Hand <steven@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 14:50:36 +0000 (15:50 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 14:32:56 +0000 (15:32 +0100)]
Watch for changes in the /vm section of the store, so that we update our
internal state if the store is configured by an external tool.
We react properly to changes in name and on_{reboot,poweroff,crash}. Reacting
correctly to vcpus, vcpu_avail, memory, and maxmem will come soon (those code
aspects are already under review). cpu_weight and bootloader to be considered.
Removed unused and semantically invalid method XendDomainInfo.setDomid.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 14:27:59 +0000 (15:27 +0100)]
Added an unwatch method, so that XendDomainInfo can unregister the watch from
outside the watch handler.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 14:27:11 +0000 (15:27 +0100)]
Move the registration of the @releaseDomain watch so that we avoid a watch
between the watch firing and the first call to refresh.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 14:25:22 +0000 (15:25 +0100)]
Downgrade the failure to change the iptables settings from fatal to an error.
Even though we are checking for a working iptables installation, it is still
possible that the user does not have the PHYSDEV matcher installed, so just
issue a diagnostic instead.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 12:38:16 +0000 (13:38 +0100)]
Merged.
jmb@artemis.home.pimb.org [Wed, 16 Nov 2005 12:34:52 +0000 (13:34 +0100)]
Dom0 might not be called Domain-0, so ask for the name in shutdown.py, instead of hardcoding it
Signed-off-by: Jody Belka <knew (at) pimb (dot) org>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 11:44:12 +0000 (12:44 +0100)]
Rename XENVER_parameters to slightly more informative
XENVER_platform_parameters.
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 11:37:29 +0000 (12:37 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 11:36:47 +0000 (12:36 +0100)]
Review checks for Error string when checking that bad arguments are handled
correctly. Many checks were looking using output.find("Error") > 1, which is
incorrect, as we do not guarantee that the word "Error" will be first in the
output (never mind the fact that strings are indexed from 0).
All these checks have been changed to compare against -1. In particular, this
should fix the failure of create_noparm_neg and create_badparm_neg on a machine
with the /etc/xen/xmdefconfig file in place, as xm create issues the
Using config file "/etc/xen/xmdefconfig" diagnostic before reporting the error.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 11:23:17 +0000 (12:23 +0100)]
Fix up start_info_t magic string to match the format of
XENVER_capabilities strings.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:41:14 +0000 (11:41 +0100)]
This fixes the following scenerio:
1) gmfn X is L2 table, and a shadow for it is set.
2) A write to page X, and the same page would be L1 table
also.(Something like the linear page table) But the shadow for L1 table
is not set.
The code path will be shadow_fault -> l1pte_write_fault. Because X is L2
table, so shadow_mark_va_out_of_sync is called , this function will
allocate a OO entry for this table, and set the writable_pl1e to -1.
Also it alloc a L1 shadow for X through shadow_map_l1_into_current_l2.
When allocating, free_out_of_sync_state will be called to free all OO
entry. However, since the OO entry allocated on
shadow_mark_va_out_of_sync has the writable_pl1e to -1, this OO entry
will never be released.
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:39:38 +0000 (11:39 +0100)]
Change ISA_IO_BASE to fix_to_virt(FIX_ISAMAP_BEGIN). It used to be
PAGE_OFFSET, which was unaltered from vanilla Linux 2.6.12.6 kernels
and caused isa_memcpy_fromio() to not read physical memory from
0xf0000 - 0xfffff correctly.
http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=393
Signed-off-by: Andrew D. Ball <aball@us.ibm.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:35:26 +0000 (11:35 +0100)]
Remove unnecessary cast of kmalloc() result.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:34:54 +0000 (11:34 +0100)]
Add definitions of some swiotlb symbols to pci-dma.c.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:34:19 +0000 (11:34 +0100)]
/dev/mem write function should use copy_from_user, not copy_to_user.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:33:13 +0000 (11:33 +0100)]
merge.
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:32:19 +0000 (11:32 +0100)]
Correctly export xencons functions to avoid ugly 'extern'
declarations.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 10:32:14 +0000 (11:32 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Wed, 16 Nov 2005 10:31:57 +0000 (11:31 +0100)]
The following patch removes every (hopefully) reference to network-limit.
Upon further investigation it sets Device parameters that don't exist any
longer, thus causing exceptions. As it looks like this is completely
obsolete, and no one is actively working on it, we should just rip it out to
end the confusion.
