sdk/emulator/qemu.git
14 years agolinux-user: commonify definitions of target typedefs
Nathan Froyd [Fri, 11 Dec 2009 17:04:46 +0000 (09:04 -0800)]
linux-user: commonify definitions of target typedefs

There's no sense in separately declaring target_{elf_greg,uid,gid,pid}_t
for every architecture.  Just declare them once with appropriate
USE_UID16 handling.

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agolinux-user: fix ELF_USE_CORE_DUMP/USE_ELF_CORE_DUMP confusion
Nathan Froyd [Fri, 11 Dec 2009 17:04:45 +0000 (09:04 -0800)]
linux-user: fix ELF_USE_CORE_DUMP/USE_ELF_CORE_DUMP confusion

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agolinux-user: use TARGET_ABI_FMT_lx to print abi_ulong types
Aurelien Jarno [Sat, 19 Dec 2009 19:28:23 +0000 (20:28 +0100)]
linux-user: use TARGET_ABI_FMT_lx to print abi_ulong types

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoroms: allow roms to be loaded at address 0
Aurelien Jarno [Sat, 19 Dec 2009 18:14:16 +0000 (19:14 +0100)]
roms: allow roms to be loaded at address 0

It was possible to load roms at address 0, but commit
632cf034b401cdd01dae253a8b577fe518e37654 started to forbid that, which
broke at least ARM versatile.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agolinux-user: cleanup force_sig() calls
Riku Voipio [Fri, 4 Dec 2009 13:16:32 +0000 (15:16 +0200)]
linux-user: cleanup force_sig() calls

Force_sig should be always called with TARGET_ signals.
Not that it really matters with SEGV, so this patch is
just for cleanup and improving consistency.

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agolinux-user: Fix mmap_lock ordering
Riku Voipio [Fri, 4 Dec 2009 13:16:31 +0000 (15:16 +0200)]
linux-user: Fix mmap_lock ordering

mmap_lock() can be called while tb_lock() is being held. To
avoid deadlock when one thread is holding mmap_lock and another
tb_lock, _always_ lock first tb_lock().

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agolinux-user: enable tb unlinking when compiled with NPTL
Riku Voipio [Fri, 4 Dec 2009 13:16:30 +0000 (15:16 +0200)]
linux-user: enable tb unlinking when compiled with NPTL

Fixes receiving signals when guest code is being executed in a tight
loop. For an example, try interrupting the following code with ctrl-c.

http://nchipin.kos.to/test-loop.c

The tight loop is ofcourse brainless, but it is also exactly how the waitpid* testcases
are implemented.

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agouser: fix compilation without -fpie
Blue Swirl [Sat, 19 Dec 2009 15:42:34 +0000 (15:42 +0000)]
user: fix compilation without -fpie

Fixes breakage by 299060a06e7c73de5cb9962edbca7f7f9e0b7f25.

Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
14 years agouser_only: compile everything with -fpie
Kirill A. Shutemov [Sat, 19 Dec 2009 15:16:08 +0000 (15:16 +0000)]
user_only: compile everything with -fpie

We really need compile _all_ sources for user target with -fpie when
use --enable-user-pie.

It's regression introduced by commit add16157d72454.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Blue Swirl <blauwirbel@gmail.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
14 years agohw/openpic.c: replace tabs by spaces
Aurelien Jarno [Sat, 19 Dec 2009 14:59:29 +0000 (15:59 +0100)]
hw/openpic.c: replace tabs by spaces

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-ppc: Use interrupts for escc
Alexander Graf [Fri, 18 Dec 2009 22:37:27 +0000 (23:37 +0100)]
target-ppc: Use interrupts for escc

When using the escc with Linux, we need interrupts. So instead of creating
a dummy device, let's just map them to the openpic we have anyways.

This makes Linux on PPC64 with console=ttyPZ0 work.

Obviously, this change needs to be reflected in openbios. Patch for that
follows this one. Please update the binary then.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-ppc: Initialize OpenPIC properly
Alexander Graf [Fri, 18 Dec 2009 22:37:26 +0000 (23:37 +0100)]
target-ppc: Initialize OpenPIC properly

When lowering an IRQ line, we search for the line we're supposed to lower.

Usually we run into an optimization there that queues up interrupts. This
queue ends with -1. Unfortunately we didn't set the first item to -1.

