hw/i386/Makefile.obj: use $(PYTHON) to run .py scripts consistently

$(PYTHON) is a Make variable which is set by configure.
In all other places over the tree, .py files are run from
Makefiles using this variable, except of a single leftover
in hw/i386/Makefile.obj (and a nearby place in there uses
$(PYTHON) correctly).  Fix this leftover too.

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Cc: Luigi Rizzo <rizzo@iet.unipi.it>
Reviewed-by:: Stefan Weil <sw@weilnetz.de>

configure: Use -B switch only for Python versions which support it

Commit 1d984a67a95d88f3e708b077dab8adeb47c38c93 added the -B switch
unconditionally. This breaks Python versions before 2.6 which don't
support that switch.

Now configure adds -B only if it is accepted by the Python interpreter.

This modification introduces a small incompatibility because -B might now
also be added when configure was called with --python=PYTHON_INTERPRETER.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qga: Fix shutdown command of guest agent to work with SysV

For now guest agent uses following command to shutdown system:
shutdown -P +0 "blabla"
but this syntax works only with shutdown command from systemd or upstart,
because SysV shutdown requires -h switch.

Following patch changes the command so it works with systemd, upstart and SysV

With upstart/systemd qga use one of thee commands, depending on 'mode' parameter:
  shutdown -P +0 "..."
  shutdown -H +0 "..."
  shutdown -r +0 "..."
SysV equivalents for these are:
  shutdown -h -P +0 "..."
  shutdown -h -H +0 "..."
  shutdown -h -r +0 "..."
and these retain their meaning with upstart/systemd.

According to FreeBSD manpages, shutdown does not accept -P and -H options. Commands should be:
  shutdown -p +0 "..."
  shutdown -h +0 "..."
  shutdown -r +0 "..."

shutdown in Solaris does not accept any of -hHpPr and does not accept time in "+0" format

Signed-off-by: Michael Avdienko <whitearchey@gmail.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

console: Remove unused debug code

The local function console_print_text_attributes is no longer used since
commit 7d6ba01c3741bc32ae252bf64a5fd3f930c2df4f.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qga: Fix compilation for old versions of MinGW

While MinGW-w64 can compile the qga code, MinGW from Debian lenny
(gcc-mingw32 4.4.2-3) shows these errors:

In file included from qga/vss-win32.c:17:
qga/vss-win32/requester.h:31:
 error: expected »=«, »,«, »;«, »asm« or »__attribute__« before »requester_init«
qga/vss-win32/requester.h:32:
 error: expected »=«, »,«, »;«, »asm« or »__attribute__« before »requester_deinit«

The macro STDAPI is unknown, so add the missing include file which
defines it.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

.travis.yml: basic compile and check recipes

This adds a build matrix definition for travis-ci.org continuous
integration service. It is usable on any public repository hosted on
GitHub. Once you have created an account signed into Travis you can
enable it on selected projects via travis-ci.org/profile. Alternatively
you can configure the service hooks on GitHub via the repository
Settings tab,then Service Hooks and selecting Travis.

Once setup Travis will automatically test every push as well as any pull
requests submitted to that repository.

The build matrix is currently split by target architecture (see TARGETS
environment variable) because a full build of QEMU can take some time.
This way you get quick feedback for any obvious errors. The additional
environment variables exist to allow additional builds to tweak the
environment. These are:

EXTRA_CONFIG - extra terms passed to configure
EXTRA_PKGS - extra dev packages to install
TEST_CMD - default "make check", can be overridden

I've confined the additional stuff to x86/x86_64 for convenience.

As Travis supports clang the main builds are done twice (once for gcc
and once for clang). However clang is disabled for the debug/trace
builds for the purposes of brevity.

Other wrinkles:

 * The lttng user-space tracing back-end is disabled
   (it is currently horribly broken)
 * The ftrace back-end doesn't run "make check"
   (it requires a mounted debugfs to work)
 * There are two debug enabled build (with and without TCG interpreter)

Signed-off-by: Alex Bennée <alex@bennee.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

pci-assign: Fix error_report of pci-stub message

Using multiple calls to error_report here means every line is
prefaced with the (potentially long) pci-assign command line
arguments.

