xen.git
19 years ago[IA64] add missing END(xen_bsw1)
awilliam@xenbuild.aw [Tue, 30 May 2006 16:30:12 +0000 (10:30 -0600)]
[IA64] add missing END(xen_bsw1)

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] VTI: Optimize thash vtlb algorithm
awilliam@xenbuild.aw [Tue, 30 May 2006 16:28:59 +0000 (10:28 -0600)]
[IA64] VTI: Optimize thash vtlb algorithm

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years ago[IA64] VTI: simple format cleanup
awilliam@xenbuild.aw [Tue, 30 May 2006 14:46:21 +0000 (08:46 -0600)]
[IA64] VTI: simple format cleanup

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years agomerge with xen-unstable.hg
awilliam@xenbuild.aw [Fri, 26 May 2006 19:41:49 +0000 (13:41 -0600)]
merge with xen-unstable.hg

19 years ago[PAE] Allow pgdirs above 4GB for paravirt guests.
kaf24@firebug.cl.cam.ac.uk [Fri, 26 May 2006 16:22:30 +0000 (17:22 +0100)]
[PAE] Allow pgdirs above 4GB for paravirt guests.
**NOTE**: This obviates the need for lowmem_emergency_pool.
Unpriv guests no longer need to be able to allocate memory
below 4GB for PAE PDPTs.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoRemove bogus inline asm of unused function memchr().
kaf24@firebug.cl.cam.ac.uk [Fri, 26 May 2006 12:53:49 +0000 (13:53 +0100)]
Remove bogus inline asm of unused function memchr().
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoCast to a signed type before negating if we really want the result to be negative.
kaf24@firebug.cl.cam.ac.uk [Fri, 26 May 2006 12:49:29 +0000 (13:49 +0100)]
Cast to a signed type before negating if we really want the result to be negative.
From: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[MINIOS] Allows the use of a C library without modifying Mini-OS source files.
kaf24@firebug.cl.cam.ac.uk [Fri, 26 May 2006 12:23:18 +0000 (13:23 +0100)]
[MINIOS] Allows the use of a C library without modifying Mini-OS source files.
From: John Ramsdell
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoMove include/asm-x86_64/e820.h to include/asm-x86_64/mach-xen/asm/e820.h
Ian.Campbell@xensource.com [Fri, 26 May 2006 10:23:55 +0000 (11:23 +0100)]
Move include/asm-x86_64/e820.h to include/asm-x86_64/mach-xen/asm/e820.h

This fixes the native build.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
--HG--
rename : linux-2.6-xen-sparse/include/asm-x86_64/e820.h => linux-2.6-xen-sparse/include/asm-x86_64/mach-xen/asm/e820.h

19 years agoNew weighted fair-share CPU scheduler w/ automatic SMP load balancing
ack@kneesa.uk.xensource.com [Fri, 26 May 2006 10:14:36 +0000 (11:14 +0100)]
New weighted fair-share CPU scheduler w/ automatic SMP load balancing
Signed-off-by: Emmanuel Ackaouy <ack@xensource.com>
19 years agoFix cscope rule in .hgignore.
kaf24@firebug.cl.cam.ac.uk [Fri, 26 May 2006 08:44:29 +0000 (09:44 +0100)]
Fix cscope rule in .hgignore.

19 years ago[IA64] support FPSWA hypercall
awilliam@xenbuild.aw [Thu, 25 May 2006 21:59:18 +0000 (15:59 -0600)]
[IA64] support FPSWA hypercall

Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
19 years agoRemove net_driver_util module. Inline the one mac-parsing function
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 21:57:44 +0000 (22:57 +0100)]
Remove net_driver_util module. Inline the one mac-parsing function
into the netfront and netback drivers. Take the opportunity to fix
the mac-parsing logic.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] support FPSWA emulation
awilliam@xenbuild.aw [Thu, 25 May 2006 21:47:33 +0000 (15:47 -0600)]
[IA64] support FPSWA emulation

Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
19 years ago[IA64] remove evtchn_ia64.c
awilliam@xenbuild.aw [Thu, 25 May 2006 21:38:16 +0000 (15:38 -0600)]
[IA64] remove evtchn_ia64.c

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

19 years ago[XEND] Wait sufficient time for memory to balloon out before creating a new domain.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 20:41:04 +0000 (21:41 +0100)]
[XEND] Wait sufficient time for memory to balloon out before creating a new domain.

