sdk/emulator/qemu.git
8 years agohw/intc/arm_gicv3: Add IRQ handling CPU interface registers
Peter Maydell [Fri, 17 Jun 2016 14:23:48 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Add IRQ handling CPU interface registers

Add the CPU interface registers which deal with acknowledging
and dismissing interrupts.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-19-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Implement CPU i/f SGI generation registers
Peter Maydell [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement CPU i/f SGI generation registers

Implement the registers in the GICv3 CPU interface which generate
new SGI interrupts.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-18-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Implement gicv3_cpuif_update()
Peter Maydell [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement gicv3_cpuif_update()

Implement the gicv3_cpuif_update() function which deals with correctly
asserting IRQ and FIQ based on the current running priority of the CPU,
the priority of the highest priority pending interrupt and the CPU's
current exception level and security state.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-17-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Implement GICv3 CPU interface registers
Peter Maydell [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement GICv3 CPU interface registers

Implement the CPU interface registers for the GICv3; these are
CPU system registers, not MMIO registers.

This commit implements all the registers which are simple
accessors for GIC state, but not those which act as interfaces
for acknowledging, dismissing or generating interrupts. (Those
will be added in a later commit.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-16-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Implement gicv3_set_irq()
Peter Maydell [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement gicv3_set_irq()

Implement the code which updates the GIC state when an interrupt
input into the GIC is asserted.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-15-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Wire up distributor and redistributor MMIO regions
Peter Maydell [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Wire up distributor and redistributor MMIO regions

Wire up the MMIO functions exposed by the distributor and the
redistributor into MMIO regions exposed by the GICv3 device.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-14-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Implement GICv3 redistributor registers
Shlomo Pongratz [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement GICv3 redistributor registers

Implement the redistributor registers of a GICv3.

Signed-off-by: Shlomo Pongratz <shlomo.pongratz@huawei.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1465915112-29272-13-git-send-email-peter.maydell@linaro.org
[PMM: significantly overhauled/rewritten:
 * use the new data structures
 * restructure register read/write to handle different width accesses
   natively, since almost all registers are 32-bit only, rather
   than implementing everything as byte accesses
 * implemented security extension support
]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/intc/arm_gicv3: Implement GICv3 distributor registers
Shlomo Pongratz [Fri, 17 Jun 2016 14:23:47 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement GICv3 distributor registers

Implement the distributor registers of a GICv3.

Signed-off-by: Shlomo Pongratz <shlomo.pongratz@huawei.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1465915112-29272-12-git-send-email-peter.maydell@linaro.org
[PMM: significantly overhauled/rewritten:
 * use the new bitmap data structures
 * restructure register read/write to handle different width accesses
   natively, since almost all registers are 32-bit only, rather
   than implementing everything as byte accesses
 * implemented security extension support
]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/intc/arm_gicv3: Implement functions to identify next pending irq
Peter Maydell [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Implement functions to identify next pending irq

Implement the GICv3 logic to recalculate the highest priority pending
interrupt for each CPU after some part of the GIC state has changed.
We avoid unnecessary full recalculation where possible.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-11-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: ARM GICv3 device framework
Shlomo Pongratz [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: ARM GICv3 device framework

This patch includes the device class itself, some ID register
value functions which will be needed by both distributor
and redistributor, and some skeleton functions for handling
interrupts coming in and going out, which will be filled in
in a subsequent patch.

Signed-off-by: Shlomo Pongratz <shlomo.pongratz@huawei.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1465915112-29272-10-git-send-email-peter.maydell@linaro.org
[PMM: pulled this patch earlier in the sequence, and left
 some code out of it for a later patch]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
8 years agohw/intc/arm_gicv3: Add vmstate descriptors
Pavel Fedin [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Add vmstate descriptors

Add state structure descriptors for the GICv3 state. We mark
the KVM GICv3 device as having a migration blocker until the
code to save and restore the state in the kernel is implemented.

Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1465915112-29272-9-git-send-email-peter.maydell@linaro.org
[PMM: Adjust to renamed struct fields; switched to using uint32_t
 array backed bitmaps; add migration blocker setting]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/intc/arm_gicv3: Move irq lines into GICv3CPUState structure
Peter Maydell [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Move irq lines into GICv3CPUState structure

Move the GICv3 parent_irq and parent_fiq pointers into the
GICv3CPUState structure rather than giving them their own array.
This will make it easy to assert the IRQ and FIQ lines for a
particular CPU interface without having to know or calculate
the CPU index for the GICv3CPUState we are working on.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-8-git-send-email-peter.maydell@linaro.org

8 years agohw/intc/arm_gicv3: Add state information
Pavel Fedin [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
hw/intc/arm_gicv3: Add state information

Add state information to GICv3 object structure and implement
arm_gicv3_common_reset().

This commit includes accessor functions for the fields which are
stored as bitmaps in uint32_t arrays.

Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1465915112-29272-7-git-send-email-peter.maydell@linaro.org
[PMM: significantly overhauled:
 * Add missing qom/cpu.h include
 * Remove legacy-only state fields (we can add them later if/when we add
   legacy emulation)
 * Use arrays of uint32_t to store the various distributor bitmaps,
   and provide accessor functions for the various set/test/etc operations
 * Add various missing register offset #defines
 * Accessor macros which combine distributor and redistributor behaviour
   removed
 * Fields in state structures renamed to match architectural register names
 * Corrected the reset value for GICR_IENABLER0 since we don't support
   legacy mode
 * Added ARM_LINUX_BOOT_IF interface for "we are directly booting a kernel in
   non-secure" so that we can fake up the firmware-mandated reconfiguration
   only when we need it
]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
8 years agotarget-arm: Add mp-affinity property for ARM CPU class
Pavel Fedin [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
target-arm: Add mp-affinity property for ARM CPU class

This allows to override default affinity IDs on a per-machine basis, and
possibility to retrieve IDs will be used by vGICv3 live migration code.

Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-6-git-send-email-peter.maydell@linaro.org

8 years agotarget-arm: Provide hook to tell GICv3 about changes of security state
Peter Maydell [Fri, 17 Jun 2016 14:23:46 +0000 (15:23 +0100)]
target-arm: Provide hook to tell GICv3 about changes of security state

The GICv3 CPU interface needs to know when the CPU it is attached
to makes an exception level or mode transition that changes the
security state, because whether it is asserting IRQ or FIQ can change
depending on these things. Provide a mechanism for letting the GICv3
device register a hook to be called on such changes.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-5-git-send-email-peter.maydell@linaro.org

8 years agotarget-arm: Define new arm_is_el3_or_mon() function
Peter Maydell [Fri, 17 Jun 2016 14:23:45 +0000 (15:23 +0100)]
target-arm: Define new arm_is_el3_or_mon() function

The GICv3 system registers need to know if the CPU is AArch64
in EL3 or AArch32 in Monitor mode. This happens to be the first
part of the check for arm_is_secure(), so factor it out into a
new arm_is_el3_or_mon() function that the GIC can also use.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-4-git-send-email-peter.maydell@linaro.org

8 years agobitops.h: Implement half-shuffle and half-unshuffle ops
Peter Maydell [Fri, 17 Jun 2016 14:23:45 +0000 (15:23 +0100)]
bitops.h: Implement half-shuffle and half-unshuffle ops

A half-shuffle operation takes a word with zeros in the high half:
 0000 0000 0000 0000 ABCD EFGH IJKL MNOP
and spreads the bits out so they are in every other bit of the word:
 0A0B 0C0D 0E0F 0G0H 0I0J 0K0L 0M0N 0O0P
A half-unshuffle performs the reverse operation.

Provide functions in bitops.h which implement these operations
for 32-bit and 64-bit inputs, and add tests for them.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-3-git-send-email-peter.maydell@linaro.org

8 years agomigration: Define VMSTATE_UINT64_2DARRAY
Peter Maydell [Fri, 17 Jun 2016 14:23:45 +0000 (15:23 +0100)]
migration: Define VMSTATE_UINT64_2DARRAY

Define a VMSTATE_UINT64_2DARRAY macro, to go with the ones we
already have for other type sizes.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Tested-by: Shannon Zhao <shannon.zhao@linaro.org>
Message-id: 1465915112-29272-2-git-send-email-peter.maydell@linaro.org

8 years agoMerge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.7-5' into...
Peter Maydell [Fri, 17 Jun 2016 13:09:46 +0000 (14:09 +0100)]
Merge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.7-5' into staging

Migration:

 - many compression/decompression fixes
 - trace improvements
 - static checker fix for detecting size mismatch in unused fields
 - fix VM save after snapshot

# gpg: Signature made Fri 17 Jun 2016 13:59:44 BST
# gpg:                using RSA key 0xEB0B4DFC657EF670
# gpg: Good signature from "Amit Shah <amit@amitshah.net>"
# gpg:                 aka "Amit Shah <amit@kernel.org>"
# gpg:                 aka "Amit Shah <amitshah@gmx.net>"
# Primary key fingerprint: 48CA 3722 5FE7 F4A8 B337  2735 1E9A 3B5F 8540 83B6
#      Subkey fingerprint: CC63 D332 AB8F 4617 4529  6534 EB0B 4DFC 657E F670

* remotes/amit-migration/tags/migration-for-2.7-5:
  vmstate-static-checker: fix size mismatch detection in unused fields
  migration: code clean up
  migration: refine the decompression code
  migration: refine the compression code
  migration: protect the quit flag by lock
  migration: refine ram_save_compressed_page
  qemu-file: Fix qemu_put_compression_data flaw
  migration: remove useless code
  migration: Fix a potential issue
  migration: Fix multi-thread compression bug
  migration: fix inability to save VM after snapshot
  migration: Trace improvements
  migration: Don't use *_to_cpup() and cpu_to_*w()

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agovmstate-static-checker: fix size mismatch detection in unused fields
Amit Shah [Fri, 17 Jun 2016 12:16:39 +0000 (17:46 +0530)]
vmstate-static-checker: fix size mismatch detection in unused fields

If a field changed from something to unused, the checker wasn't flagging
if the field size mismatched.  This was noticed in:

http://thread.gmane.org/gmane.comp.emulators.qemu/419802

where the 4->1 size change along with field name change to 'unused'
wasn't being flagged.  Fix this.