Use a single error_printf to preserve the intended formatting.
Since this code path is always preceded by an error_report call,
we don't lose the command line reporting.

Signed-off-by: Cole Robinson <crobinso@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qapi: Fix comment for create-type to match code.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

vl: fix build when configured with no graphic support

The following error occurs when building with no graphic output support:

  vl.c: In function ‘main’:
  vl.c:2829:19: error: variable ‘ds’ set but not used [-Werror=unused-but-set-variable]
       DisplayState *ds;
                     ^
  cc1: all warnings being treated as errors

To reproduce this issue, just run:

  $ ./configure \
      --disable-curses \
      --disable-sdl \
      --disable-cocoa \
      --disable-gtk \
      --disable-vnc \
      --enable-werror
  $ make vl.o

Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

usb: drop unused USBNetState.inpkt field

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qemu-char: add missing characters used in keymaps

This patch adds all missing characters used in regional keymap
files which already exist in QEMU. I checked for the missing
characters by going through all of the keymaps and matching that
with records in vnc_keysym.h. If the key wasn't found I looked
it up in libxkbcommon library [1]. If I understood it correctly
this is also the same place where most of the keymaps were
exported from according to the comment on the first line in those
files. I was able to find all symbols except "quotebl" used
in Netherland keymap.

I tested this update with Czech keyboard by myself. I also asked
Matej Serc to test Slovenian keyboard layout - he reported problems
with it few days ago on this mailing list. Both layouts seems
to work fine. I wasn't able to test the remaining layouts but
since this change doesn't modify any existing symbols, just adds
new ones, I don't expect any sideeffects.

[1] http://cgit.freedesktop.org/xorg/lib/libxkbcommon

Signed-off-by: Jan Krupa <jkrupa@suse.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qemu-char: add support for U-prefixed symbols

This patch adds support for Unicode symbols in keymap files. This
feature was already used in some keyboard layouts in QEMU generated
from XKB (e.g. Arabic) but it wasn't implemented in QEMU source code.

There is no need for check of validity of the hex string after U character
because strtol returns 0 in case the conversion was unsuccessful.

Signed-off-by: Jan Krupa <jkrupa@suse.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qemu-char: add Czech keymap file

This patch adds Czech keyboard layout to available keymap files
and Makefile.

Signed-off-by: Jan Krupa <jkrupa@suse.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

qemu-char: add Czech characters to VNC keysyms

This patch adds missing Czech characters to the VNC keysym table.

Signed-off-by: Jan Krupa <jkrupa@suse.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

Update version for 1.7.0-rc0 release

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

sdl: Reverse support for video mode setting

Currently, If the setting of video mode failed, qemu will exit. It
should go back to the previous setting if the new screen resolution
failed. This patch fixes LP#1216368, add support to revert to existing
surface for the failure of video mode setting.

Reported-by: Sascha Krissler <sascha@srlabs.de>
Signed-off-by: Lei Li <lilei@linux.vnet.ibm.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1378285636-7091-1-git-send-email-lilei@linux.vnet.ibm.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

timers: fix stop/cont with -icount

Stop/cont commands are broken with -icount due to a deadlock.  The
real problem is that the computation of timers_state.cpu_ticks_offset
makes no sense with -icount enabled: we set it to an icount clock value
in cpu_disable_ticks, and subtract a TSC (or similar, whatever
cpu_get_real_ticks happens to return) value in cpu_enable_ticks.

The fix is simple.  timers_state.cpu_ticks_offset is only used
together with cpu_get_real_ticks, so we can use cpu_get_real_ticks
in cpu_disable_ticks.  There is no need to update cpu_ticks_prev
at the time cpu_disable_ticks is called; instead, we can do it
the next time cpu_get_ticks is called.