This patch fixes this, making interrupts work on PPC64.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoUpdate version and changelog to sync with stable branch
Anthony Liguori [Sat, 19 Dec 2009 14:24:34 +0000 (08:24 -0600)]
Update version and changelog to sync with stable branch

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoUpdate to SeaBIOS 0.5.0
Anthony Liguori [Sat, 19 Dec 2009 03:26:15 +0000 (21:26 -0600)]
Update to SeaBIOS 0.5.0

The only change is updating the makefile but that way we're carrying an official
release.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoRevert "monitor: Convert do_migrate_set_speed() to QObject"
Anthony Liguori [Fri, 18 Dec 2009 20:51:49 +0000 (14:51 -0600)]
Revert "monitor: Convert do_migrate_set_speed() to QObject"

This reverts commit 3a4921047d514c336531c304bec362c4d138ebcc.

From Luiz:

  do_migrate_set_speed() accepts a suffix for the 'value' argument and this is
  not good for QMP.  We will have to add a new argument type to handle that and
  this will have to wait for 0.13.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoe1000: Don't muck with PCI commmand register
Anthony Liguori [Fri, 18 Dec 2009 19:58:56 +0000 (13:58 -0600)]
e1000: Don't muck with PCI commmand register

Otherwise, the driver does not work in Linux after the INT_DISABLE changes in
PCI.

Michael Tsirkin had a patch to do this, I'm not sure what happened to it.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: do_balloon(): Use 'M' argument type
Luiz Capitulino [Fri, 18 Dec 2009 15:25:05 +0000 (13:25 -0200)]
monitor: do_balloon(): Use 'M' argument type

This makes do_balloon() accept megabyte values from the user
Monitor while accepting byte values for QMP.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: Introduce 'M' argument type
Luiz Capitulino [Fri, 18 Dec 2009 15:25:04 +0000 (13:25 -0200)]
monitor: Introduce 'M' argument type

This is a target long value in megabytes which should be
converted to bytes.

It will be used by handlers which accept a megabyte value
when in "user mode".

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQMP: Update spec file
Luiz Capitulino [Fri, 18 Dec 2009 15:25:03 +0000 (13:25 -0200)]
QMP: Update spec file

- Remove "draft" status
- Change default success response to be json-object
- Change error and event data member to be a json-object
- Update examples
- Add new section "Compatibility Considerations"
- Other fixes and clarifications

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQMP: Update README file
Luiz Capitulino [Fri, 18 Dec 2009 15:25:02 +0000 (13:25 -0200)]
QMP: Update README file

- Fix output description
- Fix command-line usage notes
- Minor improvements

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQMP: Assure that returned data is a QDict
Luiz Capitulino [Fri, 18 Dec 2009 15:25:01 +0000 (13:25 -0200)]
QMP: Assure that returned data is a QDict

This is for debug purposes only.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQMP: Return an empty dict by default
Luiz Capitulino [Fri, 18 Dec 2009 15:25:00 +0000 (13:25 -0200)]
QMP: Return an empty dict by default

Currently, when a regular command doesn't have any data to output,
QMP will emit:

{ "return": "OK" }

Returning an empty dict is better though, because dicts can support
some protocol changes in a compatible way.

So, with this commit we will return:

{ "return": {} }

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQMP: Only handle converted commands
Luiz Capitulino [Fri, 18 Dec 2009 15:24:59 +0000 (13:24 -0200)]
QMP: Only handle converted commands

Looks like I dropped this check when addressing the 'query-'
commands request.

QMP should only handle converted commands, obviously.

Reported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agotarget-ppc: fix ppc32 kvm build
Alexander Graf [Sat, 19 Dec 2009 00:58:59 +0000 (01:58 +0100)]
target-ppc: fix ppc32 kvm build

My segment sync patch broke compilation on PPC32, because it was trying to
sync the SLB even though ppc32 CPUs don't have an SLB.

So let's only sync it when we're on a PP64 one!

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agousb-musb: convert fifo to 8bit and add more registers
Riku Voipio [Thu, 3 Dec 2009 13:56:08 +0000 (15:56 +0200)]
usb-musb: convert fifo to 8bit and add more registers

Convert musb fifo to 8bit to allow 8/16/32bit access

MUSB allows reading and writing to the fifo in 32/16/8 bit
width. The Linux kernel does this sometimes, most usually at
the end of writing the packet to allow packet to end at a
odd bytecount.