Signed-off-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <d7ec03a9b2edfa0616764887a51ba8f64fdd3f68.1466165736.git.amit.shah@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: code clean up
Liang Li [Thu, 5 May 2016 07:32:59 +0000 (15:32 +0800)]
migration: code clean up

Use 'QemuMutex comp_done_lock' and 'QemuCond comp_done_cond' instead
of 'QemuMutex *comp_done_lock' and 'QemuCond comp_done_cond'. To keep
consistent with 'QemuMutex decomp_done_lock' and
'QemuCond comp_done_cond'.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-10-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: refine the decompression code
Liang Li [Thu, 5 May 2016 07:32:58 +0000 (15:32 +0800)]
migration: refine the decompression code

The current code for multi-thread decompression is not clear,
especially in the aspect of using lock. Refine the code
to make it clear.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-9-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: refine the compression code
Liang Li [Thu, 5 May 2016 07:32:57 +0000 (15:32 +0800)]
migration: refine the compression code

The current code for multi-thread compression is not clear,
especially in the aspect of using lock. Refine the code
to make it clear.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-8-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: protect the quit flag by lock
Liang Li [Thu, 5 May 2016 07:32:56 +0000 (15:32 +0800)]
migration: protect the quit flag by lock

quit_comp_thread and quit_decomp_thread are accessed by several
thread, it's better to protect them with locks. We use a per
thread flag to replace the global one, and the new flag is protected
by a lock.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-7-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: refine ram_save_compressed_page
Liang Li [Thu, 5 May 2016 07:32:55 +0000 (15:32 +0800)]
migration: refine ram_save_compressed_page

Use qemu_put_compression_data to do the compression directly
instead of using do_compress_ram_page, avoid some data copy.
very small improvement, at the same time, add code to check
if the compression is successful.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-6-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoqemu-file: Fix qemu_put_compression_data flaw
Liang Li [Thu, 5 May 2016 07:32:54 +0000 (15:32 +0800)]
qemu-file: Fix qemu_put_compression_data flaw

Current qemu_put_compression_data can only work with no writable
QEMUFile, and can't work with the writable QEMUFile. But it does
not provide any measure to prevent users from using it with a
writable QEMUFile.

We should fix this flaw to make it works with writable QEMUFile.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Suggested-by: Juan Quintela <quintela@redhat.com>
Message-Id: <1462433579-13691-5-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: remove useless code
Liang Li [Thu, 5 May 2016 07:32:53 +0000 (15:32 +0800)]
migration: remove useless code

page_buffer is set twice repeatedly, remove the previous set.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <1462433579-13691-4-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: Fix a potential issue
Liang Li [Thu, 5 May 2016 07:32:52 +0000 (15:32 +0800)]
migration: Fix a potential issue

At the end of live migration and before vm_start() on the destination
side, we should make sure all the decompression tasks are finished, if
this can not be guaranteed, the VM may get the incorrect memory data,
or the updated memory may be overwritten by the decompression thread.
Add the code to fix this potential issue.

Suggested-by: David Alan Gilbert <dgilbert@redhat.com>
Suggested-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-3-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: Fix multi-thread compression bug
Liang Li [Thu, 5 May 2016 07:32:51 +0000 (15:32 +0800)]
migration: Fix multi-thread compression bug

Recently, a bug related to multiple thread compression feature for
live migration is reported. The destination side will be blocked
during live migration if there are heavy workload in host and
memory intensive workload in guest, this is most likely to happen
when there is one decompression thread.

Some parts of the decompression code are incorrect:
1. The main thread receives data from source side will enter a busy
loop to wait for a free decompression thread.
2. A lock is needed to protect the decomp_param[idx]->start, because
it is checked in the main thread and is updated in the decompression
thread.

Fix these two issues by following the code pattern for compression.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Reported-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Tested-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1462433579-13691-2-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: fix inability to save VM after snapshot
Denis V. Lunev [Wed, 15 Jun 2016 15:06:43 +0000 (18:06 +0300)]
migration: fix inability to save VM after snapshot

The following sequence of operations fails:
    virsh start vm
    virsh snapshot-create vm
    virshh save vm --file file
with the following error
    error: Failed to save domain vm to file
    error: internal error: unable to execute QEMU command 'migrate':
    There's a migration process in progress

The problem is that qemu_savevm_state() calls migrate_init() which sets
migration state to MIGRATION_STATUS_SETUP and never cleaned it up.
This patch do the job.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Amit Shah <amit.shah@redhat.com>
Message-Id: <1466003203-26263-1-git-send-email-den@openvz.org>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: Trace improvements
Dr. David Alan Gilbert [Tue, 14 Jun 2016 09:36:26 +0000 (10:36 +0100)]
migration: Trace improvements

A couple of improvements to tracing that have come out of helping
people with migration problems:
  * vmstate_n_elems trace the count/name - for when you have problems
    getting array counts right
  * vmstate_subsection_load_bad - add the idstr, for when you receive a
    subsection you weren't expecting.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1465896986-16132-1-git-send-email-dgilbert@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: Don't use *_to_cpup() and cpu_to_*w()
Peter Maydell [Fri, 10 Jun 2016 16:09:22 +0000 (17:09 +0100)]
migration: Don't use *_to_cpup() and cpu_to_*w()

The *_to_cpup() and cpu_to_*w() functions just compose a pointer
dereference with a byteswap. Instead use ld*_p() and st*_p(),
which handle potential pointer misalignment and avoid the need
to cast the pointer.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1465574962-2710-1-git-send-email-peter.maydell@linaro.org>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.7-20160617' into staging
Peter Maydell [Fri, 17 Jun 2016 11:36:27 +0000 (12:36 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.7-20160617' into staging

ppc patch queue for 2016-06-17

Here's the current accumulated set of spapr, ppc and related patches.
  * The big thing in here is CPU hotplug for spapr
    - This includes a number of acked generic changes adding new
      infrastructure for hotplugging cpu cores
  * A number of TCG bug fixes are also included
  * This adds a new testcase to make it harder to accidentally break
    Macintosh (and other openbios) platforms