The change to cpu_disable_ticks is the important part of the patch.
The rest modifies the code to always check timers_state.cpu_ticks_prev,
even when the ticks are not advancing (i.e. the VM is stopped).  It also
makes a similar change to cpu_get_clock_locked, so that the code remains
similar for cpu_get_ticks and cpu_get_clock_locked.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1382977938-13844-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

e1000/rtl8139: update HMP NIC when every bit is written

We currently just update the HMP NIC info when the last bit of macaddr
is written. This assumes that guest driver will write all the macaddr
from bit 0 to bit 5 when it changes the macaddr, this is the current
behavior of linux driver (e1000/rtl8139cp), but we can't do this
assumption.

The macaddr that is used for rx-filter will be updated when every bit
is changed. This patch updates the e1000/rtl8139 nic to update HMP NIC
info when every bit is changed. It will be same as virtio-net.

Signed-off-by: Amos Kong <akong@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Message-id: 1383650238-16015-1-git-send-email-akong@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

virtio-net: only delete bh that existed

We delete without check whether it existed during exit. This will lead NULL
pointer deference since it was created conditionally depends on guest driver
status and features. So add a check of existence before trying to delete it.

Cc: qemu-stable@nongnu.org
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383728288-28469-1-git-send-email-jasowang@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

rtc: remove dead SQW IRQ code

This was once introduced by commit 100d9891d6 but was never used in-tree
and then got broken by commit 32e0c8260d. Time to clean up.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Message-id: 520B6A27.4040207@siemens.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

vga: fix invalid read after free

After calling dpy_gfx_replace_surface(s->con, surface), the outer
surface is invalid.

==5370== Invalid read of size 4
==5370==    at 0x460229: surface_bits_per_pixel (console.h:250)
==5370==    by 0x466A81: get_depth_index (vga.c:1173)
==5370==    by 0x467EC2: vga_draw_graphic (vga.c:1718)
==5370==    by 0x4687A5: vga_update_display (vga.c:1914)
==5370==    by 0x2A782E: qxl_hw_update (qxl.c:1766)
==5370==    by 0x3EB83B: graphic_hw_update (console.c:254)
==5370==    by 0x3FBE31: qemu_spice_display_refresh (spice-display.c:418)
==5370==    by 0x2A7D01: display_refresh (qxl.c:1886)
==5370==    by 0x3EEE1C: dpy_refresh (console.c:1436)
==5370==    by 0x3EB543: gui_update (console.c:192)
==5370==    by 0x3C43B3: timerlist_run_timers (qemu-timer.c:488)
==5370==    by 0x3C4416: qemu_clock_run_timers (qemu-timer.c:499)
==5370==  Address 0x22ffb1e0 is 0 bytes inside a block of size 56 free'd
==5370==    at 0x4A074C4: free (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==5370==    by 0x4245FC: free_and_trace (vl.c:2771)
==5370==    by 0x50899AE: g_free (gmem.c:252)
==5370==    by 0x3EE8D3: qemu_free_displaysurface (console.c:1332)
==5370==    by 0x3EEDB7: dpy_gfx_replace_surface (console.c:1427)
==5370==    by 0x467EB6: vga_draw_graphic (vga.c:1714)
==5370==    by 0x4687A5: vga_update_display (vga.c:1914)
==5370==    by 0x2A782E: qxl_hw_update (qxl.c:1766)
==5370==    by 0x3EB83B: graphic_hw_update (console.c:254)
==5370==    by 0x3FBE31: qemu_spice_display_refresh (spice-display.c:418)
==5370==    by 0x2A7D01: display_refresh (qxl.c:1886)
==5370==    by 0x3EEE1C: dpy_refresh (console.c:1436)

Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1383664554-15248-1-git-send-email-marcandre.lureau@gmail.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

tests: fix 64-bit int literals for 32-bit hosts

On 32-bit hosts:

  CC    tests/test-opts-visitor.o
tests/test-opts-visitor.c: In function 'test_value':
tests/test-opts-visitor.c:128: warning: integer constant is too large for 'long' type
  CC    tests/test-bitops.o
tests/test-bitops.c:34: warning: integer constant is too large for 'long' type
tests/test-bitops.c:35: warning: integer constant is too large for 'long' type
tests/test-bitops.c:35: warning: integer constant is too large for 'long' type
  CC    tests/endianness-test.o
tests/endianness-test.c:47: warning: integer constant is too large for 'long' type

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1383669768-23926-1-git-send-email-stefanha@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

docs/memory.txt: Clarify and expand priority/overlap documentation

The documentation of how overlapping memory regions behave and how
the priority system works was rather brief, and confusion about
priorities seems to be quite common for developers trying to understand
how the memory region system works, so expand and clarify it.
This includes a worked example with overlaps, documentation of the
behaviour when an overlapped container has "holes", and mention
that it's valid for a region to have both MMIO callbacks and
subregions (and how this interacts with priorities when it does).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1381848154-31602-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

configure: detect endian via compile test

This avoids needing to execute a program and keeping an (incomplete)
list when cross-compiling.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Tested-by: James Hogan <james.hogan@imgtec.com> [mips]
Message-id: 1372649418-4987-1-git-send-email-vapier@gentoo.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

tests: fix memleak in error path test for input visitor

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1383676551-18806-3-git-send-email-xiawenc@linux.vnet.ibm.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

qapi: fix memleak by adding implict struct functions in dealloc visitor

Otherwise member "base" is leaked in a qapi_free_STRUCTURE() call.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1383676551-18806-2-git-send-email-xiawenc@linux.vnet.ibm.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_32wu()

Replace the legacy cpu_to_32wu() with stl_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-10-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_be64wu()

Replace the legacy cpu_to_be64wu() with stq_be_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-9-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_be32wu()

Replace the legacy cpu_to_be32wu() with stl_be_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-8-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_be16wu()

Replace the legacy cpu_to_be16wu() with stw_be_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-7-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove be32_to_cpupu()

Replace the legacy be32_to_cpupu() with ldl_be_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-6-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove le32_to_cpupu()

Replace the legacy le32_to_cpupu() with ldl_le_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-5-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove le16_to_cpupu()

Replace the legacy le16_to_cpupu() with lduw_le_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-4-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_le32wu()

Replace the legacy cpu_to_le32wu() with stl_le_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-3-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

bswap.h: Remove cpu_to_le16wu()

Replace the legacy cpu_to_le16wu() with stw_le_p().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1383669517-25598-2-git-send-email-peter.maydell@linaro.org
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

Merge remote-tracking branch 'afaerber/tags/qom-devices-for-anthony' into staging

QOM device refactorings

* QTest coverage for all machines
* QOM realize for Milkymist UART
* QOM realize for ARM MPCore
* device_add bug fixes and cleanups
* QOM for PCMCIA/MicroDrive (last legacy IDE device)

# gpg: Signature made Tue 05 Nov 2013 09:07:03 AM PST using RSA key ID 3E7E013F
# gpg: Can't check signature: public key not found

# By Andreas Färber (49) and others
# Via Andreas Färber
* afaerber/tags/qom-devices-for-anthony: (54 commits)
  pcmcia/pxa2xx: QOM'ify PXA2xxPCMCIAState
  ide: Drop ide_init2_with_non_qdev_drives()
  microdrive: Coding Style cleanups
  pcmcia: QOM'ify PCMCIACardState and MicroDriveState
  pxa: Fix typo "dettach"
  qom: Fix pointer to int property helpers' documentation
  qdev-monitor: Inline qdev_init() for device_add
  qdev-monitor: Avoid qdev as variable name
  qdev: Drop misleading qdev_free() function
  qdev-monitor: Unref device when device_add fails
  qdev-monitor: Fix crash when device_add is called with abstract driver
  qdev-monitor: Clean up qdev_device_add() variable naming
  arm11mpcore: Split off RealView MPCore
  arm11mpcore: Prepare for QOM embedding
  arm11mpcore: Convert mpcore_rirq_state to QOM realize
  realview_gic: Prepare for QOM embedding
  realview_gic: Convert to QOM realize
  arm11mpcore: Convert ARM11MPCorePriveState to QOM realize
  arm11mpcore: Split off SCU device
  arm11mpcore: Create container MemoryRegion in instance_init
  ...

pcmcia/pxa2xx: QOM'ify PXA2xxPCMCIAState

Turn it into a SysBusDevice and use a container MemoryRegion.