Signed-off-by: Sean Dague <sean@dague.net>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:30:32 +0000 (11:30 +0100)]
Fix function that shouldn't be globally defined.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:29:52 +0000 (11:29 +0100)]
Trivial coding style fix.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:28:12 +0000 (11:28 +0100)]
Add function declarations, and ensure we prevent mismatch
of function definition vs. declaration.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Wed, 16 Nov 2005 10:27:19 +0000 (11:27 +0100)]
Replace 0 with NULL when assigning pointer fields/parameters.
Signed-off-by: Vincent Hanquez <vincent@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 18:46:48 +0000 (19:46 +0100)]
Fix 64-bit build that I broke by making acpi table mapping
fixup dependent on CONFIG_XEN_X86 (that only xen/i386
defines).
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 18:19:02 +0000 (19:19 +0100)]
Make watchStart and watchMain global functions rather than classmethods,
meaning that we no longer need to prefix all the field accesses with cls.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 18:09:30 +0000 (19:09 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 18:08:11 +0000 (19:08 +0100)]
Have a watch callback return 0 or 1 depending upon whether it would like to
continue to receive watches. This means that it is not necessary to probe
around in the xswatch internals to unregister a watch.
Tidy up the hotplug watch handling, moving the nested function out to a
separate place (I don't think that this was a problem in the end, but it was
making me nervous as I was debugging the recent race condition, and I find it
clearer out of the class in any case. Pass an integer code representing
hotplug status, once it has been parsed from the store, as there are now a few
different places we can detect failure, and it's cleaner to pass a code from
those places.
Small tidy up to XendDomain, removing the unused releaseDomain field.
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 17:47:39 +0000 (18:47 +0100)]
Remove the handling of TypeError inside gather. It is not thrown by int()
when given a non-convertible value, as claimed by the comment, as the exception
ValueError is thrown in that situation. TypeError is only thrown on int(None),
which cannot actually take place in this code, because of the explicit check
for None.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 17:43:28 +0000 (18:43 +0100)]
Ensure we initialise the cpu_present_map before
topology_init() is called. In latest Linux kernels it
iterates over cpu_present_map rather than cpu_possible_map.
Also, save/restore should really iterate over
possible cpus, not present ones (not that it really matters,
but it's a small cleanup).
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 16:30:55 +0000 (17:30 +0100)]
This patch fixes dom0 SMP vcpu hotplug. Currently, domains without
config files (e.g. dom0) don't set info['vcpus'] correctly resulting in
incorrect cpu availablity values in the store. This prevents hotplug
from functioning.
Signed-off-by: Ryan Harper <ryanh@us.ibm.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 16:28:05 +0000 (17:28 +0100)]
This patch fixes a number of vcpu related issues.
1. xm vcpu-list now shows info for all vcpus by using
info['max_vcpu_id']+1 as end of the range to query
vcpu_info.
2. Add new vcpu fields, online_vcpus, max_vcpu_id to
XendDomainInfo.sxpr()
3. Remove filter_cpump() which isn't needed now that
the per vcpu cpumap field is correct.
4. Update xm/main.py to use online_vcpus as the number
vcpus to display in list.
Signed-off-by: Ryan Harper <ryanh@us.ibm.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 16:22:42 +0000 (17:22 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 16:22:04 +0000 (17:22 +0100)]
Fix race between xspy_read_watch and xspy_watch, by placing the watch in the
xshandle's list of watches *before* registering the watch with Xend.
Closes bug #392.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 16:02:37 +0000 (17:02 +0100)]
acpi change is Xen/x86 specific, so make it dependent on
CONFIG_XEN_X86.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 15:57:02 +0000 (16:57 +0100)]
Upgrade Xen's msr.h to include definitions from Linux 2.6.14.
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 15:24:31 +0000 (16:24 +0100)]
Merged.
danms@us.ibm.com [Tue, 15 Nov 2005 15:07:44 +0000 (16:07 +0100)]
Reduce the time required to start xm-test domains by allowing the console
to retry connecting to the DomU multiple times. A polling approach instead
of a wait-and-hope-it-works approach.
