sdk/emulator/qemu.git
8 years agotarget-i386: Set constant model_id for qemu64/qemu32/athlon
Eduardo Habkost [Sat, 9 Apr 2016 19:44:20 +0000 (16:44 -0300)]
target-i386: Set constant model_id for qemu64/qemu32/athlon

Newer PC machines don't set hw_version, and older machines set
model-id on compat_props explicitly, so we don't need the
x86_cpudef_setup() code that sets model_id using
qemu_hw_version() anymore.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agopc: Set CPU model-id on compat_props for pc <= 2.4
Eduardo Habkost [Sat, 9 Apr 2016 19:26:38 +0000 (16:26 -0300)]
pc: Set CPU model-id on compat_props for pc <= 2.4

Instead of relying on x86_cpudef_setup() calling
qemu_hw_version(), just make old machines set model-id explicitly
on compat_props for qemu64, qemu32, and athlon. This will allow
us to eliminate x86_cpudef_setup() later.

Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agoosdep: Move default qemu_hw_version() value to a macro
Eduardo Habkost [Sat, 9 Apr 2016 19:42:44 +0000 (16:42 -0300)]
osdep: Move default qemu_hw_version() value to a macro

The macro will be used by code that will stop calling
qemu_hw_version() at runtime and just need a constant value.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agotarget-i386: kvm: Use X86XSaveArea struct for xsave save/load
Eduardo Habkost [Mon, 23 Nov 2015 12:43:26 +0000 (10:43 -0200)]
target-i386: kvm: Use X86XSaveArea struct for xsave save/load

Instead of using offset macros and bit operations in a uint32_t
array, use the X86XSaveArea struct to perform the loading/saving
operations in kvm_put_xsave() and kvm_get_xsave().

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agotarget-i386: Use xsave structs for ext_save_area
Eduardo Habkost [Sat, 28 Nov 2015 16:32:26 +0000 (14:32 -0200)]
target-i386: Use xsave structs for ext_save_area

This doesn't introduce any change in the code, as the offsets and
struct sizes match what was present in the table. This can be
validated by the QEMU_BUILD_BUG_ON lines on target-i386/cpu.h,
which ensures the struct sizes and offsets match the existing
values in ext_save_area.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agotarget-i386: Define structs for layout of xsave area
Eduardo Habkost [Thu, 19 Nov 2015 18:52:33 +0000 (16:52 -0200)]
target-i386: Define structs for layout of xsave area

Add structs that define the layout of the xsave areas used by
Intel processors. Add some QEMU_BUILD_BUG_ON lines to ensure the
structs match the XSAVE_* macros in target-i386/kvm.c and the
offsets and sizes at target-i386/cpu.c:ext_save_areas.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Mon, 23 May 2016 15:15:51 +0000 (16:15 +0100)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

* NMI cleanups (Bandan)
* RAMBlock/Memory cleanups and fixes (Dominik, Gonglei, Fam, me)
* first part of linuxboot support for fw_cfg DMA (Richard)
* IOAPIC fix (Peter Xu)
* iSCSI SG_IO fix (Vadim)
* Various infrastructure bug fixes (Zhijian, Peter M., Stefan)
* CVE fixes (Prasad)

# gpg: Signature made Mon 23 May 2016 16:06:18 BST using RSA key ID 78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"

* remotes/bonzini/tags/for-upstream: (24 commits)
  cpus: call the core nmi injection function
  nmi: remove x86 specific nmi handling
  target-i386: add a generic x86 nmi handler
  coccinelle: add g_assert_cmp* to macro file
  iscsi: pass SCSI status back for SG_IO
  esp: check dma length before reading scsi command(CVE-2016-4441)
  esp: check command buffer length before write(CVE-2016-4439)
  scripts/signrom.py: Check for magic in option ROMs.
  scripts/signrom.py: Allow option ROM checksum script to write the size header.
  Remove config-devices.mak on 'make clean'
  cpus.c: Use pthread_sigmask() rather than sigprocmask()
  memory: remove unnecessary masking of MemoryRegion ram_addr
  memory: Drop FlatRange.romd_mode
  memory: Remove code for mr->may_overlap
  exec: adjust rcu_read_lock requirement
  memory: drop find_ram_block()
  vl: change runstate only if new state is different from current state
  ioapic: clear remote irr bit for edge-triggered interrupts
  ioapic: keep RO bits for IOAPIC entry
  target-i386: key sfence availability on CPUID_SSE, not CPUID_SSE2
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agocpus: call the core nmi injection function
Bandan Das [Fri, 20 May 2016 16:28:37 +0000 (12:28 -0400)]
cpus: call the core nmi injection function

We can call the common function here directly since
x86 specific actions will be taken care of by the arch
specific nmi handler

Signed-off-by: Bandan Das <bsd@redhat.com>
Message-Id: <1463761717-26558-4-git-send-email-bsd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonmi: remove x86 specific nmi handling
Bandan Das [Fri, 20 May 2016 16:28:36 +0000 (12:28 -0400)]
nmi: remove x86 specific nmi handling

nmi_monitor_handle is wired to call the x86 nmi
handler. So, we can directly use it at call sites.

Signed-off-by: Bandan Das <bsd@redhat.com>
Message-Id: <1463761717-26558-3-git-send-email-bsd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agotarget-i386: add a generic x86 nmi handler
Bandan Das [Fri, 20 May 2016 16:28:35 +0000 (12:28 -0400)]
target-i386: add a generic x86 nmi handler

Instead of having x86 ifdefs in core nmi code, this
change adds a arch specific handler that the nmi common
code can call.

Signed-off-by: Bandan Das <bsd@redhat.com>
Message-Id: <1463761717-26558-2-git-send-email-bsd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agococcinelle: add g_assert_cmp* to macro file
Paolo Bonzini [Wed, 18 May 2016 09:11:55 +0000 (11:11 +0200)]
coccinelle: add g_assert_cmp* to macro file

This helps applying semantic patches to unit tests.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoiscsi: pass SCSI status back for SG_IO
Vadim Rozenfeld [Fri, 13 May 2016 11:03:22 +0000 (13:03 +0200)]
iscsi: pass SCSI status back for SG_IO

Signed-off-by: Vadim Rozenfeld <vrozenfe@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoesp: check dma length before reading scsi command(CVE-2016-4441)
Prasad J Pandit [Thu, 19 May 2016 10:39:31 +0000 (16:09 +0530)]
esp: check dma length before reading scsi command(CVE-2016-4441)

The 53C9X Fast SCSI Controller(FSC) comes with an internal 16-byte
FIFO buffer. It is used to handle command and data transfer.
Routine get_cmd() uses DMA to read scsi commands into this buffer.
Add check to validate DMA length against buffer size to avoid any
overrun.

