Remove noisy printf when KVM masks CPU features

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

fix asynchronous ioctls

posix_aio_read expect aio requests to return the number of bytes
requests to be successfull, so we need to fake this up for ioctls.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

kvm: Trim cpu features not supported by kvm

Remove cpu features that are not supported by kvm from the cpuid features
reported to the guest.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix x86 feature modifications for features that set multiple bits

QEMU allows adding or removing cpu features by using the syntax '-cpu +feature'
or '-cpu -feature'.  Some cpuid features cause more than one bit to be set or
cleared; but QEMU stops after just one bit has been modified, causing the
feature bits to be inconsistent.

Fix by allowing all feature bits corresponding to a given name to be set.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Make x86 cpuid feature names available in file scope

To be used later.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

kvm: Add support for querying supported cpu features

kvm does not support all cpu features; add support for dunamically querying
the supported feature set.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Introduce kvm_check_extension to check if KVM extensions are supported

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

e1000: Do not reinit pci config space to 0

pci_register_device already mallocs the pci config space buffer filled
with zeroes.

Doing this again breaks some default config space writes like
setting the subsystem vendor id and subsystem device id.

Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

AIO deletion race fix

When deleting an fd event there is a chance the object doesn't get
deleted, but only ->deleted set positive and deleted somewhere later.

Now, if we create a handler for the fd again before the actual
deletion occurs, we end up writing data into an object that has
->deleted set, which is obviously wrong.

I see two ways to fix this:

1. Don't return ->deleted objects in the search
2. Unset ->deleted in the search

This patch implements 1. which feels safer to do. It fixes AIO issues
I've seen with curl, as libcurl unsets fd event listeners pretty
frequently.

Signed-off-by: Alexander Graf <alex@csgraf.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

reset state for load_linux

The linux loader is just an option rom like any other, just with
some special requirements. Right now, our option rom resetting
mechanism is not being applied to it. As a result, users using
-kernel will not be able to successfully reboot their machines

This patch fixes it by saving all the data we generated in
the load_linux() function, to be used later by the option rom
resetting mechanism.

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

register reset handler for option_roms

Currently, boot options are not preserved across a system reset.
option roms can modify themselves, or can for instance restore the real
int 0x19 vector after they tried to boot from it.

To properly do that, we need a reset handler registered to deal with option
roms. This patch is based on current version on qemu-kvm.git

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix cluster freeing in qcow2

Need to drop QCOW_OFLAG_COPIED from a cluster pointer before freeing it.

Add an explanation how thing meant to work.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Enable power button even generation.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Drop CONFIG_GDBSTUB

This is no user-flippable switch, and no arch makes use of disabling
gdbstub support. So it's pointless to keep the related #ifdefs and
configure hunks around - and risking breakages like 711c410fdd again.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Remove superfluous #includes.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Use target_phys_addr_t, not target_ulong.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Replace cpu_abort with hw_error

Signed-off-by: Paul Brook <paul@codesourcery.com>

Fix disassembler memory accesses

Sparc disassembler wants to check previous addresses for some stuff
and this may actually cause faults to be generated to the guest if the
address is close to page start, because of the function used for the
memory access.

Fix by changing ldub_code to cpu_memory_rw_debug, which doesn't trigger
exceptions.

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Suppress make directory messages.

We already print a directory prefix in non-verbose mode, so there's no
point printing a messages when recursive make enters/leaves a directory.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Fix PL110 framebuffer byteswapping in 32bpp mode.

Signed-off-by: Paul Brook <paul@codesourcery.com>

ETRAX: Simplify serport control logic.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Drop _t for the serial type.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Use NULL for subwidth serport reg accesses.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Drop the _t for the internal PIC type.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Simplify the interrupt controller model.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

ETRAX: Simplify interrupt signaling.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

Fix compiler warnings in nwfpe code.

Signed-off-by: Paul Brook <paul@codesourcery.com>

target-mips: proper sign extension for 'SUBU rd, zero, rt'

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

target-mips: fix comments about SUB/DSUB

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

qemu-options.hx: fix description of cache=none suboption

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Suppress type mismatch warnings in VDE code.

Signed-off-by: Paul Brook <paul@codesourcery.com>

minor fix of pci_register_bus()

keep first_bus linked list correct.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>

use PCI_HEADER_TYPE.

use symbolic value instead of 0x0e and related value.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>

Fix typos in comments in exec.c

This patch fixes several typos in comments in exec.c:

            longet -> longer
       recommanded -> recommended
        ajustments -> adjustments
   inconsistancies -> inconsistencies
           phsical -> physical
       positionned -> positioned
       succesfully -> successfully
      regon_offset -> region_offset

and also:

      start_region -> start_addr

Signed-off-by: Stuart Brady <stuart.brady@gmail.com>

Optimize cmp x, 0 case

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Reindent

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Handle NULL bdrv.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Improve instruction name comments for easier searching

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Clarify: dmmuregs[1] is not a typo

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Optimize operations with immediate parameters

Fix Sparc64 sign extension problems

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Fix hw/acpi.c build w/ DEBUG enabled

Trivial build warning/fixes when the local DEBUG define is enabled.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Make sure not to fall through on error in loadvm

This is from the KVM tree

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Pci nic: pci_register_device can fail

The pci_register_device() call in PCI nic initialization routines can
fail.  Handle this failure and propagate a meaningful error message to
the user instead of generating a SEGV.

Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Fix serial option with -drive

This is from the KVM tree.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Build fix for !CONFIG_GDBSTUB case

Once CONFIG_GDBSTUB not configured, compile will generate error.

Signed-off-by: Xiantao Zhang <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

suport device driver initialization model

According to PnP specification, Appendix B, Option ROMs
that support DDIM (device driver initialization model) should
have their memory space writeable.

KVM deviates from us here, by removing the IO_MEM_ROM flag,
to allow for PCI option ROMs (they require DDIM). However,
there's absolutely no reason we can't do the same.

Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

fix virtio_blk_inhdr comment

The inhdr is at the end of the S/G list, not the beginning.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

qemu-io: Optionally verify only part of read data

There are reasonable test cases where a read must span areas that are not
uniformly filled with one pattern but contains several parts. This makes -P
useless for them currently.

Introducing additional options which determine the part of the read data that
should be verified with the given pattern allows to check such reads.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Hardware watchdog

Here is an updated hardware watchdog patch, which should fix
everything that was raised about the previous version ...

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

net: Fix -net socket parameter checks

My commit ea053add700d8abe203cd79a9ffb082aee4eabc0 broke -net socket by
overwriting an intermediate buffer in the added check_param. Fix this
by switching check_param to automatic buffer allocation and release, ie.
callers no longer have to worry about providing a scratch buffer.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

kvm: Avoid COW if KVM MMU is asynchronous

Avi Kivity wrote:
> Suggest wrapping in a function and hiding it deep inside kvm-all.c.
>

Done in v2:

---------->

If the KVM MMU is asynchronous (kernel does not support MMU_NOTIFIER),
we have to avoid COW for the guest memory. Otherwise we risk serious
breakage when guest pages change there physical locations due to COW
after fork. Seen when forking smbd during runtime via -smb.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vnc: kill leftover debug statement.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vnc: no need to set force_update for incremental update requests.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vnc: fix server surface pixel format.

Format must be identical to the guest surface, we can't work with
the 32 bpp used by the default surface allocator.

Without this patch vnc doesn't get the conversions right when sending
pixel data to the client.  The bug triggers if
  (a) the client doesn't support WMVi, and
  (b) the guest screen depth is != 32 bpp.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

virtio-blk: add SGI_IO passthru support

[had the qemu list address wrong the first time, reply to this message,
 not the previous if you were on Cc]

Add support for SG_IO passthru (packet commands) to the virtio-blk
backend.  Conceptually based on an older patch from Hannes Reinecke
but largely rewritten to match the code structure and layering in
virtio-blk.

Note that currently we issue the hose SG_IO synchronously.  We could
easily switch to async I/O, but that would required either bloating
the VirtIOBlockReq by the size of struct sg_io_hdr or an additional
memory allocation for each SG_IO request.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

ppc: include cache-utils.o in BLOCK_OBJS

On ppc, cutils.o needs cache-utils.o or an undefined reference to
qemu_cache_conf results.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vnc: fix segfault

Move down cmp_bytes initialization.  Must be after vga_hw_update()
call, because that one might change the screen depth.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

net: Avoid gcc'ism in net_host_device_add

>> +    if (net_client_init(device, opts ? : "") < 0) {
>
> Is this a gcc extension?  Do we want to introduce this construct to the
> code base.

Valid remark, fix below.

Thanks,
Jan

-------->

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

vnc: windup keypad keys for qemu console emulation

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

kvm: Relax aligment check of kvm_set_phys_mem

There is no need to reject an unaligned memory region registration if
the region will be I/O memory and it will not split an existing KVM
slot. This fixes KVM support on PPC.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Remove redundant cirrus vga ram functions.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Remove redundant #define

Suppress a warning
audio/dsoundaudio.c:35:1: warning: "WIN32_LEAN_AND_MEAN" redefined
<command line>:4:1: warning: this is the location of the previous definition

Signed-off-by: Alex Ivanov <void@aleksoft.net>

Remove cpu_get_io_memory_{read,write}.

Signed-off-by: Paul Brook <paul@codesourcery.com>

Remove vga LFB accessor hacks.

These are redundant now that we remap the LFB ram area.

Signed-off-by: Paul Brook <paul@codesourcery.com>

gitignore qemu-options.texi

$subj

François.

Signed-off-by: François Revol <revol@free.fr>

Fix qemu_event_init

Falling through to "fail" made qemu_event_init() close the pipe fds
immediately again, breaking timer event notification.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Fix powerpc 604 reset vector

According to 604eUM_book (see 8.3.3 Reset inputs p8-54), the IP bit is set
for hreset and the vector is at offset 0x100 from the exception prefix.