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:56:47 +0000 (15:56 +0100)]
Have the vif scripts expect the offline event, not the remove event. With
the remove event, the script environment variable has disappeared, so we must
hook on the offline event instead. This fix was put into xen-backend.rules,
but not the vif scripts, so they were out of sync, and the interfaces were not
being properly closed down.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:54:51 +0000 (15:54 +0100)]
Added link_exists function, for neatness. Check for the existence of the pdev
before issuing the diagnostic about the missing vdev. Having a pdev but no
vdev is a strong indicator that the network-bridge script has run once already,
and so there is nothing to worry about. Having the vdev and no corresponding
pdev is still diagnosed.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:44:37 +0000 (15:44 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:36:42 +0000 (15:36 +0100)]
This patch is largely stylistic, and gets rid of StuddlyCaps formatting in
all the param variables, make consistant use of bold and italics in the the
man pages to make it easier to read, and adds a few minor content additions.
It makes actually reading the man page a lot easier.
Signed-off-by: Sean Dague <sean@dague.net>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:27:48 +0000 (15:27 +0100)]
This patch adds more content around the scheduler portions of the xm.1 man
page. Most of this content came from the xensource wiki or the sedf doc in
the Xen tree. It still needs more work, but it is a good first start.
Signed-off-by: Sean Dague <sean@dague.net>
emellor@leeni.uk.xensource.com [Tue, 15 Nov 2005 14:26:37 +0000 (15:26 +0100)]
This patch is another in the continued effort to flesh out the man pages, it
adds pretty reasonable text to most of the code xen domain and host commands
(device and scheduler will come in the future).
Signed-off-by: Sean Dague <sean@dague.net>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 14:11:24 +0000 (15:11 +0100)]
Update hgignore list.
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 14:09:58 +0000 (15:09 +0100)]
The new userland monitoring tool, XenMon.
Signed-off-by: Rob Gardner <rob.gardner@hp.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 13:59:59 +0000 (14:59 +0100)]
Extra tracing in Xen to support XenMon.
Signed-off-by: Rob Gardner <rob.gardner@hp.com>
I also appied some fixes and simplifications to the tracing.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 10:36:44 +0000 (11:36 +0100)]
Previous changeset trivially broke ioemu/helper2.c. Fix it.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 10:34:29 +0000 (11:34 +0100)]
This patch fixed keyboard issue for linux kernel 2.6 in vmx guest.
linux 2.6 kernel keyboard use a timer to poll the kbd state,in
virtualization environment,the kbd interrupt injection is very likely
to happend just after kbd state query in the kbd timer and thus
will reverse sequence of scan code.
Also fix env->send_event SMP issue.
Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
Signed-off-by: Edwin Zhai<edwin.zhai@intel.com>
Signed-off-by: Eddion Dong <eddie.dong@intel.com>
Signed-off-by: Asit Mallick <asit.k.mallick@intel.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 10:28:33 +0000 (11:28 +0100)]
Properly declare ioapic_update_EOI().
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Tue, 15 Nov 2005 10:28:01 +0000 (11:28 +0100)]
Initialise secondary CPUs via an arch_initcall rather than
a subsys_initcall. This ensures that initialisation occurs
before topology_init().
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 18:55:40 +0000 (19:55 +0100)]
get_page_type() must clean up writable pagetable state
before failing, otherwise functions like set_gdt() can
spuriously fail because a page appears to be writable
despite there being pending changes to mapping of that
page in ptwr state.
This should fix the reported cases of crashes in
vcpu_prepare in xenlinux.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 18:45:32 +0000 (19:45 +0100)]
Better fix for flushing conflicting batched ptwr state
before emulating a PT access.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 17:27:11 +0000 (18:27 +0100)]
Flush writable pagetable state before emulating a PT
update. Avoids possibility of updating a PTE temporarily
marked writable by ptwr batching logic, which can corrupt
page reference counts. Aiee!
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 17:13:38 +0000 (18:13 +0100)]
Use make_lowmem_page_readonly/writable() in preference to the
generic functions where appropriate. This prevents us using the
generic functions early during boot, when pte_pfn() does not work
(because max_mapnr is not initialised).
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 14:21:16 +0000 (15:21 +0100)]
Failure to revalidate a writable pagetable page should crash
the offending domain, not merely pause it. Also, callers can
now specify another domain to domain_crash().