On the ES7000 when Dom0 boots up with all of system memory and you try
to bring up a DomU with more than 2GB of memory, xend times out before
the memory is freed causing the domain not to be created. This patch
increases the timeout depending on the amount of memory that needs to be
freed. It also places a cap on the sleep time so that it does not grow
without a limit. This fixes bug# 650

http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=3D650

Signed-off-by: Aravindh Puthiyaparambil
<aravindh.puthiyaparambil@unisys.com>

19 years agoAllow compat hypercalls to be made via privcmd (to support older userspaces).
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 20:24:47 +0000 (21:24 +0100)]
Allow compat hypercalls to be made via privcmd (to support older userspaces).
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
19 years agoFix VMEXIT tracing.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 20:18:55 +0000 (21:18 +0100)]
Fix VMEXIT tracing.
Signed-off-by: Bruce Rogers <brogers@novell.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[HVM] Support multiple HVM time device models coming soon.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 15:00:36 +0000 (16:00 +0100)]
[HVM] Support multiple HVM time device models coming soon.
        1: Adopt an abstract layer for periodic time on top of different
HV time device models like PIT, RTC and ACPI time.
        2: Extract VMM knowledge from PIT DM code and keep close with
qemu for easy maintain in future.
        3: Use guest time (TSC) to drive PIT CLK that helps us to remove
a lot of extra logic previously introduced to convert from host time to
guest time.
        4: Some cleanup for SMP code like move cache_tsc_offset to per VP.

Signed-off-by: Eddie Dong <eddie.dong@intel.com>
19 years agoFix x86/64 build after previous changeset.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 15:00:09 +0000 (16:00 +0100)]
Fix x86/64 build after previous changeset.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoRemove redundant test in grant-table interface code.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 14:55:07 +0000 (15:55 +0100)]
Remove redundant test in grant-table interface code.
Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>
19 years agoChange the x86_emulate() interface to pack all context arguments
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 14:52:38 +0000 (15:52 +0100)]
Change the x86_emulate() interface to pack all context arguments
into a context structure. This context can then be handed to
callback operations.

Based on an original patch from Mats Petersson <mats.petersson@amd.com>

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[BALLOON] Do not allow target to be set below two percent of maximum
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:54:43 +0000 (14:54 +0100)]
[BALLOON] Do not allow target to be set below two percent of maximum
memory size.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoDo not include asm/fixmap.h directly from non-arch-specific Linux files.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:47:23 +0000 (14:47 +0100)]
Do not include asm/fixmap.h directly from non-arch-specific Linux files.
Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>
19 years ago[XEND] Remember cpu-pinning information across guest restarts.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:45:40 +0000 (14:45 +0100)]
[XEND] Remember cpu-pinning information across guest restarts.
Signed-off-by: Qing He <qing.he@intel.com>
19 years agoFix xc_tbuf.c indentation and stop turning failure to change
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:42:07 +0000 (14:42 +0100)]
Fix xc_tbuf.c indentation and stop turning failure to change
trace-buffer size into a fatal error.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoEnsure trace-buffer size is set back to zero if trace buffers can't be allocated.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:37:55 +0000 (14:37 +0100)]
Ensure trace-buffer size is set back to zero if trace buffers can't be allocated.
Signed-off-by: Atsushi Sakai <sakaia@jp.fujitsu.com>
19 years ago[NET] front: Use netif_carrier to model backend-connection state.
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:29:21 +0000 (14:29 +0100)]
[NET] front: Use netif_carrier to model backend-connection state.