# gpg: Signature made Fri 17 Jun 2016 07:35:29 BST
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-for-2.7-20160617:
  spapr: implement query-hotpluggable-cpus callback
  hmp: Add 'info hotpluggable-cpus' HMP command
  QMP: Add query-hotpluggable-cpus
  spapr: CPU hot unplug support
  spapr: CPU hotplug support
  spapr: convert boot CPUs into CPU core devices
  spapr: Move spapr_cpu_init() to spapr_cpu_core.c
  spapr: Abstract CPU core device and type specific core devices
  qom: API to get instance_size of a type
  spapr_drc: Prevent detach racing against attach for CPU DR
  xics,xics_kvm: Handle CPU unplug correctly
  cpu: Abstract CPU core type
  qdev: hotplug: Introduce HotplugHandler.pre_plug() callback
  target-ppc: Fix rlwimi, rlwinm, rlwnm
  vfio: Fix broken EEH
  target-ppc: Bug in BookE wait instruction
  ppc / sparc: Add a tester for checking whether OpenBIOS runs successfully
  hw/ppc/spapr: Silence deprecation message in qtest mode

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Fri, 17 Jun 2016 10:25:46 +0000 (11:25 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc, pci, virtio: new features, cleanups, fixes

Beginning of reconnect support for vhost-user.
Misc cleanups and fixes.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Fri 17 Jun 2016 01:28:39 BST
# gpg:                using RSA key 0x281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream:
  MAINTAINERS: add Marcel to PCI
  msi_init: change return value to 0 on success
  fix some coding style problems
  pci core: assert ENOSPC when add capability
  test: start vhost-user reconnect test
  tests: append i386 tests
  vhost-net: save & restore vring enable state
  vhost-net: save & restore vhost-user acked features
  vhost-net: do not crash if backend is not present
  vhost-user: disconnect on start failure
  qemu-char: add qemu_chr_disconnect to close a fd accepted by listen fd
  tests/vhost-user-bridge: workaround stale vring base
  tests/vhost-user-bridge: add client mode
  vhost-user: add ability to know vhost-user backend disconnection
  pci: fix pci_requester_id()

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Conflicts:
tests/Makefile.include

8 years agospapr: implement query-hotpluggable-cpus callback
Igor Mammedov [Fri, 10 Jun 2016 00:59:08 +0000 (06:29 +0530)]
spapr: implement query-hotpluggable-cpus callback

It returns a list of present/possible to hotplug CPU
objects with a list of properties to use with
device_add.

in spapr case returned list would looks like:
-> { "execute": "query-hotpluggable-cpus" }
<- {"return": [
     { "props": { "core": 8 }, "type": "POWER8-spapr-cpu-core",
       "vcpus-count": 2 },
     { "props": { "core": 0 }, "type": "POWER8-spapr-cpu-core",
       "vcpus-count": 2,
       "qom-path": "/machine/unattached/device[0]"}
   ]}'

TODO:
  add 'node' property for core <-> numa node mapping

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agohmp: Add 'info hotpluggable-cpus' HMP command
Bharata B Rao [Fri, 10 Jun 2016 00:59:07 +0000 (06:29 +0530)]
hmp: Add 'info hotpluggable-cpus' HMP command

This is the HMP equivalent for QMP query-hotpluggable-cpus.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
[dwg: Fixed problem with printf formats on 32-bit host]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoQMP: Add query-hotpluggable-cpus
Igor Mammedov [Fri, 10 Jun 2016 00:59:06 +0000 (06:29 +0530)]
QMP: Add query-hotpluggable-cpus

It will allow mgmt to query present and hotpluggable CPU objects,
it is required from a target platform that wishes to support command
to implement and set MachineClass.query_hotpluggable_cpus callback,
which will return a list of possible CPU objects with options that
would be needed for hotplugging possible CPU objects.

There are:
'type': 'str' - QOM CPU object type for usage with device_add
'vcpus-count': 'int' - number of logical VCPU threads per
                        CPU object (mgmt needs to know)

and a set of optional fields that are to used for hotplugging a CPU
objects and would allows mgmt tools to know what/where it could be
hotplugged;
[node],[socket],[core],[thread]

For present CPUs there is a 'qom-path' field which would allow mgmt to
inspect whatever object/abstraction the target platform considers
as CPU object.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr: CPU hot unplug support
Bharata B Rao [Fri, 10 Jun 2016 00:59:05 +0000 (06:29 +0530)]
spapr: CPU hot unplug support

Remove the CPU core device by removing the underlying CPU thread devices.
Hot removal of CPU for sPAPR guests is achieved by sending the hot unplug
notification to the guest. Release the vCPU object after CPU hot unplug so
that vCPU fd can be parked and reused.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr: CPU hotplug support
Bharata B Rao [Fri, 10 Jun 2016 00:59:04 +0000 (06:29 +0530)]
spapr: CPU hotplug support

Set up device tree entries for the hotplugged CPU core and use the
exising RTAS event logging infrastructure to send CPU hotplug notification
to the guest.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr: convert boot CPUs into CPU core devices
Bharata B Rao [Fri, 10 Jun 2016 00:59:03 +0000 (06:29 +0530)]
spapr: convert boot CPUs into CPU core devices

Introduce sPAPRMachineClass.dr_cpu_enabled to indicate support for
CPU core hotplug. Initialize boot time CPUs as core deivces and prevent
topologies that result in partially filled cores. Both of these are done
only if CPU core hotplug is supported.