Convert the fifo to 8bit allows removing lots of shifts
which shows that the fifo is more natural as 8bit.

While at it, add multiple missing register definitions and
and cleanup debug prints.

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agohandle SD CMD5 without error messages
Juha Riihimäki [Thu, 3 Dec 2009 13:56:07 +0000 (15:56 +0200)]
handle SD CMD5 without error messages

From: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoMake USB hid devices self-powered
Riku Voipio [Thu, 3 Dec 2009 13:56:06 +0000 (15:56 +0200)]
Make USB hid devices self-powered

Simplifies power budget negotiation.

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoGive a error when running out of iomem areas.
Riku Voipio [Thu, 3 Dec 2009 13:56:05 +0000 (15:56 +0200)]
Give a error when running out of iomem areas.

The limit of iomem areas is quite low. Without the
debug print, it is quite hard to figure out why more
devices are not getting registered.

Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agofix networking on win32 host
Juha Riihimäki [Thu, 3 Dec 2009 13:56:04 +0000 (15:56 +0200)]
fix networking on win32 host

At least under some mingw compilers slirp networking fails without declaring
these fields packed.

From: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agofix pidfile option to work in WIN32
Juha Riihimäki [Thu, 3 Dec 2009 13:56:03 +0000 (15:56 +0200)]
fix pidfile option to work in WIN32

Explicit read/write locking pidfile under WIN32 is bit extreme
nobody get the chance to read the pidfile. Convert to a write-only lock.

Also, creating pidfile was disabled along with daemonize under
WIN32. Enable it, but do not enable daemon support which doesn't
exist under WIN32 atm.

From: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoFix win32 log file location
Juha Riihimäki [Thu, 3 Dec 2009 13:56:02 +0000 (15:56 +0200)]
Fix win32 log file location

/tmp doesn't exist under win32. Ease the pain of win32 development slightly.

From: Juha Riihimäki <juha.riihimaki@nokia.com>
Signed-off-by: Riku Voipio <riku.voipio@nokia.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoUpdate SeaBIOS to include PCI based option rom loading
Anthony Liguori [Fri, 18 Dec 2009 15:42:56 +0000 (09:42 -0600)]
Update SeaBIOS to include PCI based option rom loading

Also remove pcbios from the tree.  It will no longer work.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoroms: remove option rom packing logic
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:12 +0000 (12:01 +0100)]
roms: remove option rom packing logic

Now that we load the option roms via fw_cfg, we can stop copying
them to the 0xc000 -> 0xe000.  The patch does just that.

Also the rom loader gets simplified as all remaining users of the
rom loader load the bits at a fixed address so the packing and
aligning logic can go away.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoroms: use new fw_cfg file xfer support.
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:11 +0000 (12:01 +0100)]
roms: use new fw_cfg file xfer support.

roms: use fw_cfg for vgabios and option rom loading, additionally to
deploying them the traditional way (copy to 0xc0000 -> 0xe0000 range).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agofw_cfg: add API for file transfer.
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:10 +0000 (12:01 +0100)]
fw_cfg: add API for file transfer.

This patch adds a file transfer interface to fw_cfg.  Intended to be
used for passing non-pci option roms and vgabios to seabios.  Namespace
is modeled after the existing cbfs filesystem support in seabios.

Reading the new FW_CFG_FILE_DIR entry returns a file list.
Fields there are in network byte order (aka bigendian).

aliguori: fix fw_cfg.h for multiboot.bin, add proper fw_cfg.h declarations,
          quiet fprintf() in fw_cfg.c

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agofw_cfg: make calls typesafe
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:09 +0000 (12:01 +0100)]
fw_cfg: make calls typesafe

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agopci romfiles: add property, add default to PCIDeviceInfo
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:08 +0000 (12:01 +0100)]
pci romfiles: add property, add default to PCIDeviceInfo

This patch adds a romfile property to the pci bus.  It allows to specify
a romfile to load into the rom bar of the pci device.  The default value
comes from a new field in PCIDeviceInfo.  The property allows to change
the file and also to disable the rom loading using an empty string.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoSupport PCI based option rom loading
Anthony Liguori [Fri, 18 Dec 2009 11:01:07 +0000 (12:01 +0100)]
Support PCI based option rom loading

Currently, we preload option roms into the option rom space in memory.  This
prevents DDIM from functioning correctly which severely limits the number
of roms we can support.