The tristate backend_state really only needs to be a boolean.  In fact
other virtualisation network drivers such as the s390 already model this
using netif_carrier_*.  So by converting backend_state over to the net
carrier flag we can tap into existing infrastructure rather than creating
new ones.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
19 years agoSeparate xenbusstate enumeration from typedef. Avoid using the typedef
kaf24@firebug.cl.cam.ac.uk [Thu, 25 May 2006 13:15:49 +0000 (14:15 +0100)]
Separate xenbusstate enumeration from typedef. Avoid using the typedef
in Linux coe as it is consider bad coding style.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] fix AFLAGS again
awilliam@xenbuild.aw [Wed, 24 May 2006 22:44:20 +0000 (16:44 -0600)]
[IA64] fix AFLAGS again

"AFLAGS = -nostdinc $CPPFLAGS" is removed

Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
Signed-off-by: Kouya Shimura <kouya@jp.fujitsu.com>
19 years agoFix IDE CD-drive PIO mode.
cl349@firebug.cl.cam.ac.uk [Wed, 24 May 2006 18:41:47 +0000 (19:41 +0100)]
Fix IDE CD-drive PIO mode.
CD drives in PIO mode don't work under Xen because of a change in Linux
between 2.6.12 and 2.6.16, as a result of the following thread:
http://lists.parisc-linux.org/pipermail/parisc-linux/2005-August/027197.html

The change breaks systems which have highmem and a swiotlb because the
ide-cd driver doesn't use the swiotlb, resulting in read/writes to/from
highmem pages in PIO mode not working any longer.  Xen kernels usually have
both highmem and a swiotlb.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years ago[IA64] acpi read-only mapping
awilliam@xenbuild.aw [Wed, 24 May 2006 17:04:33 +0000 (11:04 -0600)]
[IA64] acpi read-only mapping

implemented dom0 acpi read-only mapping

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] GNTMAP_readonly support linux part
awilliam@xenbuild.aw [Wed, 24 May 2006 16:59:28 +0000 (10:59 -0600)]
[IA64] GNTMAP_readonly support linux part

add read only mapping support.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] GNTMAP_readonly support xen part
awilliam@xenbuild.aw [Wed, 24 May 2006 16:56:47 +0000 (10:56 -0600)]
[IA64] GNTMAP_readonly support xen part

add grant table GNTMAP_readonly support.
introduce ASSIGN_readonly flags for read only page assignment to
pseudo physical address space.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] support DOMID_XEN and DOMID_IO of foreign domain page mapping
awilliam@xenbuild.aw [Wed, 24 May 2006 16:39:55 +0000 (10:39 -0600)]
[IA64] support DOMID_XEN and DOMID_IO of foreign domain page mapping

support of foreign domain page mapping of DOMID_XEN and DOMID_IO.
This patch is needed for xentrace and xenmon.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] fix dirty_bit in xenivt.S
awilliam@xenbuild.aw [Wed, 24 May 2006 15:59:04 +0000 (09:59 -0600)]
[IA64] fix dirty_bit in xenivt.S

fix dirty_bit in linux-2.6-xen-sparse/arch/ia64/xen/xenivt.S
there was itc.d left which wasn't para-virtualized.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] remove unused audio driver from buildconfigs
awilliam@xenbuild.aw [Wed, 24 May 2006 14:56:52 +0000 (08:56 -0600)]
[IA64] remove unused audio driver from buildconfigs

There'll be better way to prevent bogus driver to register with
an empty irq line. But for now, disable in buildconfig is easier.

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

19 years ago[IA64] swap XEN_HYPER_ITC_D and XEN_HYPER_ITC_I
awilliam@xenbuild.aw [Wed, 24 May 2006 14:50:34 +0000 (08:50 -0600)]
[IA64] swap XEN_HYPER_ITC_D and XEN_HYPER_ITC_I

Signed-off-by Dietmar Hahn <dietmar.hahn@fujitsu-siemens.com>

19 years ago[IA64] Add ptc.l emulation
awilliam@xenbuild.aw [Wed, 24 May 2006 14:48:21 +0000 (08:48 -0600)]
[IA64] Add ptc.l emulation

This patch adds support for ptc.l emulation for xenlinux.
In 2.6.16 kernel it flush_tlb_range will call global_tlb_purge
directly, so ptc.l shouldn't be used when CONFIG_SMP enable.
But in order to enhance performance (maybe), 2.6.17 kernel in
smp environment will do mm check first. If mm is current->active_mm
and the mm (corresponding process) just runs on the local processor,
kernel only needs to do ptc.l at local processor instead of global purge.
So ptc.l emulation is necessary for 2.6.17 kernel.

Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Zhang Xiantao <xiantao.zhang@intel.com>
19 years agoRemove unnecessary x86-specific mapping code from privcmd interface.
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 13:49:24 +0000 (14:49 +0100)]
Remove unnecessary x86-specific mapping code from privcmd interface.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[XENSTORE] Make use of /proc/xen/xsd_{port,kva} private to the Linux implementation.
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 13:24:57 +0000 (14:24 +0100)]
[XENSTORE] Make use of /proc/xen/xsd_{port,kva} private to the Linux implementation.
Signed-off-by: John Levon <john.levon@sun.com>
19 years ago[PTRACE] Fix ptrace for hvm guests.
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 13:12:40 +0000 (14:12 +0100)]
[PTRACE] Fix ptrace for hvm guests.
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
19 years agoEvent-channel driver: Use for_each_possible_cpu instead
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 11:01:25 +0000 (12:01 +0100)]
Event-channel driver: Use for_each_possible_cpu instead
of for_each_cpu , or statically initialize data when possible.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
19 years agoFix kcons_write and kcons_write_dom0 CodingStyle.
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 10:57:37 +0000 (11:57 +0100)]
Fix kcons_write and kcons_write_dom0 CodingStyle.
Use roundup_pow_of_two().

Signed-off-by: Chris Wright <chrisw@sous-sol.org>
19 years agoUse DEFINE_SPINLOCK in Linux drivers, conforming to coding style.
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 10:56:40 +0000 (11:56 +0100)]
Use DEFINE_SPINLOCK in Linux drivers, conforming to coding style.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
19 years agoRemove broken inline-asm string functions from Xen/x86. The
kaf24@firebug.cl.cam.ac.uk [Wed, 24 May 2006 10:52:25 +0000 (11:52 +0100)]
Remove broken inline-asm string functions from Xen/x86. The
generic implementations in string.c are quite sufficient.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] work around ENOSYS vs EINVAL in ioremap
awilliam@xenbuild.aw [Tue, 23 May 2006 23:05:27 +0000 (17:05 -0600)]
[IA64] work around ENOSYS vs EINVAL in ioremap

We should not panic the domain for an EINVAL return from the ioremap
hyerpcall.  This was only meant for ENOSYS.  Long term this is probably
not the best place for this kind of sanity checking.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years ago[IA64] Add -rc support to sparse-merge
awilliam@xenbuild.aw [Tue, 23 May 2006 21:10:27 +0000 (15:10 -0600)]
[IA64] Add -rc support to sparse-merge

Signed-off-by: Aron Griffis <aron@hp.com>
19 years ago[IA64] make xenLinux/ia64 privcmd mmap not to use dom0 memory
awilliam@xenbuild.aw [Tue, 23 May 2006 21:09:21 +0000 (15:09 -0600)]
[IA64] make xenLinux/ia64 privcmd mmap not to use dom0 memory

xenLinux/ia64 privcmd mmap uses pseudo physical address space.
it used alloc_pages() to allocate the space.
It wastes dom0 memory and sometimes several hundreds megabytes is
allocated depending on domU memory size.
With this patch xenLinux/ia64 trys to find the region which can be
used safely and uses the reasion.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoAllow a guest to determine its VM UUID.
emellor@leeni.uk.xensource.com [Tue, 23 May 2006 15:23:10 +0000 (16:23 +0100)]
Allow a guest to determine its VM UUID.

Use xenstore-read $(xenstore-read vm)/uuid.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
19 years ago[IA64] Final patch to make evtchn fully working for xen/ia64
awilliam@xenbuild.aw [Tue, 23 May 2006 15:17:57 +0000 (09:17 -0600)]
[IA64] Final patch to make evtchn fully working for xen/ia64

OK, this is the last one of patch sets, which enables xen/ia64 to work
on event channel mechanism fully. Now user may observe /proc/interrupts
of dom0 as:

           CPU0
 34:         12        Phys-irq  ide0
 39:          0        Phys-irq  acpi
 45:        322        Phys-irq  serial
 48:     115006        Phys-irq  peth0
 49:      16269        Phys-irq  ioc0
 50:         31        Phys-irq  ioc1
 51:          2        Phys-irq  ehci_hcd:usb1
 52:          0        Phys-irq  uhci_hcd:usb2
 53:         55        Phys-irq  uhci_hcd:usb3
256:          0     Dynamic-irq  RESCHED0
257:          0     Dynamic-irq  IPI0
258:      44572     Dynamic-irq  timer0
259:       2316     Dynamic-irq  xenbus
260:       8304     Dynamic-irq  blkif-backend
261:      25947     Dynamic-irq  vif3.0
ERR:          0

Based on this patch, dom0's performance is not affected, while domU's
network output is seen with several percents increase. More important,
now xen/ia64 can progress a base step towards driver domain support!