Fixes CVE-2016-4441.

Reported-by: Li Qiang <liqiang6-s@360.cn>
Cc: qemu-stable@nongnu.org
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-Id: <1463654371-11169-3-git-send-email-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoesp: check command buffer length before write(CVE-2016-4439)
Prasad J Pandit [Thu, 19 May 2016 10:39:30 +0000 (16:09 +0530)]
esp: check command buffer length before write(CVE-2016-4439)

The 53C9X Fast SCSI Controller(FSC) comes with an internal 16-byte
FIFO buffer. It is used to handle command and data transfer. While
writing to this command buffer 's->cmdbuf[TI_BUFSZ=16]', a check
was missing to validate input length. Add check to avoid OOB write
access.

Fixes CVE-2016-4439.

Reported-by: Li Qiang <liqiang6-s@360.cn>
Cc: qemu-stable@nongnu.org
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-Id: <1463654371-11169-2-git-send-email-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscripts/signrom.py: Check for magic in option ROMs.
Richard W.M. Jones [Wed, 11 May 2016 21:06:46 +0000 (22:06 +0100)]
scripts/signrom.py: Check for magic in option ROMs.

Because of the risk that compilers might not emit the asm() block at
the beginning of the option ROM, check that the ROM contains the
required magic signature.

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Message-Id: <1463000807-18015-3-git-send-email-rjones@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscripts/signrom.py: Allow option ROM checksum script to write the size header.
Richard W.M. Jones [Wed, 11 May 2016 21:06:45 +0000 (22:06 +0100)]
scripts/signrom.py: Allow option ROM checksum script to write the size header.

Modify the signrom.py script so that if the size byte in the header is
0 (ie. not set) then the script will set the size.  If the size byte
is non-zero then we do the same as before, so this doesn't require
changes to any existing ROM sourcecode.

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Message-Id: <1463000807-18015-2-git-send-email-rjones@redhat.com>

8 years agoRemove config-devices.mak on 'make clean'
Peter Maydell [Tue, 17 May 2016 11:27:31 +0000 (12:27 +0100)]
Remove config-devices.mak on 'make clean'

Our dependency mechanism works like this:
 * on first build there is neither a .o nor a .d
 * we create the .d as a side effect of creating the .o
 * for rebuilds we know when we need to update the .o,
   which also updates the .d

This system requires that you're never in a situation where there is
a .o file but no .d (because then we will never realise we need to
build the .d, and we will not have the dependency information about
when to rebuild the .o).

This is working fine for our object files, but we also try to use it
for $TARGET/config-devices.mak (where the dependency file is
in $TARGET-config-devices.mak.d). Unfortunately "make clean" doesn't
remove config-devices.mak, which means that it puts us in the
forbidden situation of "object file exists but not its .d file".
This in turn means that we will fail to notice when we need to rebuild:
  mkdir build/depbug
  (cd build/depbug && '../../configure')
  make -C build/depbug -j8
  make -C build/depbug clean
  echo "CONFIG_CANARY = y" >> default-configs/arm-softmmu.mak
  make -C build/depbug
  grep CANARY build/depbug/aarch64-softmmu/config-devices.mak

The CANARY token should show up in config-devices.mak but does not.

Fix this bug by making "make clean" delete the config-devices.mak files.
config-all-devices.mak doesn't have the same problem since it has
no .d file, but delete it too, since it is created by "make" and
logically should be removed by "make clean".

(Note that it is important not to remove config-devices.mak until
after we have recursively run 'make clean' in the subdirectories.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <1463484451-22979-1-git-send-email-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agocpus.c: Use pthread_sigmask() rather than sigprocmask()
Peter Maydell [Mon, 16 May 2016 17:33:59 +0000 (18:33 +0100)]
cpus.c: Use pthread_sigmask() rather than sigprocmask()

On Linux, sigprocmask() and pthread_sigmask() are in practice the
same thing (they only set the signal mask for the calling thread),
but the documentation states that the behaviour of sigprocmask() in a
multithreaded process is undefined. Use pthread_sigmask() instead
(which is what we do in almost all places in QEMU that alter the
signal mask already).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <1463420039-29761-1-git-send-email-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: remove unnecessary masking of MemoryRegion ram_addr
Paolo Bonzini [Tue, 1 Mar 2016 09:44:50 +0000 (10:44 +0100)]
memory: remove unnecessary masking of MemoryRegion ram_addr

mr->ram_block->offset is already aligned to both host and target size
(see qemu_ram_alloc_internal).  Remove further masking as it is
unnecessary.

Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: Drop FlatRange.romd_mode
Fam Zheng [Fri, 25 Mar 2016 10:10:29 +0000 (18:10 +0800)]
memory: Drop FlatRange.romd_mode

Its value is alway set to mr->romd_mode, so the removed comparisons are
fully superseded by "a->mr == b->mr".

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1458900629-2334-3-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: Remove code for mr->may_overlap
Fam Zheng [Fri, 25 Mar 2016 10:10:28 +0000 (18:10 +0800)]
memory: Remove code for mr->may_overlap

The collision check does nothing and hasn't been used. Remove the
variable together with related code.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1458900629-2334-2-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoexec: adjust rcu_read_lock requirement
Gonglei [Tue, 10 May 2016 02:05:00 +0000 (10:05 +0800)]
exec: adjust rcu_read_lock requirement

qemu_ram_unset_idstr() doesn't need rcu lock anymore,
meanwhile make the range of rcu lock in
qemu_ram_set_idstr() as small as possible.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Message-Id: <1462845901-89716-3-git-send-email-arei.gonglei@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: drop find_ram_block()
Gonglei [Tue, 10 May 2016 02:04:59 +0000 (10:04 +0800)]
memory: drop find_ram_block()

On the one hand, we have already qemu_get_ram_block() whose function
is similar. On the other hand, we can directly use mr->ram_block but
searching RAMblock by ram_addr which is a kind of waste.

Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-Id: <1462845901-89716-2-git-send-email-arei.gonglei@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agovl: change runstate only if new state is different from current state
Li Zhijian [Thu, 14 Apr 2016 03:25:52 +0000 (11:25 +0800)]
vl: change runstate only if new state is different from current state

Previously, qemu will abort at following scenario:
(qemu) stop
(qemu) system_reset
(qemu) system_reset
(qemu) 2016-04-13T20:54:38.979158Z qemu-system-x86_64: invalid runstate transition: 'prelaunch' -> 'prelaunch'

Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1460604352-18630-1-git-send-email-lizhijian@cn.fujitsu.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoioapic: clear remote irr bit for edge-triggered interrupts
Peter Xu [Tue, 10 May 2016 10:21:22 +0000 (18:21 +0800)]
ioapic: clear remote irr bit for edge-triggered interrupts

This is to better emulate IOAPIC version 0x1X hardware. Linux kernel
leveraged this "feature" to do explicit EOI since EOI register is still
not introduced at that time. This will also fix the issue that level
triggered interrupts failed to work when IR enabled (tested with Linux
kernel version 4.5).

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1462875682-1349-3-git-send-email-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoioapic: keep RO bits for IOAPIC entry
Peter Xu [Tue, 10 May 2016 10:21:21 +0000 (18:21 +0800)]
ioapic: keep RO bits for IOAPIC entry

Currently IOAPIC RO bits can be written. To be better aligned with
hardware, we should let them read-only.

Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1462875682-1349-2-git-send-email-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agotarget-i386: key sfence availability on CPUID_SSE, not CPUID_SSE2
Paolo Bonzini [Mon, 16 May 2016 09:11:29 +0000 (11:11 +0200)]
target-i386: key sfence availability on CPUID_SSE, not CPUID_SSE2

sfence was introduced before lfence and mfence.  This fixes Linux
2.4's measurement of checksumming speeds for the pIII_sse
algorithm:

md: linear personality registered as nr 1
md: raid0 personality registered as nr 2
md: raid1 personality registered as nr 3
md: raid5 personality registered as nr 4
raid5: measuring checksumming speed
   8regs     :   384.400 MB/sec
   32regs    :   259.200 MB/sec
invalid operand: 0000
CPU:    0
EIP:    0010:[<c0240b2a>]    Not tainted
EFLAGS: 00000246
eax: c15d8000   ebx: 00000000   ecx: 00000000   edx: c15d5000
esi: 8005003b   edi: 00000004   ebp: 00000000   esp: c15bdf50
ds: 0018   es: 0018   ss: 0018
Process swapper (pid: 1, stackpage=c15bd000)
Stack: 00000000 00000000 00000000 00000000 00000000 00000000 00000000
00000000
       00000000 00000000 00000000 00000000 00000000 00000000 00000000
00000000
       00000000 00000206 c0241c6c 00001000 c15d4000 c15d7000 c15d4000
c15d4000
Call Trace:    [<c0241c6c>] [<c0105000>] [<c0241db4>] [<c010503b>]
[<c0105000>]
  [<c0107416>] [<c0105030>]

Code: 0f ae f8 0f 10 04 24 0f 10 4c 24 10 0f 10 54 24 20 0f 10 5c
 <0>Kernel panic: Attempted to kill init!

Reported-by: Stefan Weil <sw@weilnetz.de>
Fixes: 121f3157887f92268a3d6169e2d4601f9292020b
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoconfigure: Allow builds with extra warnings
Stefan Weil [Thu, 28 Apr 2016 21:33:41 +0000 (23:33 +0200)]
configure: Allow builds with extra warnings

The clang compiler supports a useful compiler option -Weverything,
and GCC also has other warnings not enabled by -Wall.

If glib header files trigger a warning, however, testing glib with
-Werror will always fail. A size mismatch is also detected without
-Werror, so simply remove it.

Cc: qemu-stable@nongnu.org
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Message-Id: <1461879221-13338-1-git-send-email-sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoi386: kvmvapic: initialise imm32 variable
Prasad J Pandit [Thu, 7 Apr 2016 07:20:08 +0000 (12:50 +0530)]
i386: kvmvapic: initialise imm32 variable

When processing Task Priorty Register(TPR) access, it could leak
automatic stack variable 'imm32' in patch_instruction().
Initialise the variable to avoid it.

Reported by: Donghai Zdh <donghai.zdh@alibaba-inc.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-Id: <1460013608-16670-1-git-send-email-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agodocs/atomics.txt: Update pointer to linux macro
Pranith Kumar [Mon, 2 May 2016 14:20:52 +0000 (10:20 -0400)]
docs/atomics.txt: Update pointer to linux macro

Add a missing end brace and update doc to point to the latest access
macro. ACCESS_ONCE() is deprecated.

Signed-off-by: Pranith Kumar <bobby.prani@gmail.com>
Message-Id: <1462198852-28694-1-git-send-email-bobby.prani@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoexec.c: Ensure right alignment also for file backed ram
Dominik Dingel [Mon, 25 Apr 2016 11:55:38 +0000 (13:55 +0200)]
exec.c: Ensure right alignment also for file backed ram

While in the anonymous ram case we already take care of the right alignment
such an alignment gurantee does not exist for file backed ram allocation.

Instead, pagesize is used for alignment. On s390 this is not enough for gmap,
as we need to satisfy an alignment up to segments.

Reported-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Message-Id: <1461585338-45863-1-git-send-email-dingel@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-usb-20160523-1' into staging
Peter Maydell [Mon, 23 May 2016 14:53:02 +0000 (15:53 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-usb-20160523-1' into staging

usb: add xen pvUSB backend, add num-ports check to ohci.

# gpg: Signature made Mon 23 May 2016 14:02:25 BST using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-usb-20160523-1:
  usb/ohci: Fix crash with when specifying too many num-ports
  xen: add pvUSB backend
  xen: write information about supported backends
  xen: introduce dummy system device

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-vga-20160523-1' into staging
Peter Maydell [Mon, 23 May 2016 13:50:40 +0000 (14:50 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-vga-20160523-1' into staging

vga: fix CVE-2016-3712 regression, misc virtio-gpu fixes.