Note: An unrelated change in the call to xics_system_init() is done
in this patch as it makes sense to use the local variable smt introduced
in this patch instead of kvmppc_smt_threads() call here.

TODO: We derive sPAPR core type by looking at -cpu <model>. However
we don't take care of "compat=" feature yet for boot time as well
as hotplug CPUs.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr: Move spapr_cpu_init() to spapr_cpu_core.c
Bharata B Rao [Fri, 10 Jun 2016 00:59:02 +0000 (06:29 +0530)]
spapr: Move spapr_cpu_init() to spapr_cpu_core.c

Start consolidating CPU init related routines in spapr_cpu_core.c. As
part of this, move spapr_cpu_init() and its dependencies from spapr.c
to spapr_cpu_core.c

No functionality change in this patch.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
[dwg: Rename TIMEBASE_FREQ to SPAPR_TIMEBASE_FREQ, since it's now in a
 public(ish) header]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr: Abstract CPU core device and type specific core devices
Bharata B Rao [Fri, 10 Jun 2016 00:59:01 +0000 (06:29 +0530)]
spapr: Abstract CPU core device and type specific core devices

Add sPAPR specific abastract CPU core device that is based on generic
CPU core device. Use this as base type to create sPAPR CPU specific core
devices.

TODO:
- Add core types for other remaining CPU types
- Handle CPU model alias correctly

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoqom: API to get instance_size of a type
Bharata B Rao [Fri, 10 Jun 2016 00:59:00 +0000 (06:29 +0530)]
qom: API to get instance_size of a type

Add an API object_type_get_size(const char *typename) that returns the
instance_size of the give typename.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agospapr_drc: Prevent detach racing against attach for CPU DR
Bharata B Rao [Thu, 12 May 2016 03:48:21 +0000 (09:18 +0530)]
spapr_drc: Prevent detach racing against attach for CPU DR

If a CPU is hot removed while hotplug of the same is still in progress,
the guest crashes. Prevent this by ensuring that detach is done only
after attach has completed.

The existing code already prevents such race for PCI hotplug. However
given that CPU is a logical DR unlike PCI and starts with ISOLATED
state, we need a logic that works for CPU too.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
               [Don't set awaiting_attach for PCI devices]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoxics,xics_kvm: Handle CPU unplug correctly
Bharata B Rao [Thu, 12 May 2016 03:48:20 +0000 (09:18 +0530)]
xics,xics_kvm: Handle CPU unplug correctly

XICS is setup for each CPU during initialization. Provide a routine
to undo the same when CPU is unplugged. While here, move ss->cs management
into xics from xics_kvm since there is nothing KVM specific in it.
Also ensure xics reset doesn't set irq for CPUs that are already unplugged.

This allows reboot of a VM that has undergone CPU hotplug and unplug
to work correctly.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agocpu: Abstract CPU core type
Bharata B Rao [Thu, 12 May 2016 03:48:16 +0000 (09:18 +0530)]
cpu: Abstract CPU core type

Add an abstract CPU core type that could be used by machines that want
to define and hotplug CPUs in core granularity.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
               [Integer core property]
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
[dwg: changed property names to 'core-id' and 'nr-threads']
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoqdev: hotplug: Introduce HotplugHandler.pre_plug() callback
Igor Mammedov [Thu, 12 May 2016 03:48:15 +0000 (09:18 +0530)]
qdev: hotplug: Introduce HotplugHandler.pre_plug() callback

pre_plug callback is to be called before device.realize() is executed.
This would allow to check/set device's properties from HotplugHandler.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agotarget-ppc: Fix rlwimi, rlwinm, rlwnm
Richard Henderson [Thu, 16 Jun 2016 19:04:04 +0000 (12:04 -0700)]
target-ppc: Fix rlwimi, rlwinm, rlwnm

In 63ae0915f8ec, I arranged to use a 32-bit rotate, without
considering the effect of a mask value that wraps around to
the high bits of the word.

Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agovfio: Fix broken EEH
Gavin Shan [Wed, 15 Jun 2016 04:28:27 +0000 (14:28 +1000)]
vfio: Fix broken EEH

vfio_eeh_container_op() is the backend that communicates with
host kernel to support EEH functionality in QEMU. However, the
functon should return the value from host kernel instead of 0
unconditionally.

dwg: Specifically the problem occurs for the handful of EEH
sub-operations which can return a non-zero, non-error result.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
[dwg: clarification to commit message]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agotarget-ppc: Bug in BookE wait instruction
Jakub Horak [Mon, 6 Jun 2016 08:47:28 +0000 (10:47 +0200)]
target-ppc: Bug in BookE wait instruction

Fixed bug in code generation for the PowerPC "wait" instruction. It
doesn't make sense to store a non-initialized register.