This patch introduces a pci_add_option_rom() which registers the
PCI_ROM_ADDRESS bar which points to our option rom.  It also converts over
the cirrus vga adapter, the rtl8139, virtio, and the e1000 to use this
new mechanism.

The result is that PXE boot functions even with three unique types of cards.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoFix backcompat for hotplug of SCSI controllers
Daniel P. Berrange [Mon, 7 Dec 2009 20:51:49 +0000 (20:51 +0000)]
Fix backcompat for hotplug of SCSI controllers

SCSI controllers have no trouble existing without any attached
disks. This could be achieved with the (legacy) monitor syntax

  pci_add pci_addr=auto storage if=scsi

This is now denied with

  scsi requires a backing file/device.
  failed to add if=scsi

There is no need for this denial and it breaks compatability
with existing QEMU usage, so remove the check for presence
of a drive.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agofdc: fix migration from 0.11
Juan Quintela [Tue, 15 Dec 2009 13:34:35 +0000 (14:34 +0100)]
fdc: fix migration from 0.11

0.11 uses as instance ide io_base, get it back

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoRevert "fdc: fix vmstate variable passed"
Juan Quintela [Tue, 15 Dec 2009 13:34:34 +0000 (14:34 +0100)]
Revert "fdc: fix vmstate variable passed"

Floppy used the io_base address to register savevm region.

This reverts commit 2966b390d0f7cd5e5b971290fcc20685de374228.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: Accept input only byte-wise
Jan Kiszka [Fri, 4 Dec 2009 13:05:29 +0000 (14:05 +0100)]
monitor: Accept input only byte-wise

This allows to suspend command interpretation and execution
synchronously, e.g. during migration.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoRevert "kvm: x86: Save/restore exception_index"
Anthony Liguori [Fri, 18 Dec 2009 14:16:30 +0000 (08:16 -0600)]
Revert "kvm: x86: Save/restore exception_index"

This reverts commit ebbc8a3d8e76d0402f8a08c10c0f32e24715d41d.

As suggested by Jan Kiszka,

  "It was obsoleted by d1793b836f8f123b961c613de1bb1c0c185c84cc and now
   saves/restores a useless field."

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agovmware: increase cursor buffer size.
Dave Airlie [Thu, 17 Dec 2009 22:08:11 +0000 (08:08 +1000)]
vmware: increase cursor buffer size.

The cursor pixmap size we calculate later ends up being 4096 dwords
long by the looks of it. This boots an F12 LiveCD now.

Signed-off-by: Dave Airlie <airlied@linux.ie>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoVMware VGA: Only enable dirty log tracking when fifo is disabled
Anthony Liguori [Thu, 17 Dec 2009 22:08:10 +0000 (08:08 +1000)]
VMware VGA: Only enable dirty log tracking when fifo is disabled

This patch enables dirty log tracking whenever it's needed and disables it
when it is not.

We unconditionally enable dirty log tracking on reset, restart dirty log
tracking when PCI IO regions are remapped, and disable/enable it based on
commands from the guest.

Rebased-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoFix VMware VGA depth computation
Anthony Liguori [Thu, 17 Dec 2009 22:08:09 +0000 (08:08 +1000)]
Fix VMware VGA depth computation

VMware VGA requires that the depth presented to the guest is the same as the
DisplaySurface that it renders to.  This is because it performs a very simple
memcpy() to blit from one surface to another.

We currently hardcode a 24-bit depth.  The surface allocator for SDL may, and
usually will, allocate a surface with a different depth causing screen
corruption.

This changes the code to allocate the DisplaySurface before initializing the
device which allows the depth of the DisplaySurface to be used instead of
hardcoding something.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoMake sure to enable dirty log tracking for VMware VGA
Anthony Liguori [Thu, 17 Dec 2009 22:08:08 +0000 (08:08 +1000)]
Make sure to enable dirty log tracking for VMware VGA

This is needed for VMware VGA to work properly under KVM.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoMake sure to enable dirty tracking of VBE vram mapping
Anthony Liguori [Thu, 17 Dec 2009 22:08:07 +0000 (08:08 +1000)]
Make sure to enable dirty tracking of VBE vram mapping

Apparently, VBE maps the VGA vram to a fixed physical location.  KVM requires
that all mappings of the VGA vram have dirty tracking enabled on them.  Any
access to the VGA vram through the VBE mapping currently fails to result in
dirty page tracking updates causing a black screen.