# gpg: Signature made Mon 23 May 2016 13:30:26 BST using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-vga-20160523-1:
  vga: add sr_vbe register set
  virtio-gpu: fix ui idx check
  virtio-gpu: use VIRTIO_GPU_MAX_SCANOUTS
  virtio-gpu: check max_outputs only
  virtio-gpu: check max_outputs value
  virtio-vga: propagate on gpu realized error
  virtio-gpu: check early scanout id

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agousb/ohci: Fix crash with when specifying too many num-ports
Thomas Huth [Mon, 23 May 2016 09:23:07 +0000 (11:23 +0200)]
usb/ohci: Fix crash with when specifying too many num-ports

QEMU currently crashes when an OHCI controller is instantiated with
too many ports, e.g. "-device pci-ohci,num-ports=100,masterbus=1".
Thus add a proper check in usb_ohci_init() to make sure that we
do not use more than OHCI_MAX_PORTS = 15 ports here.

Ticket: https://bugs.launchpad.net/qemu/+bug/1581308
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-id: 1463995387-11710-1-git-send-email-thuth@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovga: add sr_vbe register set
Gerd Hoffmann [Tue, 17 May 2016 08:54:54 +0000 (10:54 +0200)]
vga: add sr_vbe register set

Commit "fd3c136 vga: make sure vga register setup for vbe stays intact
(CVE-2016-3712)." causes a regression.  The win7 installer is unhappy
because it can't freely modify vga registers any more while in vbe mode.

This patch introduces a new sr_vbe register set.  The vbe_update_vgaregs
will fill sr_vbe[] instead of sr[].  Normal vga register reads and
writes go to sr[].  Any sr register read access happens through a new
sr() helper function which will read from sr_vbe[] with vbe active and
from sr[] otherwise.

This way we can allow guests update sr[] registers as they want, without
allowing them disrupt vbe video modes that way.

Cc: qemu-stable@nongnu.org
Reported-by: Thomas Lamprecht <thomas@lamprecht.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1463475294-14119-1-git-send-email-kraxel@redhat.com

8 years agoxen: add pvUSB backend
Juergen Gross [Thu, 12 May 2016 14:13:41 +0000 (16:13 +0200)]
xen: add pvUSB backend

Add a backend for para-virtualized USB devices for xen domains.

The backend is using host-libusb to forward USB requests from a
domain via libusb to the real device(s) passed through.

Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@citrix.com>
Message-id: 1463062421-613-4-git-send-email-jgross@suse.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-gpu: fix ui idx check
Marc-André Lureau [Thu, 19 May 2016 10:26:00 +0000 (12:26 +0200)]
virtio-gpu: fix ui idx check

Fix off-by-one value check (0 is the first scanout).

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-7-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoxen: write information about supported backends
Juergen Gross [Thu, 12 May 2016 14:13:40 +0000 (16:13 +0200)]
xen: write information about supported backends

Add a Xenstore directory for each supported pv backend. This will allow
Xen tools to decide which backend type to use in case there are
multiple possibilities.

The information is added under
/local/domain/<backend-domid>/device-model/<domid>/backends
before the "running" state is written to Xenstore. Using a directory
for each backend enables us to add parameters for specific backends
in the future.

This interface is documented in the Xen source repository in the file
docs/misc/qemu-backends.txt

In order to reuse the Xenstore directory creation already present in
hw/xen/xen_devconfig.c move the related functions to
hw/xen/xen_backend.c where they fit better.

Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@citrix.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Message-id: 1463062421-613-3-git-send-email-jgross@suse.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-gpu: use VIRTIO_GPU_MAX_SCANOUTS
Marc-André Lureau [Thu, 19 May 2016 10:25:59 +0000 (12:25 +0200)]
virtio-gpu: use VIRTIO_GPU_MAX_SCANOUTS

The value is defined in virtio_gpu.h already (changing from 4 to 16).

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-6-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoxen: introduce dummy system device
Juergen Gross [Thu, 12 May 2016 14:13:39 +0000 (16:13 +0200)]
xen: introduce dummy system device

Introduce a new dummy system device serving as parent for virtual
buses. This will enable new pv backends to introduce virtual buses
which are removable again opposed to system buses which are meant
to stay once added.

Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Anthony PERARD <anthony.perard@citrix.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Message-id: 1463062421-613-2-git-send-email-jgross@suse.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-gpu: check max_outputs only
Marc-André Lureau [Thu, 19 May 2016 10:25:58 +0000 (12:25 +0200)]
virtio-gpu: check max_outputs only

The scanout id should not be above the configured num_scanouts.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-5-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-gpu: check max_outputs value
Marc-André Lureau [Thu, 19 May 2016 10:25:57 +0000 (12:25 +0200)]
virtio-gpu: check max_outputs value

The value must be less than VIRTIO_GPU_MAX_SCANOUT.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-4-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-vga: propagate on gpu realized error
Marc-André Lureau [Thu, 19 May 2016 10:25:56 +0000 (12:25 +0200)]
virtio-vga: propagate on gpu realized error

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-3-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovirtio-gpu: check early scanout id
Marc-André Lureau [Thu, 19 May 2016 10:25:55 +0000 (12:25 +0200)]
virtio-gpu: check early scanout id

Before accessing the g->scanout array, in order to avoid potential
out-of-bounds access.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1463653560-26958-2-git-send-email-marcandre.lureau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/ehabkost/tags/machine-pull-request' into staging
Peter Maydell [Mon, 23 May 2016 09:30:41 +0000 (10:30 +0100)]
Merge remote-tracking branch 'remotes/ehabkost/tags/machine-pull-request' into staging

Machine Core queue, 2016-05-20

# gpg: Signature made Fri 20 May 2016 21:26:49 BST using RSA key ID 984DC5A6
# gpg: Good signature from "Eduardo Habkost <ehabkost@redhat.com>"