No difference in this area between 604 and 604e.

Signed-off-by: Tristan Gingold <gingold@adacore.com>

Fix PPC reset

Silence missing pkg-config error messages

sparc64 fix context value for ITLB fault

Revert previous change to get_physical_address_code:
I/D MMU context register is shared, so using dmmuregs[1] is correct

Signed-off-by: Igor V. Kovalenko <igor.v.kovalenko@gmail.com>

sparc64 fix TLB match code

TLB match code must respect page size, otherwise 4M page mappings may
be not found.

Also correct a typo in get_physical_address_code which should use IMMU
registers.

Signed-off-by: Igor V. Kovalenko <igor.v.kovalenko@gmail.com>
--
Kind regards,
Igor V. Kovalenko

Synch to VGA

[RESEND] Fix vga segfaults or screen corruption with large memory guests

page0 and friends are ram addresses; a smaller size will overflow and
cause a segfault or random corruption.

Change them to ram_addr_t.

Signed-off-by: Avi Kivity <avi@redhat.com>

build system: Silence failing configure tests

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

fix net.c compile warning

Fix net.c compile warning:

CC net.o
net.c: In function \91net_slirp_redir\92:
net.c:623: warning: format not a string literal and no format arguments

Signed-off-by: Robert Reif <reif@earthlink.net>

Recognize V9 stores and CAS accesses as writes

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Fix a warning in sparc64-linux-user build

Merge branch 'master' of blueswir1@git.sv.gnu.org:/srv/git/qemu

sparc64 support TSB related MMU registers

Posting updated patch to the list...

>>> On Fri, Apr 24, 2009 at 9:42 PM, Blue Swirl <blauwirbel@gmail.com> wrote:
>>>  >
>>>  > Nice, though I didn't notice any visible improvement in my tests.
>>>
>>> This early in boot process there is not much to output; and I test
>>>  recent kernel which may use different startup sequence.
>>>  I modified openbios cif handler to output arguments and I now can see
>>>  visible difference.
>>>
>>>
>>>  >
>>>  > About the patch, there are a few problems:
>>>  > - it breaks Sparc32
>>>
>>> You mean it stops working?
>>
>> Does not even build.

Fixed now.

>>>  > - commented out code is ugly
>>>  > - if and else should be on the same line as '{' or '}'
>>>  > - long lines should be wrapped
>>>  > - in the line:
>>>  > +    return (((tag_access_register & 0x1fff)<<48)|(tag_access_register >> 22));
>>>  >  there should be white space between ) and << and 48.
>>>  >
>>>

>>
>> Also the ")|(" in between is crowded.
>>
>> Maybe the coding style does not describe this well enough.

BTW Supplying indent template would be great.

Please see the updated patch qemu-sparc64-tsb-asi-2.patch attached.

--
Kind regards,
Igor V. Kovalenko

CRIS: cris_mmu_result_t -> cris_mmu_result

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>

Fix Win32 warning

Fix compilation when x86_64 is defined

Fix warning

block-vvfat: fix a warning

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

qemu: introduce iothread (Marcelo Tosatti)

Fill in the hooks and introduce iothread.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7248 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: handle stop request in main loop (Marcelo Tosatti)

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7247 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: refactor tcg cpu execution loop (Marcelo Tosatti)

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7246 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: use debug_requested global instead of cpu_exec return (Marcelo Tosatti)

Necessary for the next refactoring patch.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7245 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: introduce lock/unlock_iothread (Marcelo Tosatti)

Hook to allow iothread to drop the global mutex.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7244 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: introduce qemu_cpu_kick (Marcelo Tosatti)

To notify cpu of pending interrupt.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7243 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: introduce qemu_init_vcpu (Marcelo Tosatti)

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7242 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: refactor main_loop (Marcelo Tosatti)

Break main loop into 3 main functions.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7241 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: factor out event notification / rearm alarm timer on main_loop_wait (Marcelo Tosatti)

Special events that have no particular event descriptor (either fd for UNIX
or HANDLE for Windows) associated with make use of an artificial one.

Factor the alarm timer notification so that it can be used for other events,
and move dyntick timer rearm to main_loop_wait.

aliguori: made sure to return a value in qemu_event_init() on win32

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7240 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: introduce qemu_init_main_loop (Marcelo Tosatti)

Hook to allow iothread initialization.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7239 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: per-arch cpu_has_work (Marcelo Tosatti)

Blue Swirl: fix Sparc32 breakage

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7238 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: mutex/thread/cond wrappers and configure tweaks (Marcelo Tosatti)

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7237 c046a42c-6fe2-441c-8c8c-71466251a162

qemu: create helper for event notification (Marcelo Tosatti)

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7236 c046a42c-6fe2-441c-8c8c-71466251a162

Fix typo, thanks to Andreas Faerber for spotting

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7235 c046a42c-6fe2-441c-8c8c-71466251a162

Use a more natural order

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@7234 c046a42c-6fe2-441c-8c8c-71466251a162