This is the true root cause of VMware VGA not working correctly under KVM and
likely also an issue with some of the std-vga black screen issues too.

Cirrus does not enable VBE so it would not be a problem when using Cirrus.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Rebased-by: Dave Airlie <airlied@redhat.com>
14 years agovmware: setup PCI BAR 2 for FIFO as per vmware spec
Dave Airlie [Thu, 17 Dec 2009 22:08:06 +0000 (08:08 +1000)]
vmware: setup PCI BAR 2 for FIFO as per vmware spec

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoqdev: improve property error reporting.
Gerd Hoffmann [Wed, 16 Dec 2009 13:22:11 +0000 (14:22 +0100)]
qdev: improve property error reporting.

Add a error message in case we fail to parse a qdev property.

Also make qemu not abort() in case setting a global property can't be
set.  This used to be a clear programming error.  The introduction of
the -global switch changed that though, so better exit instead (after
printing the new error message).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agofix vga names in default_list
Gerd Hoffmann [Wed, 16 Dec 2009 12:35:19 +0000 (13:35 +0100)]
fix vga names in default_list

Fix mismerge between 64465297 and 556cd098.

Cc: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agousb-host: check mon before using it.
Gerd Hoffmann [Tue, 15 Dec 2009 10:43:02 +0000 (11:43 +0100)]
usb-host: check mon before using it.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agousb-net: use qdev for -usbdevice
Gerd Hoffmann [Tue, 15 Dec 2009 09:20:57 +0000 (10:20 +0100)]
usb-net: use qdev for -usbdevice

Rebased to master, adapted to device renaming by armbru,
no other changes.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoCheck rom_load_all() return value.
Gerd Hoffmann [Mon, 14 Dec 2009 15:07:35 +0000 (16:07 +0100)]
Check rom_load_all() return value.

Check rom_load_all() return value.
Also don't make option rom loading failure fatal.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agodefaults: update device_list[]
Gerd Hoffmann [Wed, 16 Dec 2009 13:25:40 +0000 (14:25 +0100)]
defaults: update device_list[]

Add isa-fdc (disables default_floppy).
Add ide-drive (disables default_cdrom).

Also walk the -global QemuOpts, so we'll catch
-global isa-fdc.drive{A,B}=<name> too.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agodefaults: split default_drive
Gerd Hoffmann [Wed, 16 Dec 2009 13:25:39 +0000 (14:25 +0100)]
defaults: split default_drive

Split default_drive into default_{floppy,cdrom,sdcard}.
Also add QEMUMachine flags to disable them per machine.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: Catch printing to non-existent monitor
Luiz Capitulino [Mon, 14 Dec 2009 20:53:24 +0000 (18:53 -0200)]
monitor: Catch printing to non-existent monitor

The monitor_vprintf() function now touches the 'mon' pointer
before calling monitor_puts(), this causes block migration
to segfault as its functions call monitor_printf() with a
NULL 'mon'.

To fix the problem this commit moves the 'mon' NULL check
from monitor_puts() to monitor_vprintf().

This can potentially hide bugs, but for some reason this has
been the behavior for a long time.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: Avoid readline functions in QMP
Luiz Capitulino [Mon, 14 Dec 2009 20:53:23 +0000 (18:53 -0200)]
monitor: Avoid readline functions in QMP

The monitor_read_command() function is readline specific
and should only be used when readline is available.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: do_balloon(): Check for errors
Luiz Capitulino [Mon, 14 Dec 2009 20:53:22 +0000 (18:53 -0200)]
monitor: do_balloon(): Check for errors

do_balloon() should check for ballooning availability as
do_info_balloon() does.

Noted by Daniel P. Berrange <berrange@redhat.com>.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agomonitor: Use 'device' in eject
Luiz Capitulino [Mon, 14 Dec 2009 20:53:21 +0000 (18:53 -0200)]
monitor: Use 'device' in eject

Monitor's eject command uses 'filename' for the device name
argument, but 'device' is a better name.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoQDict: Fix size update
Luiz Capitulino [Mon, 14 Dec 2009 20:53:20 +0000 (18:53 -0200)]
QDict: Fix size update

Key replacement should not update the dictionary's size.