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Mon, 14 Nov 2005 11:28:47 +0000 (12:28 +0100)]
Fix check for a working iptables installation. Checking for the iptables
binary is not sufficient, because the user may not have the appropriate
modules installed.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Mon, 14 Nov 2005 11:12:29 +0000 (12:12 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Mon, 14 Nov 2005 11:11:41 +0000 (12:11 +0100)]
Added line of debugging to try and catch hotplug script race.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 10:47:38 +0000 (11:47 +0100)]
Declare local variables at the start of a block, not in the middle
(ISO C constraint).
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 10:36:42 +0000 (11:36 +0100)]
Clean up xenstore/console shared page initialisation, which is
now handled solely by the domain builder.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 10:35:50 +0000 (11:35 +0100)]
Revert accidentally applied changes from changeset
7783:
5aad7e145e501fbfb346954950a33b042a963633.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 10:11:11 +0000 (11:11 +0100)]
Trivial Makefile cleanup for ioemu.
emellor@leeni.uk.xensource.com [Mon, 14 Nov 2005 10:06:25 +0000 (11:06 +0100)]
Allocate MAC addresses using Xensource Inc's new OUI.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Mon, 14 Nov 2005 10:05:34 +0000 (11:05 +0100)]
If /sbin/ isn't in the path, udev rules will erroneously not get
installed.
Signed-off-by: Nivedita Singhvi <niv@us.ibm.com>
emellor@leeni.uk.xensource.com [Sat, 12 Nov 2005 17:59:56 +0000 (18:59 +0100)]
The various changes to move to using info instead of passing the config
sxpr everywhere broke the use of a boot loader for booting with. The
attached fixes things up so that
a) use of a bootloader ==> kernel not specified in the config file
b) don't cleanup bootloader files until after the domain has booted
c) rerun bootloader on reboot
Signed-off-by: Jeremy Katz <katzj@redhat.com>
emellor@leeni.uk.xensource.com [Sat, 12 Nov 2005 17:52:16 +0000 (18:52 +0100)]
Further changes to xm.1 man page.
Signed-off-by: Sean Dague <sean@dague.net>
emellor@leeni.uk.xensource.com [Sat, 12 Nov 2005 17:41:37 +0000 (18:41 +0100)]
From Charles Coffing <ccoffing@novell.com>:
"make uninstall" was not cleaning up some udev rules. Also, on x86_64
there was stuff left in /usr/lib/xen, not just /usr/$LIBDIR/xen.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Sat, 12 Nov 2005 17:32:19 +0000 (18:32 +0100)]
The following patch is work in progress to add more meat to the xm.1 man
page. Expect more content to trickle in over time.
Signed-off-by: Sean Dague <sean@dague.net>
dan@guaranine.beaverton.ibm.com [Sat, 12 Nov 2005 17:25:12 +0000 (18:25 +0100)]
This test was failing for he wrong reason. Calls to block-attach after the
first *should* report failure, so the test should not fail because of this.
The test now fails because multiple calls to block-attach seem to detach the
block devicem, which is wrong.
Signed-off-by: Dan Smith <danms@us.ibm.com>
emellor@leeni.uk.xensource.com [Sat, 12 Nov 2005 16:51:14 +0000 (17:51 +0100)]
Disable use of mmap. This allows Xenstored to run with an NFS-mounted
/var/lib/xenstored. Closes bug #400. This fix no doubt has a performance hit,
so I have opened bug #401 to discuss a better fix.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 20:59:33 +0000 (21:59 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 20:59:05 +0000 (21:59 +0100)]
New Xenbus driver structure, bringing an explicit state machine to the table.
This will make the drivers more robust to failure of the other end, and makes
current driver state more visible. It also allows us to bring very similar
pieces of code from the drivers together into the xenbus driver. This should
be the start of a fix to the problem with block and net device hotplugging,
because the code was previously confused about the difference between the
connection state of the kernel device layer, and the state of the xenbus
communication. This confusion is addressed here.
Much of this work is influenced by Rusty Russell's skeleton driver, recently
posted to xen-devel, as that piece made suggestions as to structure and
arrangement that have been taken on board, albeit pushed down into the xenbus
layer.
A new file xenbus_client.c has been added. This file implements the interface
between the xenbus layer and the layer specific to the driver. Access to
lower layers is still allowed for those who do not need all the facilities of
the new driver.
xenbus_dev_error has moved from xenbus_xs.c into xenbus_client.c, and
xenbus_dev_ok has been removed, in favour of xenbus_switch_state.
DevController.py now closes down a device by writing a change to the Closing
state on the backend, to allow the frontend time to close down properly. Its
also writes the frontend and backend details within the same transaction.