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

19 years agoFix log message written when xenstored corrupts its store.
emellor@leeni.uk.xensource.com [Tue, 23 May 2006 15:17:28 +0000 (16:17 +0100)]
Fix log message written when xenstored corrupts its store.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
19 years agoAdded get_permissions implementation.
emellor@leeni.uk.xensource.com [Tue, 23 May 2006 15:14:49 +0000 (16:14 +0100)]
Added get_permissions implementation.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
19 years agoFix xspy_get_permissions -- you can't increment the perms pointer when
emellor@leeni.uk.xensource.com [Tue, 23 May 2006 15:14:03 +0000 (16:14 +0100)]
Fix xspy_get_permissions -- you can't increment the perms pointer when
iterating and expect to be able to free that pointer afterwards.

Signed-off-by: Ewan Mellor <ewan@xensource.com>
19 years ago[IA64] Clean up pirq bind code
awilliam@xenbuild.aw [Tue, 23 May 2006 14:56:13 +0000 (08:56 -0600)]
[IA64] Clean up pirq bind code

This patch clears some pirq related logic as a last prep step
for the final evtchn to be effective. Same, this patch doesn't
break existing logic.

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

19 years ago[IA64] Add event injection logic
awilliam@xenbuild.aw [Tue, 23 May 2006 14:34:48 +0000 (08:34 -0600)]
[IA64] Add event injection logic

Add event inject logic. Because up to this point there's no place
to register callback, this patch doesn't break existing working flow.

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

19 years ago[IA64] Add do_callback_ops
awilliam@xenbuild.aw [Tue, 23 May 2006 14:24:09 +0000 (08:24 -0600)]
[IA64] Add do_callback_ops

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

19 years ago[IA64] Import linux header files
awilliam@xenbuild.aw [Tue, 23 May 2006 14:18:48 +0000 (08:18 -0600)]
[IA64] Import linux header files

Import hw_irq.h and irq.h from linux

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

19 years agoMove Xen CPU hotplug code into new file cpu_hotplug.c.
kaf24@firebug.cl.cam.ac.uk [Tue, 23 May 2006 11:05:08 +0000 (12:05 +0100)]
Move Xen CPU hotplug code into new file cpu_hotplug.c.
Cleaner and allows use by architectures which aren't currently
using Xen's smpboot.c.

Based on an original patch by Tristan Gingold.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] update default build configs
awilliam@xenbuild.aw [Mon, 22 May 2006 21:08:26 +0000 (15:08 -0600)]
[IA64] update default build configs

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years agomerge with xen-unstable.hg
awilliam@xenbuild.aw [Mon, 22 May 2006 20:13:38 +0000 (14:13 -0600)]
merge with xen-unstable.hg

19 years ago[IA64] Work around auto-ballooning changes.
kaf24@firebug.cl.cam.ac.uk [Mon, 22 May 2006 16:30:25 +0000 (17:30 +0100)]
[IA64] Work around auto-ballooning changes.
Original patch from Kevin Tian at Intel.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoFix x86/64 by passing the required length to reserve_bootmem instead of the end address
Ian.Campbell@xensource.com [Mon, 22 May 2006 16:03:24 +0000 (17:03 +0100)]
Fix x86/64 by passing the required length to reserve_bootmem instead of the end address

Fixup a warning since the machine_e820 and memmap variables are only used if
CONFIG_XEN_PRIVILEGED_GUEST.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years ago[IA64] Enable VTi network by defalt
awilliam@xenbuild.aw [Mon, 22 May 2006 14:53:26 +0000 (08:53 -0600)]
[IA64] Enable VTi network by defalt

Due to dom0_vp open, we can enable VTi's network in configuration file now.