Add a link<pcmcia-card> property to the PCMCIACardState.

Signed-off-by: Andreas Färber <afaerber@suse.de>

ide: Drop ide_init2_with_non_qdev_drives()

All its users have finally been converted.

Signed-off-by: Andreas Färber <afaerber@suse.de>

microdrive: Coding Style cleanups

Add missing braces.

Signed-off-by: Andreas Färber <afaerber@suse.de>

pcmcia: QOM'ify PCMCIACardState and MicroDriveState

Turn PCMCIACardState into a device.
Move callbacks to new PCMCIACardClass.

Derive TYPE_MICRODRIVE from TYPE_PCMCIA_CARD.
Replace ide_init2_with_non_qdev_drives().

Signed-off-by: Othmar Pasteka <pasteka@kabsi.at>
Signed-off-by: Andreas Färber <afaerber@suse.de>

pxa: Fix typo "dettach"

Signed-off-by: Andreas Färber <afaerber@suse.de>

qom: Fix pointer to int property helpers' documentation

Relocate to alongside the other object_property_add_* helpers while at it.

Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev-monitor: Inline qdev_init() for device_add

For historic reasons, qdev_init() unparents the device on failure.
Inline this to make the error paths clearer and consistent.

Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev-monitor: Avoid qdev as variable name

Prepares for bringing error cleanup code into canonical QOM form.

Includes a whitespace removal after curly brace by Stefan.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev: Drop misleading qdev_free() function

The qdev_free() function name is misleading since all the function does
is unlink the device from its parent.  The device is not necessarily
freed.

The device will be freed when its QObject refcount reaches zero.  It is
usual for the parent (bus) to hold the final reference but there are
cases where something else holds a reference so "free" is a misleading
name.

Call object_unparent(obj) directly instead of having a qdev wrapper
function.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev-monitor: Unref device when device_add fails

qdev_device_add() leaks the created device upon failure.  I suspect this
problem crept in because qdev_free() unparents the device but does not
drop a reference - confusing name.

Cc: qemu-stable@nongnu.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev-monitor: Fix crash when device_add is called with abstract driver

User is able to crash running QEMU when following monitor
command is called:

 device_add intel-hda-generic

Crash is caused by assertion in object_initialize_with_type()
when type is abstract.

Checking if type is abstract before instance is created in
qdev_device_add() allows to prevent crash on incorrect user input.

Cc: qemu-stable@nongnu.org
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qdev-monitor: Clean up qdev_device_add() variable naming

Avoid confusion between object (obj) and object class (oc).
Tidy DeviceClass variable while at it (k -> dc).

Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Split off RealView MPCore

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Prepare for QOM embedding

Move state struct, type constant and cast macro to a new header.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Convert mpcore_rirq_state to QOM realize

Embed ARM11MPCorePriveState and RealViewGICState and replace SysBus
initfn with realizefn.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

realview_gic: Prepare for QOM embedding

Move state struct, type constant and cast macro to a new header.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

realview_gic: Convert to QOM realize

Embed GICState and replace SysBus initfn with realizefn.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Convert ARM11MPCorePriveState to QOM realize

Embed child devices and replace SysBus initfn with realizefn.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Split off SCU device

Inspired by a9scu.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Create container MemoryRegion in instance_init

This allows to map the region directly after object initialization.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Drop unused fields

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

arm11mpcore: Fix typo in MemoryRegion name

"mpcode" -> "mpcore"

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

a9scu: Build only once

It does not have a target or ARMCPU dependency.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

a15mpcore: Prepare for QOM embedding

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a15mpcore: Convert to QOM realize

Turn SysBusDevice initfn into a QOM realizefn.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a15mpcore: Embed GICState

This covers both emulated and KVM GIC.