Signed-off-by: Jakub Horak <thement@ibawizard.net>
[dwg: revised commit message]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoppc / sparc: Add a tester for checking whether OpenBIOS runs successfully
Thomas Huth [Tue, 14 Jun 2016 13:57:56 +0000 (15:57 +0200)]
ppc / sparc: Add a tester for checking whether OpenBIOS runs successfully

Since the mac99 and g3beige PowerPC machines recently broke without
being noticed, it would be good to have a tester for "make check"
that detects such issues immediately. A simple way to test the firmware
of these machines is to use the "-prom-env" parameter of QEMU. This
parameter can be used to put some Forth code into the 'boot-command'
firmware variable which then can signal success to the tester by
writing a magic value to a known memory location. And since some of the
Sparc machines are also using OpenBIOS, they are now tested with this
prom-env-tester, too.

Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
[dwg: Removed sparc64, because it trips a TCG bug on 32-bit hosts]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoMAINTAINERS: add Marcel to PCI
Michael S. Tsirkin [Mon, 13 Jun 2016 20:06:32 +0000 (23:06 +0300)]
MAINTAINERS: add Marcel to PCI

Marcel is reviewing PCI patches anyway, things will
be easier if people remember to Cc him.

Cc: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agomsi_init: change return value to 0 on success
Cao jin [Fri, 10 Jun 2016 09:54:32 +0000 (17:54 +0800)]
msi_init: change return value to 0 on success

No caller use its return value as msi capability offset, also in order
to make its return behaviour consistent with msix_init().

cc: Michael S. Tsirkin <mst@redhat.com>
cc: Paolo Bonzini <pbonzini@redhat.com>
cc: Hannes Reinecke <hare@suse.de>
cc: Markus Armbruster <armbru@redhat.com>
cc: Marcel Apfelbaum <marcel@redhat.com>

Acked-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agofix some coding style problems
Cao jin [Fri, 10 Jun 2016 09:54:23 +0000 (17:54 +0800)]
fix some coding style problems

It has:
1. More newlines make the code block well separated.
2. Add more comments for msi_init.
3. Fix a indentation in vmxnet3.c.
4. ioh3420 & xio3130_downstream: put PCI Express capability init function
   together, make it more readable.

cc: Michael S. Tsirkin <mst@redhat.com>
cc: Markus Armbruster <armbru@redhat.com>
cc: Marcel Apfelbaum <marcel@redhat.com>
cc: Dmitry Fleytman <dmitry@daynix.com>
cc: Jason Wang <jasowang@redhat.com>

Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agopci core: assert ENOSPC when add capability
Cao jin [Fri, 10 Jun 2016 09:54:22 +0000 (17:54 +0800)]
pci core: assert ENOSPC when add capability

ENOSPC is programming error, assert it for debugging.

cc: Michael S. Tsirkin <mst@redhat.com>
cc: Marcel Apfelbaum <marcel@redhat.com>
cc: Markus Armbruster <armbru@redhat.com>

Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Cao jin <caoj.fnst@cn.fujitsu.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agotest: start vhost-user reconnect test
Marc-André Lureau [Mon, 6 Jun 2016 16:45:08 +0000 (18:45 +0200)]
test: start vhost-user reconnect test

This is a simple reconnect test, that simply checks if vhost-user
reconnection is possible and restore the state. A more complete test
would actually manipulate and check the ring contents (such extended
testing would benefit from the libvhost-user proposed in QEMU list to
avoid duplication of ring manipulations)

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agotests: append i386 tests
Marc-André Lureau [Mon, 6 Jun 2016 16:45:07 +0000 (18:45 +0200)]
tests: append i386 tests

Do not overwrite x86-64 tests, re-enable vhost-user-test.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovhost-net: save & restore vring enable state
Marc-André Lureau [Mon, 6 Jun 2016 16:45:06 +0000 (18:45 +0200)]
vhost-net: save & restore vring enable state

A driver may change the vring enable state at run time but vhost-user
backend may not be present (a contrived example is when the backend is
disconnected and the device is reconfigured after driver rebinding)

Restore the vring state when the vhost-user backend is started, so it
can process the ring.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovhost-net: save & restore vhost-user acked features
Marc-André Lureau [Mon, 6 Jun 2016 16:45:05 +0000 (18:45 +0200)]
vhost-net: save & restore vhost-user acked features

The initial vhost-user connection sets the features to be negotiated
with the driver. Renegotiation isn't possible without device reset.

To handle reconnection of vhost-user backend, ensure the same set of
features are provided, and reuse already acked features.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovhost-net: do not crash if backend is not present
Marc-André Lureau [Mon, 6 Jun 2016 16:45:04 +0000 (18:45 +0200)]
vhost-net: do not crash if backend is not present

Do not crash when backend is not present while enabling the ring. A
following patch will save the enabled state so it can be restored once
the backend is started.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovhost-user: disconnect on start failure
Marc-André Lureau [Mon, 6 Jun 2016 16:45:03 +0000 (18:45 +0200)]
vhost-user: disconnect on start failure

If the backend failed to start (for example feature negociation failed),
do not exit, but disconnect the char device instead. Slightly more
robust for reconnect case.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agoqemu-char: add qemu_chr_disconnect to close a fd accepted by listen fd
Tetsuya Mukawa [Mon, 6 Jun 2016 16:45:02 +0000 (18:45 +0200)]
qemu-char: add qemu_chr_disconnect to close a fd accepted by listen fd

The patch introduces qemu_chr_disconnect(). The function is used for
closing a fd accepted by listen fd. Though we already have qemu_chr_delete(),
but it closes not only accepted fd but also listen fd. This new function
is used when we still want to keep listen fd.