This commit also adds a test for the bug.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoqdev: Improve uni-north device names
Markus Armbruster [Mon, 14 Dec 2009 09:41:21 +0000 (10:41 +0100)]
qdev: Improve uni-north device names

Switch to the names suggested by Blue Swirl.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoAvoid permanently disabled QEMU monitor when UNIX migration fails
Daniel P. Berrange [Fri, 11 Dec 2009 21:01:14 +0000 (21:01 +0000)]
Avoid permanently disabled QEMU monitor when UNIX migration fails

If a UNIX migration command is attempt to a UNIX socket which does
not exist, then the monitor is suspended, but never resumed. This
prevents any further use of the monitor

* migration-unix.c: Only call migrate_fd_monitor_suspend() once
  connected to the UNIX socket.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoFix loading of ELF multiboot kernels
Kevin Wolf [Fri, 4 Dec 2009 16:19:25 +0000 (17:19 +0100)]
Fix loading of ELF multiboot kernels

The multiboot implementation assumed that there is only one program header
(which contains the entry point) and that the entry point is at the start of
the code. This doesn't hold true generally and caused too little data to be
loaded.

Fix the loading code to pass the whole loaded data to the Multiboot Option ROM.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoRevert "Rename DriveInfo.onerror to on_write_error" (fix mismerge)
Kevin Wolf [Fri, 4 Dec 2009 11:06:32 +0000 (12:06 +0100)]
Revert "Rename DriveInfo.onerror to on_write_error" (fix mismerge)

Part of the first patch of the -drive rerror series has been merged once more
on top of the rest of the series. This effectively disables the rerror option
and always goes with the default value. Reverting the commit re-enables the
option.

This reverts commit fc072ec4df0996682dfbff6c735e2bbc0d93132f.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoqemu-io: Fix memory leak
Kevin Wolf [Wed, 18 Nov 2009 09:42:59 +0000 (10:42 +0100)]
qemu-io: Fix memory leak

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoFix thinko in linuxboot.S
Paolo Bonzini [Sun, 13 Dec 2009 10:36:40 +0000 (11:36 +0100)]
Fix thinko in linuxboot.S

The %gs segment that was used was not matching the comments.
I just moved the GDT descriptor on the stack instead.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agotarget-i386: Fix evaluation of DR7 register
Jan Kiszka [Mon, 14 Dec 2009 11:26:27 +0000 (12:26 +0100)]
target-i386: Fix evaluation of DR7 register

hw_breakpoint_type and hw_breakpoint_len used the wrong index multiplier
to extract type and len.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agokvm: x86: Use separate exception_injected CPUState field
Jan Kiszka [Mon, 14 Dec 2009 11:26:17 +0000 (12:26 +0100)]
kvm: x86: Use separate exception_injected CPUState field

Marcelo correctly remarked that there are usage conflicts between QEMU
core code and KVM /wrt exception_index. So spend a separate field and
also save/restore it properly.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agovnc: hextile: do not generate ForegroundSpecified and SubrectsColoured tiles
Anthony Liguori [Mon, 14 Dec 2009 20:11:56 +0000 (14:11 -0600)]
vnc: hextile: do not generate ForegroundSpecified and SubrectsColoured tiles

This violates the RFB specification (section 6.6.4).  It happens to work with
most clients but it's still wrong.

Reported-by: Yaniv Kaul <ykaul@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoRevert "pci: interrupt disable bit support"
Anthony Liguori [Mon, 14 Dec 2009 17:36:53 +0000 (11:36 -0600)]
Revert "pci: interrupt disable bit support"

This reverts commit 0ea5709a32085f7d14901a09d12bd35f9b267607.

Per discussion with Michael Tsirkin, this is too risky for 0.12

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
14 years agoS390: Bail out without KVM
Alexander Graf [Fri, 18 Dec 2009 15:29:04 +0000 (16:29 +0100)]
S390: Bail out without KVM

Currently only the S390 KVM target works. To keep users from accidently not
using KVM, let's not even initialize the machine when KVM is not used.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoS390: Don't tell guest we're updating config space
Alexander Graf [Thu, 17 Dec 2009 12:06:09 +0000 (13:06 +0100)]
S390: Don't tell guest we're updating config space

Currently we always set the "config space changed" bit to 1 when triggering
any virtio interrupt. While that worked in 2.6.27, newer kernels interpret
that value as "only the config space changed and nothing else happened".