Prepares for QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a15mpcore: Split off instance_init

Prepares for QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9mpcore: Prepare for QOM embedding

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9mpcore: Convert to QOM realize

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9mpcore: Embed ARMMPTimerState

Prepares for QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

arm_mptimer: Convert to QOM realize

Split the SysBusDevice initfn into instance_init and realizefn.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9mpcore: Embed A9SCUState

Prepares for QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9scu: QOM cleanups

Rename A9SCUState::busdev field to parent_obj and turn realizefn into an
instance_init function to allow early MMIO mapping.

Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

a9mpcore: Embed GICState

Prepares for conversion to QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

arm_gic: Extract headers hw/intc/arm_gic{,_common}.h

Rename NCPU to GIC_NCPU and move GICState away from gic_internal.h.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <afaerber@suse.de>

a9mpcore: Split off instance_init

Prepares for QOM realize.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Andreas Färber <andreas.faerber@web.de>

milkymist-uart: Use Device::realize instead of SysBusDevice::init

Use of SysBusDevice::init is deprecated. Use Device::realize instead.

Also introduce TypeInfo::instance_init milkymist_uart_init().

Reported-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

qtest: Prepare QOM machine tests

Instantiate all [*] machines per target, so that they get a bit of test
coverage at all. This has proven helpful during QOM refactorings.

[*] ppcemb target contains some non-working non-embedded machines, and
ppc405 CPUs are not available there either.
i386 and x86_64 do not cover pc*-x.y or xenfv.

Signed-off-by: Andreas Färber <afaerber@suse.de>

leon3: Don't enforce use of -bios with qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

shix: Don't require firmware presence for qtest

Adopt error_report() while at it.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Andreas Färber <afaerber@suse.de>

shix: Drop debug output

Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Andreas Färber <afaerber@suse.de>

milkymist: Suppress -kernel/-bios/-drive error for qtest

Acked-by: Michael Walle <michael@walle.cc>
Signed-off-by: Andreas Färber <afaerber@suse.de>

an5206: Don't enforce use of kernel for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

mcf5208: Don't enforce use of kernel for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

axis_dev88: Don't enforce use of kernel for qtest

Reviewed-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

armv7m: Don't enforce use of kernel for qtest

Adopt error_report().

Signed-off-by: Andreas Färber <afaerber@suse.de>

exynos4_boards: Silence lack of -smp 2 warning for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

omap_sx1: Don't enforce use of kernel or flash for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

palm: Don't enforce loading ROM or kernel for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

z2: Don't enforce use of -pflash for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

gumstix: Don't enforce use of -pflash for qtest

Signed-off-by: Andreas Färber <afaerber@suse.de>

mainstone: Don't enforce use of -pflash for qtest

Simply skip flash setup for now.

Also drop useless debug output.

Signed-off-by: Andreas Färber <afaerber@suse.de>

puv3: Turn puv3_load_kernel() into a no-op for qtest without -kernel

Replacing the assert() with more user-friendly error handling is left
for a follow-up.

Signed-off-by: Andreas Färber <afaerber@suse.de>

mips_mipssim: Silence BIOS loading warning for qtest

Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Andreas Färber <afaerber@suse.de>

Merge tag 'for_anthony' of git://git./virt/kvm/mst/qemu

pci, pc, pvpanic bug fixes

This fixes strange pvpanic behaviour: you had to
pause to let VM continue (and potentially reboot on panic
if enabled).

This also fixes two bugs reported by Andreas.
One is a long-standing bug exposed by recent pci changes,
the other affects old piix machine types and was caused
by recent acpi changes.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>

ossaudio: do not enable by default

Modern Linux's no longer support /dev/dsp so enabling it by
default causes audio failures on newer Linux distros.