Signed-off-by: Tetsuya Mukawa <mukawa@igel.co.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agotests/vhost-user-bridge: workaround stale vring base
Marc-André Lureau [Mon, 6 Jun 2016 16:45:01 +0000 (18:45 +0200)]
tests/vhost-user-bridge: workaround stale vring base

This patch is a similar solution to what Yuanhan Liu/Huawei Xie have
suggested for DPDK. When vubr quits (killed or crashed), a restart of
vubr would get stale vring base from QEMU. That would break the kernel
virtio net completely, making it non-work any more, unless a driver
reset is done.

So, instead of getting the stale vring base from QEMU, Huawei suggested
we could get a proper one from used->idx. This works because the queues
packets are processed in order.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agotests/vhost-user-bridge: add client mode
Marc-André Lureau [Mon, 6 Jun 2016 16:45:00 +0000 (18:45 +0200)]
tests/vhost-user-bridge: add client mode

If -c is specified, vubr will try to connect to the socket instead of
listening for connections.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agovhost-user: add ability to know vhost-user backend disconnection
Tetsuya Mukawa [Mon, 6 Jun 2016 16:44:59 +0000 (18:44 +0200)]
vhost-user: add ability to know vhost-user backend disconnection

Current QEMU cannot detect vhost-user backend disconnection. The
patch adds ability to know it.
To know disconnection, add watcher to detect G_IO_HUP event. When
G_IO_HUP event is detected, the disconnected socket will be read
to cause a CHR_EVENT_CLOSED.

Signed-off-by: Tetsuya Mukawa <mukawa@igel.co.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Victor Kaplansky <victork@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agopci: fix pci_requester_id()
Peter Xu [Tue, 17 May 2016 11:26:10 +0000 (19:26 +0800)]
pci: fix pci_requester_id()

This fix SID verification failure when IOMMU IR is enabled with PCI
bridges. Existing pci_requester_id() is more like getting BDF info
only. Renaming it to pci_get_bdf(). Meanwhile, we provide the correct
implementation to get requester ID. VT-d spec 5.1.1 is a good reference
to go, though it talks only about interrupt delivery, the rule works
exactly the same for non-interrupt cases.

Currently, there are three use cases for pci_requester_id():

- PCIX status bits: here we need BDF only, not requester ID. Replacing
  with pci_get_bdf().
- PCIe Error injection and MSI delivery: for both these cases, we are
  looking for requester IDs. Here we should use the new impl.

To avoid a PCI walk every time we send MSI message, one requester_id
cache field is added to PCIDevice to cache the result when initialize
PCI device.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
8 years agohw/ppc/spapr: Silence deprecation message in qtest mode
Thomas Huth [Tue, 14 Jun 2016 17:23:03 +0000 (19:23 +0200)]
hw/ppc/spapr: Silence deprecation message in qtest mode

When running "make check", there is currently always an error message
saying "spapr-pci-vfio-host-bridge is deprecated". This happens because
the QOM tests are instantiating all possible devices, and the error
message is currently located in the instance_init() function of the
device. Since it is legal for the tests to instantiate a device without
using it, the error message should be silenced when we're running in
test mode.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
8 years agoMerge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
Peter Maydell [Thu, 16 Jun 2016 16:58:45 +0000 (17:58 +0100)]
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

* KVM startup speedup (Chao Peng)
* configure fixes and cleanups (David, Thomas)
* ctags fix (Sergey)
* NBD cleanups (Peter, Eric)
* "-L help" command line option (Richard)
* More esp.c bugfixes (me, Prasad)
* KVM_CAP_MAX_VCPU_ID support (Greg)

# gpg: Signature made Thu 16 Jun 2016 17:39:10 BST
# gpg:                using RSA key 0xBFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream: (29 commits)
  vl: smp_parse: cleanups
  scsi: esp: make cmdbuf big enough for maximum CDB size
  scsi: esp: clean up handle_ti/esp_do_dma if s->do_cmd
  scsi: esp: respect FIFO invariant after message phase
  scsi: esp: check buffer length before reading scsi command
  nbd: Avoid magic number for NBD max name size
  nbd: Detect servers that send unexpected error values
  nbd: Clean up ioctl handling of qemu-nbd -c
  nbd: Group all Linux-specific ioctl code in one place
  nbd: Reject unknown request flags
  nbd: Improve server handling of bogus commands
  nbd: Quit server after any write error
  nbd: More debug typo fixes, use correct formats
  nbd: Use BDRV_REQ_FUA for better FUA where supported
  vl.c: Add '-L help' which lists data dirs.
  KVM: use KVM_CAP_MAX_VCPU_ID
  scsi-disk: Use (unsigned long) typecasts when using "%lu" format string
  target-i386: kvm: cache KVM_GET_SUPPORTED_CPUID data
  nbd: simplify the nbd_request and nbd_reply structs
  nbd: Don't use cpu_to_*w() functions
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agovl: smp_parse: cleanups
Andrew Jones [Fri, 10 Jun 2016 17:40:12 +0000 (19:40 +0200)]
vl: smp_parse: cleanups

No functional changes; only some code movement and removal of
dead code (impossible conditions). Also, max_cpus can be
initialized to 1, like smp_cpus, because it's either set by the
user or set to smp_cpus, when smp_cpus is set by the user, or
set to 1, when nothing is set.