From: Zhang Xiantao <xiantao.zhang@intel.com>
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years agoPatch linux so that /proc/stat shows all -- not just online -- CPUs.
ack@kneesa.uk.xensource.com [Mon, 22 May 2006 14:42:54 +0000 (15:42 +0100)]
Patch linux so that /proc/stat shows all -- not just online -- CPUs.
This way, tools such as TOP don't freak out when CPUs are hotplugged.

19 years agoFix x86/64 -xenU build. probe_roms() is only defined if building a privleged
Ian.Campbell@xensource.com [Mon, 22 May 2006 14:03:46 +0000 (15:03 +0100)]
Fix x86/64 -xenU build. probe_roms() is only defined if building a privleged
or !xen build.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoRemove DOM0_PHYSICAL_MEMORY_MAP dom0 op.
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:23:21 +0000 (09:23 +0100)]
Remove DOM0_PHYSICAL_MEMORY_MAP dom0 op.

The addition of the e820 style memory_op sub calls make this operation
obsolete.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoConvert x86/64 Linux to use the new memory map hypercall.
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:23:15 +0000 (09:23 +0100)]
Convert x86/64 Linux to use the new memory map hypercall.

This change removes several of the differences between the bare-metal
and Xen versions.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoConvert x86/32 Linux to use the new memory map hypercall.
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:23:03 +0000 (09:23 +0100)]
Convert x86/32 Linux to use the new memory map hypercall.

This change removes several of the differences between the bare-metal
and Xen versions.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoNew memory_op subops which return the apparent or actual physical address map.
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:22:18 +0000 (09:22 +0100)]
New memory_op subops which return the apparent or actual physical address map.

The new subops return a memory map in e820 format. This will allow the
removal of some Xen special casing in the Linux port by using the same
code as native.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoAdd include/asm-x86_64/e820.h to sparse tree
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:21:58 +0000 (09:21 +0100)]
Add include/asm-x86_64/e820.h to sparse tree

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years agoRe-order code in preparation for future patches.
Ian.Campbell@xensource.com [Mon, 22 May 2006 08:21:01 +0000 (09:21 +0100)]
Re-order code in preparation for future patches.

There is no change to the functions which have moved.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
19 years ago[LOADER] More sanity checks when parsing Elf images to avoid
kaf24@firebug.cl.cam.ac.uk [Sun, 21 May 2006 19:15:58 +0000 (20:15 +0100)]
[LOADER] More sanity checks when parsing Elf images to avoid
out-of-bounds array accesses when loading the image.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[ACM] Replace the union structure parameter of the ACM hypercalls
kaf24@firebug.cl.cam.ac.uk [Sun, 21 May 2006 18:05:31 +0000 (19:05 +0100)]
[ACM] Replace the union structure parameter of the ACM hypercalls
with command-specific request structures. It aligns the ACM
hypercalls with the way parameters are passed in the event channel
hypercalls.

Advantages include backward-compatibility regarding old guests when
new calls are added and clarity of the code.

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

19 years agoFinish separating the typedefs from structure definitions in Xen public headers.
kaf24@firebug.cl.cam.ac.uk [Sun, 21 May 2006 17:57:42 +0000 (18:57 +0100)]
Finish separating the typedefs from structure definitions in Xen public headers.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
19 years ago[LOADER] Another change to the elf_paddr_offset/virt_base logic.
kaf24@firebug.cl.cam.ac.uk [Sun, 21 May 2006 17:55:14 +0000 (18:55 +0100)]
[LOADER] Another change to the elf_paddr_offset/virt_base logic.
This avoids using zero to indicate 'undefined'. Instead we use
an explicit boolean to indicate whether suitable defaults should
be used. As well as making the logic clearer (I believe) it also
avoids problems if anyone ever explicitly specifies virt_base
(and also elf_paddr_offset) as zero.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] Enable P2M/VP by default
awilliam@xenbuild.aw [Sun, 21 May 2006 13:49:46 +0000 (07:49 -0600)]
[IA64] Enable P2M/VP by default

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years ago[IA64] add missing ia64_trace_syscall privop
awilliam@xenbuild.aw [Sun, 21 May 2006 13:31:02 +0000 (07:31 -0600)]
[IA64] add missing ia64_trace_syscall privop

This patch adds a missing privop definition for ia64_trace_syscall
when CONFIG_XEN is unset.  It also cleans up some inconsistent
spacing.