Since we usually trigger interrupts to tell the guest that something did
happen, we just not tell it the config space changed for now until we
implement the correct callback for that.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoadd default virtcon initialization
Alexander Graf [Thu, 17 Dec 2009 12:06:08 +0000 (13:06 +0100)]
add default virtcon initialization

When going through the default devices, we don't initialize the virtio
console, unless we're doing -nographic.

I suppose that's just a leftover from the recent code restructuring, so
let's put it in.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoS390: Loop through virtio console devices
Alexander Graf [Thu, 17 Dec 2009 12:06:07 +0000 (13:06 +0100)]
S390: Loop through virtio console devices

We used to always create one single virtio console device. This breaks when
either zero of multiple virtio console devices are requested, so let's use
the same code as on x86.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-s390: Fail on unknown instructions
Alexander Graf [Thu, 17 Dec 2009 12:56:47 +0000 (13:56 +0100)]
target-s390: Fail on unknown instructions

We were being a bit too nice and didn't give the guest an invalid instruction
interrupt.

While that works, it's not exactly the fastest thing to do, since now the
guest doesn't know that we're not really implementing that instruction, so it
continues doing it.

We run into this with the set_page_unstable hint instruction. So let's bail out
in these cases.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoosdep: Fix runtime failure on older Linux kernels
Andre Przywara [Fri, 18 Dec 2009 09:45:07 +0000 (10:45 +0100)]
osdep: Fix runtime failure on older Linux kernels

If QEMU finds newer kernel header files on compilation time, it will use
advertised features like pipe2 or SOCK_CLOEXEC by just doing a compile test.
If later the executables are executed on an older kernel (<2.6.27,
like Xen Dom0 2.6.18), then QEMU will fail on opening sockets and creating
pipes and returns the rather unspecific "qemu_init_main_loop failed".
This patch fixes this by checking the return values of these calls
for EINVAL and ENOSYS and falling back to the older versions automatically.

Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agoFix a make -j race
Juergen Lock [Mon, 14 Dec 2009 20:54:43 +0000 (21:54 +0100)]
Fix a make -j race

Make libuser.a depend on $(GENERATED_HEADERS) too so make -j won't start
building it before the headers exist.  (There may be more bugs like this
but at least this makes (g)make -j4 started from scratch on a quadcore
now always complete here again.)

Signed-off-by: Juergen Lock <nox@jelal.kn-bremen.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix float32_to_s vs zero exponent.
Richard Henderson [Mon, 14 Dec 2009 01:50:46 +0000 (17:50 -0800)]
target-alpha: Fix float32_to_s vs zero exponent.

There was a bug in float32_to_s that incorrectly mapped a zero exponent
to 0x38.  This meant 0.0f != 0.  At the same time, fix a generic type
punning bug in helper_memory_to_s and helper_s_to_memory.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix cvtlq.
Richard Henderson [Mon, 14 Dec 2009 01:48:55 +0000 (17:48 -0800)]
target-alpha: Fix cvtlq.

We were missing the 0xc0000000 mask, leading to incorrect results.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix generic ctz64.
Richard Henderson [Mon, 14 Dec 2009 01:47:25 +0000 (17:47 -0800)]
target-alpha: Fix generic ctz64.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Implement fp branch/cmov inline.
Richard Henderson [Mon, 14 Dec 2009 01:46:17 +0000 (17:46 -0800)]
target-alpha: Implement fp branch/cmov inline.

The old fcmov implementation had a typo:
-        tcg_gen_mov_i64(cpu_fir[rc], cpu_fir[ra]);
which moved the condition, not the second source, to the destination.

But it's also easy to implement the simplified fp comparison inline.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Add placeholders for missing userspace PALcalls.
Richard Henderson [Mon, 14 Dec 2009 01:18:47 +0000 (17:18 -0800)]
target-alpha: Add placeholders for missing userspace PALcalls.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-mips: No MIPS16 support for 4Kc, 4KEc cores
Stefan Weil [Tue, 15 Dec 2009 13:03:03 +0000 (14:03 +0100)]
target-mips: No MIPS16 support for 4Kc, 4KEc cores

Fix regression introduced by d19954f46dfc262612c30e9534e660e953049487.