The timeout on the hotplug scripts has been cranked down from 2 minutes to 5
seconds.
The diagnose script has been updated to parse the new states.
The blktab and tpm drivers have yet to be ported to this structure.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
smh22@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 18:11:13 +0000 (19:11 +0100)]
Attached patch allows PAE xenlinux to run in the shadow mode using
log-dirty guest-refcount, which is required to support
save/restore/relocate. We can turn on/off the mode every 5-sec interval
while doing kernel build (make -j4), for example, and it survives for
hours.
We are still restoring log-dirty mode for x86_64 xenlinux, which is not
very stable right now, but I believe it should be done very soon.
We also checked that it did not break 64-bit VMX domains, which uses
different sub-mode of the shadow mode.
Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
smh22@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 18:02:49 +0000 (19:02 +0100)]
Fix make_page_readonly/make_page_writeable on PAE guests - previous behaviour
was incorrect when machine address > 4GB. We believe this fixes bugzilla #267.
Also tidy up and fix show_page_walk.
Signed-off-by: Steven Hand <steven@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 17:01:44 +0000 (18:01 +0100)]
Rename the commands inside the hotplug scripts to match those used by hotplug
and udev. Pass all commands through from xen-backend.agent or
xen-backend.rules rather than second-guessing the individual scripts. This
should make it easier to tear down devices cleanly, especially the dangling
vifs we are seeing at the moment.
Cope with a missing type node inside the block script, as we don't want to
resurrect a removed node by writing a hotplug error message.
Log failure of iptables command.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 16:59:20 +0000 (17:59 +0100)]
New location for code common to both netfront and netback, with the MAC parsing
pulled here.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 16:56:02 +0000 (17:56 +0100)]
Don't call iptables if it is not installed. This code is only there to ensure
that packets between domU and the bridge are not firewalled, so if iptables is
not installed, then there is no worry.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 16:04:17 +0000 (17:04 +0100)]
Fix function call syntax in vif-route and vif-nat.
Function calls in shell script should not contain parenthesis.
Signed-off-by: Murillo Fernandes Bernardes <mfb@br.ibm.com>
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 14:36:16 +0000 (15:36 +0100)]
Merged.
emellor@leeni.uk.xensource.com [Fri, 11 Nov 2005 14:36:07 +0000 (15:36 +0100)]
Chmod a+x.
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 13:54:20 +0000 (14:54 +0100)]
Missed two files from NULL-pointer-check changeset.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 11:55:47 +0000 (12:55 +0100)]
Remove unnecessary NULL checks before freeing memory blocks.
The following functions check for NULL internally:
free, kfree, xfree, free_xenheap_page[s]
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 10:16:46 +0000 (11:16 +0100)]
Fix application of OS patches to deal better with failure.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 10:01:32 +0000 (11:01 +0100)]
Max xen skbuff allocation is now 64KB. kmem caches only
get populated with objects on first allocation, so if large
caches end up being unused, they don't actually cost us
very much.
Signed-off-by: Keir Fraser <keir@xensource.com>
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 09:47:43 +0000 (10:47 +0100)]
Merge with xen-unstable-ia64.hg
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 09:46:36 +0000 (10:46 +0100)]
nloopbacks default is now 8. So vifnum of greater than 7 requires
an adjustment to nloopbacks. Warning comment updated.
Signed-off-by: Nivedita Singhvi (niv@us.ibm.com)
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 09:41:25 +0000 (10:41 +0100)]
Add MOVSX instr support to VMX MMIO decoder.
Signed-off-by: Xin Li <xin.b.li@intel.com>
kaf24@firebug.cl.cam.ac.uk [Fri, 11 Nov 2005 09:39:20 +0000 (10:39 +0100)]
Remove unnecessary ifdef in defn of xprintk. Install
xenguest.h.
Signed-off-by: Joseph Cihula <joseph.cihula@intel.com>
djm@kirby.fc.hp.com [Thu, 10 Nov 2005 21:24:29 +0000 (15:24 -0600)]
Enable gcc3.4 with CONFIG_SMP (by Alex Williamson)
djm@kirby.fc.hp.com [Thu, 10 Nov 2005 20:02:01 +0000 (14:02 -0600)]
Blkfront no longer built by default, need CONFIG_XEN_BLKDEV_FRONTEND