Signed-off-by: Aron Griffis <aron@hp.com>
19 years ago[IA64] make it possible to disable CONFIG_XEN
awilliam@xenbuild.aw [Sun, 21 May 2006 13:28:05 +0000 (07:28 -0600)]
[IA64] make it possible to disable CONFIG_XEN

This patch helps distro inclusion efforts by making it possible to
build a xen-patched Linux kernel with CONFIG_XEN disabled.

Signed-off-by: Aron Griffis <aron@hp.com>
19 years ago[IA64] sparse-merge: check SPARSEDIR for modifications
awilliam@xenbuild.aw [Sun, 21 May 2006 13:24:51 +0000 (07:24 -0600)]
[IA64] sparse-merge: check SPARSEDIR for modifications

Check SPARSEDIR for modifications before allowing the merge.
This removes the need for the FIXME.  Additionally clean up
empty diffs at the end of the run.

Signed-off-by: Aron Griffis <aron@hp.com>
19 years ago[IA64] minor fix for sparse-merge
awilliam@xenbuild.aw [Sun, 21 May 2006 13:22:58 +0000 (07:22 -0600)]
[IA64] minor fix for sparse-merge

Here's another sparse-merge patch to allow ARCH to be set on the
command-line, and to add -C to the final hg update since it's probably
switching between branches.

Signed-off-by: Aron Griffis <aron@hp.com>
19 years ago[LOADER] Remove check for VIRT_BASE existence: it's not needed
kaf24@firebug.cl.cam.ac.uk [Sun, 21 May 2006 08:55:15 +0000 (09:55 +0100)]
[LOADER] Remove check for VIRT_BASE existence: it's not needed
to prevent xend crashign (when Aravindh's latest patch is
applied) and breaks ia64. Also, ensure we set a reasonable
default for elf_paddr_offset.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[BUILDER] Always build initial PAE pagetables with a fourth mid-level
kaf24@firebug.cl.cam.ac.uk [Sat, 20 May 2006 15:25:28 +0000 (16:25 +0100)]
[BUILDER] Always build initial PAE pagetables with a fourth mid-level
pgdir. This is required for Xen to be able to insert its own protected
mappings. At the same time, clean up 32b and pae pagetable-size
calculations to use the helper macro already used for x86/64.

Based on an original patch from Bruce Rogers at Novell.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoDon't require sed to support the -r option.
cl349@firebug.cl.cam.ac.uk [Fri, 19 May 2006 21:08:32 +0000 (22:08 +0100)]
Don't require sed to support the -r option.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years agoMake the latest network-bridge tweaks work in a busybox environment too.
cl349@firebug.cl.cam.ac.uk [Fri, 19 May 2006 21:00:28 +0000 (22:00 +0100)]
Make the latest network-bridge tweaks work in a busybox environment too.

Signed-off-by: David Lively <dlively@virtualiron.com>
19 years ago[SHADOW] Crash the guest rather than BUGing the whole system when out of memory.
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 16:50:32 +0000 (17:50 +0100)]
[SHADOW] Crash the guest rather than BUGing the whole system when out of memory.
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
19 years agoMerged.
emellor@leeni.uk.xensource.com [Fri, 19 May 2006 16:31:34 +0000 (17:31 +0100)]
Merged.

19 years agoExpose the exceptions thrown by xen.lowlevel.xc so they can be caught.
Anthony Liguori [Fri, 19 May 2006 16:31:28 +0000 (17:31 +0100)]
Expose the exceptions thrown by xen.lowlevel.xc so they can be caught.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
19 years ago[MINIOS] PAE support.
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 15:52:52 +0000 (16:52 +0100)]
[MINIOS] PAE support.
Signed-off-by: Aravindh Puthiyaparambil <aravindh.puthiyaparambil@unisys.com>
19 years agoAdd a read-verify block device test to xm-test.
harry@localhost.localdomain [Fri, 19 May 2006 15:30:21 +0000 (16:30 +0100)]
Add a read-verify block device test to xm-test.