* remotes/ehabkost/tags/machine-pull-request: (21 commits)
  Use &error_fatal when initializing crypto on qemu-{img,io,nbd}
  vl: Use &error_fatal when parsing monitor options
  vl: Use &error_fatal when parsing VNC options
  machine: add properties to compat_props incrementaly
  vl: Simplify global property registration
  vl: Make display_remote a local variable
  vl: Move DisplayType typedef to vl.c
  vl: Make display_type a local variable
  vl: Replace DT_NOGRAPHIC with machine option
  milkymist: Move DT_NOGRAPHIC check outside milkymist_tmu2_create()
  spice: Initialization stubs on qemu-spice.h
  gtk: Initialization stubs
  cocoa: cocoa_display_init() stub
  sdl: Initialization stubs
  curses: curses_display_init() stub
  vnc: Initialization stubs
  vl: Add DT_COCOA DisplayType value
  vl: Replace *_vga_available() functions with class_names field
  vl: Table-based select_vgahw()
  vl: Use exit(1) when requested VGA interface is unavailable
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoUse &error_fatal when initializing crypto on qemu-{img,io,nbd}
Eduardo Habkost [Thu, 12 May 2016 14:10:04 +0000 (11:10 -0300)]
Use &error_fatal when initializing crypto on qemu-{img,io,nbd}

In addition to making the code simpler, this will replace the
long error messages:
  cannot initialize crypto: Unable to initialize GNUTLS library: [...]
  cannot initialize crypto: Unable to initialize gcrypt
with shorter messages:
  Unable to initialize GNUTLS library: [...]
  Unable to initialize gcrypt

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Use &error_fatal when parsing monitor options
Eduardo Habkost [Thu, 12 May 2016 14:10:00 +0000 (11:10 -0300)]
vl: Use &error_fatal when parsing monitor options

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Use &error_fatal when parsing VNC options
Eduardo Habkost [Thu, 12 May 2016 14:09:59 +0000 (11:09 -0300)]
vl: Use &error_fatal when parsing VNC options

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agomachine: add properties to compat_props incrementaly
Igor Mammedov [Thu, 28 Jan 2016 10:58:08 +0000 (11:58 +0100)]
machine: add properties to compat_props incrementaly

Switch to adding compat properties incrementaly instead of
completly overwriting compat_props per machine type.
That removes data duplication which we have due to nested
[PC|SPAPR]_COMPAT_* macros.

It also allows to set default device properties from
default foo_machine_options() hook, which will be used
in following patch for putting VMGENID device as
a function if ISA bridge on pc/q35 machines.

Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
[ehabkost: Fixed CCW_COMPAT_* and PC_COMPAT_0_* defines]
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Simplify global property registration
Eduardo Habkost [Thu, 28 Jan 2016 15:11:04 +0000 (13:11 -0200)]
vl: Simplify global property registration

There's no need to use qdev_prop_register_global_list() and an
array, if we are registering a single GlobalProperty struct. Use
qdev_prop_register_global() instead.

Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Make display_remote a local variable
Eduardo Habkost [Tue, 19 Apr 2016 19:55:28 +0000 (16:55 -0300)]
vl: Make display_remote a local variable

The variable is used only inside main(), so it can be local.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Move DisplayType typedef to vl.c
Eduardo Habkost [Tue, 19 Apr 2016 19:55:27 +0000 (16:55 -0300)]
vl: Move DisplayType typedef to vl.c

Now the type is only used inside vl.c and doesn't need to be in a
header file.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Make display_type a local variable
Eduardo Habkost [Tue, 19 Apr 2016 19:55:26 +0000 (16:55 -0300)]
vl: Make display_type a local variable

Now display_type is only used inside main(), and don't need to be a
global variable.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Replace DT_NOGRAPHIC with machine option
Eduardo Habkost [Tue, 19 Apr 2016 19:55:25 +0000 (16:55 -0300)]
vl: Replace DT_NOGRAPHIC with machine option

All DisplayType values are just UI options that don't affect any
hardware emulation code, except for DT_NOGRAPHIC. Replace
DT_NOGRAPHIC with DT_NONE plus a new "-machine graphics=on|off"
option, so hardware emulation code don't need to use the
display_type variable.

Cc: Michael Walle <michael@walle.cc>
Cc: Blue Swirl <blauwirbel@gmail.com>
Cc: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agomilkymist: Move DT_NOGRAPHIC check outside milkymist_tmu2_create()
Eduardo Habkost [Tue, 19 Apr 2016 19:55:24 +0000 (16:55 -0300)]
milkymist: Move DT_NOGRAPHIC check outside milkymist_tmu2_create()

DT_NOGRAPHIC handling will be moved to a MachineState field, and
it will be easier to change milkymist_init() to check that field.

Cc: Michael Walle <michael@walle.cc>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agospice: Initialization stubs on qemu-spice.h
Eduardo Habkost [Tue, 19 Apr 2016 19:55:23 +0000 (16:55 -0300)]
spice: Initialization stubs on qemu-spice.h

This reduces the number of CONFIG_SPICE #ifdefs in vl.c.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agogtk: Initialization stubs
Eduardo Habkost [Tue, 19 Apr 2016 19:55:22 +0000 (16:55 -0300)]
gtk: Initialization stubs

This reduces the number of CONFIG_GTK #ifdefs in vl.c.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agococoa: cocoa_display_init() stub
Eduardo Habkost [Tue, 19 Apr 2016 19:55:21 +0000 (16:55 -0300)]
cocoa: cocoa_display_init() stub

One less #ifdef in vl.c.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agosdl: Initialization stubs
Eduardo Habkost [Tue, 19 Apr 2016 19:55:20 +0000 (16:55 -0300)]
sdl: Initialization stubs

This reduces the number of CONFIG_SDL #ifdefs in vl.c.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agocurses: curses_display_init() stub
Eduardo Habkost [Tue, 19 Apr 2016 19:55:19 +0000 (16:55 -0300)]
curses: curses_display_init() stub

One less #ifdef in vl.c.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovnc: Initialization stubs
Eduardo Habkost [Tue, 19 Apr 2016 19:55:18 +0000 (16:55 -0300)]
vnc: Initialization stubs

This reduces the number of CONFIG_VNC #ifdefs in the vl.c code.

The only user-visible difference is that this will make QEMU
complain about syntax when using "-display vnc" ("VNC requires a
display argument vnc=<display>") even if CONFIG_VNC is disabled.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Add DT_COCOA DisplayType value
Eduardo Habkost [Tue, 19 Apr 2016 19:55:17 +0000 (16:55 -0300)]
vl: Add DT_COCOA DisplayType value

Instead of reusing DT_SDL for Cocoa, use DT_COCOA to indicate
that a Cocoa display was requested.

configure already ensures CONFIG_COCOA and CONFIG_SDL are never
set at the same time. The only case where DT_SDL is used outside
a #ifdef CONFIG_SDL block is in the no_frame/alt_grab/ctrl_grab
check. That means the only user-visible change is that we will
start printing a warning if the SDL-specific options are used in
Cocoa mode. This is a bugfix, because no_frame/alt_grab/ctrl_grab
are not used by Cocoa code.