Signed-off-by: Anthony Liguori <aliguori@amazon.com>
Tested-by: Andreas Färber <afaerber@suse.de>
Message-id: 1383497154-9271-1-git-send-email-aliguori@amazon.com

Merge remote-tracking branch 'spice/spice.v76' into staging

# By Gerd Hoffmann
# Via Gerd Hoffmann
* spice/spice.v76:
  qxl: replace pipe signaling with bottom half

Message-id: 1383656322-24150-1-git-send-email-kraxel@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

Merge remote-tracking branch 'mst/tags/for_anthony' into staging

pci, pc, pvpanic bug fixes

This fixes strange pvpanic behaviour: you had to
pause to let VM continue (and potentially reboot on panic
if enabled).

This also fixes two bugs reported by Andreas.
One is a long-standing bug exposed by recent pci changes,
the other affects old piix machine types and was caused
by recent acpi changes.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Mon 04 Nov 2013 05:42:46 AM PST using RSA key ID D28D5469
# gpg: Can't check signature: public key not found

# By Michael S. Tsirkin (2) and Paolo Bonzini (1)
# Via Michael S. Tsirkin
* mst/tags/for_anthony:
  vl: allow "cont" from panicked state
  exec: limit system memory size
  pc: disable acpi info for isapc and old pc machine

Message-id: 1383572851-28326-1-git-send-email-mst@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

Merge remote-tracking branch 'kraxel/e820.1' into staging

# By Gerd Hoffmann
# Via Gerd Hoffmann
* kraxel/e820.1:
  pc: register e820 entries for ram
  pc: add etc/e820 fw_cfg file

Message-id: 1383567431-13540-1-git-send-email-kraxel@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

vl: allow "cont" from panicked state

After reporting the GUEST_PANICKED monitor event, QEMU stops the VM.
The reason for this is that events are edge-triggered, and can be lost if
management dies at the wrong time.  Stopping a panicked VM lets management
know of a panic even if it has crashed; management can learn about the
panic when it restarts and queries running QEMU processes.  The downside
is of course that the VM will be paused while management is not running,
but that is acceptable if it only happens with explicit "-device pvpanic".

Upon learning of a panic, management (if configured to do so) can pick a
variety of behaviors: leave the VM paused, reset it, destroy it.  In
addition to all of these behaviors, it is possible to dump the VM core
from the host.

However, right now, the panicked state is irreversible, and can only be
exited by resetting the machine.  This means that any policy decision
is entirely in the hands of the host.  In particular there is no way to
use the "reboot on panic" option together with pvpanic.

This patch makes the panicked state reversible (and removes various
workarounds that were there because of the state being irreversible).
With this change, management has a wider set of possible policies: it
can just log the crash and leave policy to the guest, it can leave the
VM paused.  In particular, the "log the crash and continue" is implemented
simply by sending a "cont" as soon as management learns about the panic.
Management could also implement the "irreversible paused state" itself.
And again, all such actions can be coupled with dumping the VM core.

Unfortunately we cannot change the behavior of 1.6.0.  Thus, even if
it uses "-device pvpanic", management should check for "cont" failures.
If "cont" fails, management can then log that the VM remained paused
and urge the administrator to update QEMU.

Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

exec: limit system memory size

The page table logic in exec.c assumes
that memory addresses are at most TARGET_PHYS_ADDR_SPACE_BITS.

But pci addresses are full 64 bit so if we try to render them ignoring
the extra bits, we get strange effects with sections overlapping each
other.

To fix, simply limit the system memory size to
 1 << TARGET_PHYS_ADDR_SPACE_BITS,
pci addresses will be rendered within that.

Cc: qemu-stable@nongnu.org
Reported-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pc: disable acpi info for isapc and old pc machine

Disable acpi build for isapc and no_kvmclock machine
types (used by xen), since acpi build currently expects pci.

Reported-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

qxl: replace pipe signaling with bottom half

qxl creates a pipe, then writes something to it to wake up the iothread
from the spice server thread to raise an irq.  These days qemu bottom
halves can be scheduled from threads and signals, so there is no reason
to do this any more.  Time to clean it up.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

pc: register e820 entries for ram

So RAM shows up in the new etc/e820 fw_cfg file.

Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>