19 years agoThis patch resizes the file system on the xm-test HVM disk.img to use
stekloff@elm3b216.beaverton.ibm.com [Fri, 19 May 2006 15:29:34 +0000 (16:29 +0100)]
This patch resizes the file system on the xm-test HVM disk.img to use
the entire space available.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoMerged.
emellor@leeni.uk.xensource.com [Fri, 19 May 2006 15:21:49 +0000 (16:21 +0100)]
Merged.

19 years agoThe enforce_dom0_cpus test sets the number of vcpus for the Dom0 and
emellor@leeni.uk.xensource.com [Fri, 19 May 2006 15:21:43 +0000 (16:21 +0100)]
The enforce_dom0_cpus test sets the number of vcpus for the Dom0 and
checks to see if they have been set correctly. We found that on
multi-proc systems, it takes a certain amount of time for the number of
vcpus to change.

This patch checks the number of vcpus inside a loop until it is correct
or 20 seconds elapse, whichever happens first. If after 20 seconds the
number of vcpus is not changed correctly, an error is raised.

Signed-off-by: Rajagopalan Subrahmanian <raj.subrahmanian@unisys.com>
19 years agoSeveral page allocations, done in the hypervisor when starting an HVM
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 15:10:52 +0000 (16:10 +0100)]
Several page allocations, done in the hypervisor when starting an HVM
domain, are not checked.  This can cause the physical machine to crash
when starting the HVM domain during low-memory conditions.

Kudos to Charles Arnold for catching the problem with
shadow_direct_map_init.

Signed-off-by: Charles Coffing <ccoffing@novell.com>
19 years ago[XEND] An empirical and more conservative memory-overhead estimate for PV and HVM...
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 15:08:51 +0000 (16:08 +0100)]
[XEND] An empirical and more conservative memory-overhead estimate for PV and HVM guests.

This patch calculates the overhead needed for HVM domains.  If HVM is
supported by the hardware, I add a little ballooning overhead to
paravirtualized VMs also, to avoid low-memory situations.  (There are
various unchecked alloc_domheap_pages calls in shadow*.c that I am
trying to avoid tripping over for now...)  The values in this patch work
fine on 32 bit; I may update them later based on feedback and/or testing
on 64 bit.

Signed-off-by: Charles Coffing <ccoffing@novell.com>
19 years ago[XEND] separate concept of initial memory size and overhead memory size.
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 15:07:36 +0000 (16:07 +0100)]
[XEND] separate concept of initial memory size and overhead memory size.

When a domain (whether para- or fully-virtualized) reports how much
overhead memory it requires (via getDomainMemory in image.py), all such
memory was immediately allocated to the domain itself.  This is
certainly incorrect for HVM domains, since additional
increase_reservation calls are made later in qemu.  Since all ballooned
memory is already taken, qemu will fail.  The fix is to treat the
requested memory size and the overhead size as separate values.  The
requested memory size is immediately allocated to the new domain; the
overhead is left unallocated for whatever else might need it later.

Signed-off-by: Charles Coffing <ccoffing@novell.com>
19 years ago[XM-TEST] Use explicit (IP id==packet size) on hping2 tests to avoid
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 15:01:08 +0000 (16:01 +0100)]
[XM-TEST] Use explicit (IP id==packet size) on hping2 tests to avoid
problems with IP id==0

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
19 years ago[NET] front: Remove duplicate maybe_wake_tx
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 14:58:46 +0000 (15:58 +0100)]
[NET] front: Remove duplicate maybe_wake_tx

Since network_tx_buf_gc unconditionally calls network_maybe_wake_tx at
the end, there is no need to call it again right after a call to
network_tx_buf_gc.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
19 years agoFix xentrace_format to deal with more than 16 CPUs.
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 14:52:35 +0000 (15:52 +0100)]
Fix xentrace_format to deal with more than 16 CPUs.
Signed-off-by: KUWAMURA Shin'ya <kuwa@jp.fujitsu.com>
19 years ago[NET] front: remove use of typedefs for structures.
kaf24@firebug.cl.cam.ac.uk [Fri, 19 May 2006 14:51:26 +0000 (15:51 +0100)]
[NET] front: remove use of typedefs for structures.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>