Cc: Andreas Färber <andreas.faerber@web.de>
Cc: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Andreas Färber <andreas.faerber@web.de>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Replace *_vga_available() functions with class_names field
Eduardo Habkost [Wed, 11 Nov 2015 20:55:56 +0000 (18:55 -0200)]
vl: Replace *_vga_available() functions with class_names field

Instead of requiring a separate function for each VGA interface,
just enumerate the corresponding class names on struct
VGAInterfaceInfo.

Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Table-based select_vgahw()
Eduardo Habkost [Wed, 28 Oct 2015 21:19:58 +0000 (19:19 -0200)]
vl: Table-based select_vgahw()

Instead of implementing separate check functions for each vga
interface type, add a table enumerating the possible VGA
interfaces.

Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agovl: Use exit(1) when requested VGA interface is unavailable
Eduardo Habkost [Wed, 11 Nov 2015 20:33:42 +0000 (18:33 -0200)]
vl: Use exit(1) when requested VGA interface is unavailable

Instead of using exit(0), use exit(1) when an unavailable VGA
interface is used in the command-line to indicate it's an error.

Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agopc-dimm: correct comment of MemoryHotplugState
Cao jin [Thu, 3 Mar 2016 09:24:12 +0000 (17:24 +0800)]
pc-dimm: correct comment of MemoryHotplugState

correct comment and remove an unused macro. commit adcb4ee6
already correct its type

Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
8 years agotci: do not include exec/exec-all.h
Paolo Bonzini [Fri, 20 May 2016 11:57:31 +0000 (13:57 +0200)]
tci: do not include exec/exec-all.h

TCI does not need the runtime definition in exec-all.h.  It only needs the
host-side definitions in tcg/tcg.h.  Now that cpu.h is not included
everywhere, this caused a failure because exec-all.h does need cpu.h
but does not include it itself.

Fix by including the intended header.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1463745452-25831-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoaspeed: include qemu/log.h
Paolo Bonzini [Fri, 20 May 2016 11:57:32 +0000 (13:57 +0200)]
aspeed: include qemu/log.h

This is not visible with the default "log" trace backend.  With other
backends however trace.h does not include qemu/log.h, resulting in
build failures.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1463745452-25831-2-git-send-email-pbonzini@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Peter Maydell [Thu, 19 May 2016 15:54:12 +0000 (16:54 +0100)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer patches

# gpg: Signature made Thu 19 May 2016 16:09:27 BST using RSA key ID C88F2FD6
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>"

* remotes/kevin/tags/for-upstream: (31 commits)
  qemu-iotests: Fix regression in 136 on aio_read invalid
  qemu-iotests: Simplify 109 with unaligned qemu-img compare
  qemu-io: Fix recent UI updates
  block: clarify error message for qmp-eject
  qemu-iotests: Some more write_zeroes tests
  qcow2: Fix write_zeroes with partially allocated backing file cluster
  qcow2: fix condition in is_zero_cluster
  block: Propagate AioContext change to all children
  block: Remove BlockDriverState.blk
  block: Don't return throttling info in query-named-block-nodes
  block: Avoid bs->blk in bdrv_next()
  block: Add bdrv_has_blk()
  block: Remove bdrv_aio_multiwrite()
  blockjob: Don't touch BDS iostatus
  blockjob: Don't set iostatus of target
  block: User BdrvChild callback for device name
  block: Use BdrvChild callbacks for change_media/resize
  block: Don't check throttled reqs in bdrv_requests_pending()
  Revert "block: Forbid I/O throttling on nodes with multiple parents for 2.6"
  block: Remove bdrv_move_feature_fields()
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'mreitz/tags/pull-block-for-kevin-2016-05-19' into queue...
Kevin Wolf [Thu, 19 May 2016 14:59:46 +0000 (16:59 +0200)]
Merge remote-tracking branch 'mreitz/tags/pull-block-for-kevin-2016-05-19' into queue-block

Block patches

# gpg: Signature made Thu May 19 16:58:53 2016 CEST using RSA key ID E838ACAD
# gpg: Good signature from "Max Reitz <mreitz@redhat.com>"

* mreitz/tags/pull-block-for-kevin-2016-05-19:
  qemu-iotests: Fix regression in 136 on aio_read invalid
  qemu-iotests: Simplify 109 with unaligned qemu-img compare
  qemu-io: Fix recent UI updates

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqemu-iotests: Fix regression in 136 on aio_read invalid
Eric Blake [Mon, 16 May 2016 16:43:03 +0000 (10:43 -0600)]
qemu-iotests: Fix regression in 136 on aio_read invalid

Commit 093ea232 removed the ability for aio_read and aio_write
to artificially inflate the invalid statistics counters for
block devices, since it no longer flags unaligned offset or
length.  Add 'aio_read -i' and 'aio_write -i' to restore
the ability, and update test 136 to use it.

Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1463416983-28318-4-git-send-email-eblake@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
8 years agoqemu-iotests: Simplify 109 with unaligned qemu-img compare
Eric Blake [Mon, 16 May 2016 16:43:02 +0000 (10:43 -0600)]
qemu-iotests: Simplify 109 with unaligned qemu-img compare

For some time now, qemu-img compare has been able to compare
unaligned images.  So we no longer need test 109's hack of
resizing to sector boundaries before invoking compare.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 1463416983-28318-3-git-send-email-eblake@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
8 years agoqemu-io: Fix recent UI updates
Eric Blake [Mon, 16 May 2016 16:43:01 +0000 (10:43 -0600)]
qemu-io: Fix recent UI updates

Commit 770e0e0e [*] tried to add 'writev -f', but didn't tweak
the getopt() call to actually let it work.  Likewise, commit
c2e001c missed implementing 'aio_write -u -z'.  The latter commit
also introduced a leak of ctx.

[*] does it sound "ech0e" in here? :)

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1463416983-28318-2-git-send-email-eblake@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Thu, 19 May 2016 14:55:08 +0000 (15:55 +0100)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

NEED_CPU_H cleanups, big enough to deserve their own pull request.