4Kc and 4KEc don't support MIPS16.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix compiler warning for gcc-4.3 (and older)
Stefan Weil [Mon, 14 Dec 2009 09:54:49 +0000 (10:54 +0100)]
target-alpha: Fix compiler warning for gcc-4.3 (and older)

"Old" compilers obviously are not able to recognise
that all cases are handled here:

qemu/target-alpha/helper.c:70: error: ‘round_mode’ may be used uninitialized in this function

A small modification helps the compiler to do its jobs.

gcc-4.4 does not need this, but is still not standard on all platforms.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Acked-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agos390: Fix buggy assignment
Stefan Weil [Mon, 14 Dec 2009 09:39:12 +0000 (10:39 +0100)]
s390: Fix buggy assignment

nd->model keeps dynamically allocated model names.
So casting of a constant string is wrong here.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Acked-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-mips: 4Kc, 4KEc cores do not support MIPS16
Stefan Weil [Tue, 15 Dec 2009 13:43:40 +0000 (14:43 +0100)]
target-mips: 4Kc, 4KEc cores do not support MIPS16

4Kc, 4KEc cores do not support MIPS16, so not only the
CP0_Config1 had to be fixed (see previous patch),
but also MIPS16 instructions must not be executed.

(Hint from Nathan Froyd, thanks).

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agomicroblaze: Print content of EAR register
Michal Simek [Mon, 7 Dec 2009 13:12:49 +0000 (14:12 +0100)]
microblaze: Print content of EAR register

I need to see EAR register in output because I need to parse irqsoff problem.

Signed-off-by: Michal Simek <monstr@monstr.eu>
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
14 years agomicroblaze: Update debug logs.
Edgar E. Iglesias [Wed, 16 Dec 2009 11:52:56 +0000 (12:52 +0100)]
microblaze: Update debug logs.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
14 years agotcg/ppc64: Fix loading of 32bit constants
malc [Tue, 15 Dec 2009 16:44:20 +0000 (19:44 +0300)]
tcg/ppc64: Fix loading of 32bit constants

Signed-off-by: malc <av1474@comtv.ru>
14 years agotarget-alpha: Fix double log_cpu_state.
Richard Henderson [Fri, 11 Dec 2009 21:23:52 +0000 (13:23 -0800)]
target-alpha: Fix double log_cpu_state.

The proper logging is handled by generic code.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix FMOV.
Richard Henderson [Fri, 11 Dec 2009 21:21:02 +0000 (13:21 -0800)]
target-alpha: Fix FMOV.

Properly handle move from the zero register.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Expand ins*h inline.
Richard Henderson [Sat, 12 Dec 2009 00:31:29 +0000 (16:31 -0800)]
target-alpha: Expand ins*h inline.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Expand msk*h inline.
Richard Henderson [Fri, 11 Dec 2009 19:58:29 +0000 (11:58 -0800)]
target-alpha: Expand msk*h inline.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Expand msk*l inline.
Richard Henderson [Fri, 11 Dec 2009 19:51:45 +0000 (11:51 -0800)]
target-alpha: Expand msk*l inline.

Similar in difficulty to ext*l, already expanded.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Expand ins*l inline.
Richard Henderson [Fri, 11 Dec 2009 18:39:56 +0000 (10:39 -0800)]
target-alpha: Expand ins*l inline.

Similar in difficulty to ext*l, already expanded.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Implement RD/WRUNIQUE in the translator
Richard Henderson [Fri, 11 Dec 2009 17:38:23 +0000 (09:38 -0800)]
target-alpha: Implement RD/WRUNIQUE in the translator

When emulating user-mode only, there's no reason to exit
the translation block to effect a call_pal.  We can generate
a move to/from the unique slot directly.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Fix fbcond branch offset.
Richard Henderson [Fri, 11 Dec 2009 17:07:29 +0000 (09:07 -0800)]
target-alpha: Fix fbcond branch offset.

The instructions use a disp21 like all other branch insns,
not the disp16 that was being passed.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Rewrite gen_ext_[hl] in terms of zapnot.
Richard Henderson [Thu, 10 Dec 2009 22:00:18 +0000 (14:00 -0800)]
target-alpha: Rewrite gen_ext_[hl] in terms of zapnot.

The architecture manual specifies the EXT instructions
in terms of the ZAPNOT operation; writing it that way in
the translator makes things a bit clearer.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
14 years agotarget-alpha: Expand zap/zapnot with immediate inline.
Richard Henderson [Thu, 10 Dec 2009 21:43:58 +0000 (13:43 -0800)]
target-alpha: Expand zap/zapnot with immediate inline.

The vast majority of zap instructions have an immediate operand,
since zapnot is the canonical method to zero-extend from u16 or u32.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>