# gpg: Signature made Thu 19 May 2016 15:42:37 BST using RSA key ID 78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"

* remotes/bonzini/tags/for-upstream: (52 commits)
  hw: clean up hw/hw.h includes
  hw: remove pio_addr_t
  cpu: move exec-all.h inclusion out of cpu.h
  exec: extract exec/tb-context.h
  hw: explicitly include qemu/log.h
  mips: move CP0 functions out of cpu.h
  arm: move arm_log_exception into .c file
  qemu-common: push cpu.h inclusion out of qemu-common.h
  acpi: do not use TARGET_PAGE_SIZE
  s390x: reorganize CSS bits between cpu.h and other headers
  dma: do not depend on kvm_enabled()
  gdbstub: remove unnecessary includes from gdbstub-xml.c
  qemu-common: stop including qemu/host-utils.h from qemu-common.h
  qemu-common: stop including qemu/bswap.h from qemu-common.h
  cpu: move endian-dependent load/store functions to cpu-all.h
  hw: cannot include hw/hw.h from user emulation
  hw: move CPU state serialization to migration/cpu.h
  hw: do not use VMSTATE_*TL
  include: poison symbols in osdep.h
  apic: move target-dependent definitions to cpu.h
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoblock: clarify error message for qmp-eject
John Snow [Wed, 18 May 2016 21:53:40 +0000 (17:53 -0400)]
block: clarify error message for qmp-eject

If you use HMP's eject but the CDROM tray is locked, you may get a
confusing error message informing you that the "tray isn't open."

As this is the point of eject, we can do a little better and help
clarify that the tray was locked and that it (might) open up later,
so try again.

It's not ideal, but it makes the semantics of the (legacy) eject
command more understandable to end users when they try to use it.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqemu-iotests: Some more write_zeroes tests
Kevin Wolf [Tue, 17 May 2016 16:37:39 +0000 (18:37 +0200)]
qemu-iotests: Some more write_zeroes tests

This covers some more write_zeroes cases which are relevant for the
recent qcow2 optimisations that check the allocation status of the
backing file for partial cluster write_zeroes requests.

This needs to be separate from 034 because we can only support qcow2 in
this test case for multiple reasons: We check the allocation status
after write_zeroes with 'qemu-img map' and the optimised behaviour that
produces zero clusters is only implemented in qcow2; second, the map
command returns offsets that are qcow2 specific; and finally, we also
use 512 byte clusters which aren't supported for formats like qed.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoqcow2: Fix write_zeroes with partially allocated backing file cluster
Kevin Wolf [Tue, 17 May 2016 16:42:23 +0000 (18:42 +0200)]
qcow2: Fix write_zeroes with partially allocated backing file cluster

In order to correctly check whether a given cluster is read as zero, we
don't only need to check whether bdrv_get_block_status_above() sets
BDRV_BLOCK_ZERO, but also if all sectors for the whole cluster have the
same status.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
8 years agoqcow2: fix condition in is_zero_cluster
Denis V. Lunev [Tue, 17 May 2016 09:15:42 +0000 (12:15 +0300)]
qcow2: fix condition in is_zero_cluster

We should check for (res & BDRV_BLOCK_ZERO) only. The situation when we
will have !(res & BDRV_BLOCK_DATA) and will not have BDRV_BLOCK_ZERO is
not possible for images with bdi.unallocated_blocks_are_zero == true.

For those images where it's false, however, it can happen and we must
not consider the data zeroed then or we would corrupt the image.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock: Propagate AioContext change to all children
Max Reitz [Tue, 17 May 2016 11:38:04 +0000 (13:38 +0200)]
block: Propagate AioContext change to all children

Instead of propagating any change of a BDS's AioContext only to its file
and backing children and letting driver-specific code do the rest, just
propagate it to all and drop the thus superfluous implementations of
bdrv_{at,de}tach_aio_context() in Quorum, blkverify and VMDK.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock: Remove BlockDriverState.blk
Kevin Wolf [Tue, 22 Mar 2016 17:38:44 +0000 (18:38 +0100)]
block: Remove BlockDriverState.blk

This patch removes the remaining users of bs->blk, which will allow us
to have multiple BBs on top of a single BDS. In the meantime, all checks
that are currently in place to prevent the user from creating such
setups can be switched to bdrv_has_blk() instead of accessing BDS.blk.

Future patches can allow them and e.g. enable users to mirror to a block
device that already has a BlockBackend on it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Don't return throttling info in query-named-block-nodes
Kevin Wolf [Tue, 22 Mar 2016 19:20:29 +0000 (20:20 +0100)]
block: Don't return throttling info in query-named-block-nodes

query-named-block-nodes should not return information that is related
to the attached BlockBackend rather than the node itself, so throttling
information needs to be removed from it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Avoid bs->blk in bdrv_next()
Kevin Wolf [Tue, 22 Mar 2016 17:58:50 +0000 (18:58 +0100)]
block: Avoid bs->blk in bdrv_next()

We need to introduce a separate BdrvNextIterator struct that can keep
more state than just the current BDS in order to avoid using the bs->blk
pointer.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Add bdrv_has_blk()
Kevin Wolf [Mon, 29 Feb 2016 09:50:38 +0000 (10:50 +0100)]
block: Add bdrv_has_blk()

In many cases we just want to know whether a BDS has at least one BB
attached, without needing to know the exact BB that is attached. In
contrast to bs->blk, this is still a valid question when more than one
BB can be attached, so just answer it by checking the parents list.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Remove bdrv_aio_multiwrite()
Kevin Wolf [Fri, 26 Feb 2016 12:50:43 +0000 (13:50 +0100)]
block: Remove bdrv_aio_multiwrite()

Since virtio-blk implements request merging itself these days, the only
remaining users are test cases for the function. That doesn't make the
function exactly useful any more.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoblockjob: Don't touch BDS iostatus
Kevin Wolf [Mon, 18 Apr 2016 13:14:11 +0000 (15:14 +0200)]
blockjob: Don't touch BDS iostatus

Block jobs don't actually make use of the iostatus for their BDSes, but
they manage a separate block job iostatus. Still, they require that it
is enabled for the source BDS and they enable it automatically for the
target and set the error handling mode - which ends up never being used
by the job.

This patch removes all of the BDS iostatus handling from the block job,
which removes another few bs->blk accesses.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblockjob: Don't set iostatus of target
Kevin Wolf [Mon, 18 Apr 2016 09:36:38 +0000 (11:36 +0200)]
blockjob: Don't set iostatus of target

When block job errors were introduced, we assigned the iostatus of the
target BDS "just in case". The field has never been accessible for the
user because the target isn't listed in query-block.

Before we can allow the user to have a second BlockBackend on the
target, we need to clean this up. If anything, we would want to set the
iostatus for the internal BB of the job (which we can always do later),
but certainly not for a separate BB which the job doesn't even use.

As a nice side effect, this gets us rid of another bs->blk use.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: User BdrvChild callback for device name
Kevin Wolf [Fri, 26 Feb 2016 09:22:16 +0000 (10:22 +0100)]
block: User BdrvChild callback for device name

In order to get rid of bs->blk for bdrv_get_device_name() and
bdrv_get_device_or_node_name(), ask all parents for their name and
simply pick the first one.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Use BdrvChild callbacks for change_media/resize
Kevin Wolf [Wed, 24 Feb 2016 14:13:35 +0000 (15:13 +0100)]
block: Use BdrvChild callbacks for change_media/resize

We want to get rid of BlockDriverState.blk in order to allow multiple
BlockBackends per BDS. Converting the device callbacks in block.c (which
assume a single BlockBackend) to per-child callbacks gets us rid of the
first few instances.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock: Don't check throttled reqs in bdrv_requests_pending()
Kevin Wolf [Tue, 22 Mar 2016 15:11:33 +0000 (16:11 +0100)]
block: Don't check throttled reqs in bdrv_requests_pending()

Checking whether there are throttled requests requires going to the
associated BlockBackend, which we want to avoid.

All users of bdrv_requests_pending() in block/io.c already call
bdrv_parent_drained_begin() first, which restarts all throttled
requests, so no throttled requests can be left here and this is removal
of dead code.

The remaining users (assertions during graph manipulation in block.c)
don't care about requests that are still queued in the BlockBackend and
haven't been issued for a BlockDriverState yet.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoRevert "block: Forbid I/O throttling on nodes with multiple parents for 2.6"
Kevin Wolf [Tue, 5 Apr 2016 09:10:50 +0000 (11:10 +0200)]
Revert "block: Forbid I/O throttling on nodes with multiple parents for 2.6"

This reverts commit 76b223200ef4fb09dd87f0e213159795eb68e7a5.

Now that I/O throttling is fully done on the BlockBackend level, there
is no reason any more to block I/O throttling for nodes with multiple
parents as the parents don't influence each other any more.

Conflicts:
block.c

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Remove bdrv_move_feature_fields()
Kevin Wolf [Mon, 29 Feb 2016 12:56:21 +0000 (13:56 +0100)]
block: Remove bdrv_move_feature_fields()

bdrv_move_feature_fields() and swap_feature_fields() are empty now, they
can be removed.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Decouple throttling from BlockDriverState
Kevin Wolf [Tue, 22 Mar 2016 12:00:08 +0000 (13:00 +0100)]
block: Decouple throttling from BlockDriverState

This moves the throttling related part of the BDS life cycle management
to BlockBackend. The throttling group reference is now kept even when no
medium is inserted.

With this commit, throttling isn't disabled and then re-enabled any more
during graph reconfiguration. This fixes the temporary breakage of I/O
throttling when used with live snapshots or block jobs that manipulate
the graph.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock/io: Quiesce parents between drained_begin/end
Kevin Wolf [Wed, 11 May 2016 12:57:23 +0000 (14:57 +0200)]
block/io: Quiesce parents between drained_begin/end

So far, bdrv_parent_drained_begin/end() was called for the duration of
the actual bdrv_drain() at the beginning of a drained section, but we
really should keep parents quiesced until the end of the drained
section.

This does not actually change behaviour at this point because the only
user of the .drained_begin/end BdrvChildRole callback is I/O throttling,
which already doesn't send any new requests after flushing its queue in
.drained_begin. The patch merely removes a trap for future users.

Reported-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Drain throttling queue with BdrvChild callback
Kevin Wolf [Tue, 22 Mar 2016 11:05:35 +0000 (12:05 +0100)]
block: Drain throttling queue with BdrvChild callback

This removes the last part of I/O throttling from block/io.c and moves
it to the BlockBackend.

Instead of having knowledge about throttling inside io.c, we can call a
BdrvChild callback .drained_begin/end, which happens to drain the
throttled requests for BlockBackend parents.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Introduce BdrvChild.opaque
Kevin Wolf [Wed, 24 Feb 2016 14:13:35 +0000 (15:13 +0100)]
block: Introduce BdrvChild.opaque

BlockBackends use it to get a back pointer from BdrvChild to
BlockBackend in any BdrvChildRole callbacks.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Move I/O throttling configuration functions to BlockBackend
Kevin Wolf [Mon, 21 Mar 2016 12:53:52 +0000 (13:53 +0100)]
block: Move I/O throttling configuration functions to BlockBackend

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Move actual I/O throttling to BlockBackend
Kevin Wolf [Mon, 21 Mar 2016 12:30:30 +0000 (13:30 +0100)]
block: Move actual I/O throttling to BlockBackend

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Move throttling fields from BDS to BB
Kevin Wolf [Mon, 21 Mar 2016 11:56:44 +0000 (12:56 +0100)]
block: Move throttling fields from BDS to BB

This patch changes where the throttling state is stored (used to be the
BlockDriverState, now it is the BlockBackend), but it doesn't actually
make it a BB level feature yet. For example, throttling is still
disabled when the BDS is detached from the BB.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: Convert throttle_group_get_name() to BlockBackend
Kevin Wolf [Mon, 21 Mar 2016 10:58:21 +0000 (11:58 +0100)]
block: Convert throttle_group_get_name() to BlockBackend

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoblock: throttle-groups: Use BlockBackend pointers internally
Kevin Wolf [Mon, 21 Mar 2016 10:30:57 +0000 (11:30 +0100)]
block: throttle-groups: Use BlockBackend pointers internally

As a first step towards moving I/O throttling to the BlockBackend level,
this patch changes all pointers in struct ThrottleGroup from referencing
a BlockDriverState to referencing a BlockBackend.

This change is valid because we made sure that throttling can only be
enabled on BDSes which have a BB attached.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>