Signed-off-by: Andrew Jones <drjones@redhat.com>
Message-Id: <1465580427-13596-2-git-send-email-drjones@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscsi: esp: make cmdbuf big enough for maximum CDB size
Prasad J Pandit [Wed, 15 Jun 2016 22:22:35 +0000 (00:22 +0200)]
scsi: esp: make cmdbuf big enough for maximum CDB size

While doing DMA read into ESP command buffer 's->cmdbuf', it could
write past the 's->cmdbuf' area, if it was transferring more than 16
bytes.  Increase the command buffer size to 32, which is maximum when
's->do_cmd' is set, and add a check on 'len' to avoid OOB access.

Reported-by: Li Qiang <liqiang6-s@360.cn>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscsi: esp: clean up handle_ti/esp_do_dma if s->do_cmd
Paolo Bonzini [Wed, 15 Jun 2016 12:29:33 +0000 (14:29 +0200)]
scsi: esp: clean up handle_ti/esp_do_dma if s->do_cmd

Avoid duplicated code between esp_do_dma and handle_ti.  esp_do_dma
has the same code that handle_ti contains after the call to esp_do_dma;
but the code in handle_ti is never reached because it is in an "else if".
Remove the else and also the pointless return.

esp_do_dma also has a partially dead assignment of the to_device
variable.  Sink it to the point where it's actually used.

Finally, assert that the other caller of esp_do_dma (esp_transfer_data)
only transfers data and not a command.  This is true because get_cmd
cancels the old request synchronously before its caller handle_satn_stop
sets do_cmd to 1.

Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscsi: esp: respect FIFO invariant after message phase
Paolo Bonzini [Tue, 14 Jun 2016 13:10:24 +0000 (15:10 +0200)]
scsi: esp: respect FIFO invariant after message phase

The FIFO contains two bytes; hence the write ptr should be two bytes ahead
of the read pointer.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscsi: esp: check buffer length before reading scsi command
Prasad J Pandit [Tue, 31 May 2016 17:53:27 +0000 (23:23 +0530)]
scsi: esp: check buffer length before reading scsi command

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() in non-DMA mode, uses 'ti_size' to read scsi
command into a buffer. Add check to validate command length against
buffer size to avoid any overrun.

Reported-by: Li Qiang <liqiang6-s@360.cn>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-Id: <1464717207-7549-1-git-send-email-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Avoid magic number for NBD max name size
Eric Blake [Wed, 11 May 2016 22:39:44 +0000 (16:39 -0600)]
nbd: Avoid magic number for NBD max name size

Declare a constant and use that when determining if an export
name fits within the constraints we are willing to support.

Note that upstream NBD recently documented that clients MUST
support export names of 256 bytes (not including trailing NUL),
and SHOULD support names up to 4096 bytes.  4096 is a bit big
(we would lose benefits of stack-allocation of a name array),
and we already have other limits in place (for example, qcow2
snapshot names are clamped around 1024).  So for now, just
stick to the required minimum, as that's easier to audit than
a full-scale support for larger names.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-12-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Detect servers that send unexpected error values
Eric Blake [Wed, 11 May 2016 22:39:43 +0000 (16:39 -0600)]
nbd: Detect servers that send unexpected error values

Add some debugging to flag servers that are not compliant to
the NBD protocol.  This would have flagged the server bug
fixed in commit c0301fcc.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alex Bligh <alex@alex.org.uk>
Message-Id: <1463006384-7734-11-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Clean up ioctl handling of qemu-nbd -c
Eric Blake [Wed, 11 May 2016 22:39:40 +0000 (16:39 -0600)]
nbd: Clean up ioctl handling of qemu-nbd -c

The kernel ioctl() interface into NBD is limited to 'unsigned long';
we MUST pass in input with that type (and not int or size_t, as
there may be platform ABIs where the wrong types promote incorrectly
through var-args).  Furthermore, on 32-bit platforms, the kernel
is limited to a maximum export size of 2T (our BLKSIZE of 512 times
a SIZE_BLOCKS constrained by 32 bit unsigned long).

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-8-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Group all Linux-specific ioctl code in one place
Eric Blake [Wed, 11 May 2016 22:39:39 +0000 (16:39 -0600)]
nbd: Group all Linux-specific ioctl code in one place

NBD ioctl()s are used to manage an NBD client session where
initial handshake is done in userspace, but then the transmission
phase is handed off to the kernel through a /dev/nbdX device.
As such, all ioctls sent to the kernel on the /dev/nbdX fd belong
in client.c; nbd_disconnect() was out-of-place in server.c.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-7-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Reject unknown request flags
Eric Blake [Wed, 11 May 2016 22:39:38 +0000 (16:39 -0600)]
nbd: Reject unknown request flags

The NBD protocol says that clients should not send a command flag
that has not been negotiated (whether by the client requesting an
option during a handshake, or because we advertise support for the
flag in response to NBD_OPT_EXPORT_NAME), and that servers should
reject invalid flags with EINVAL.  We were silently ignoring the
flags instead.  The client can't rely on our behavior, since it is
their fault for passing the bad flag in the first place, but it's
better to be robust up front than to possibly behave differently
than the client was expecting with the attempted flag.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alex Bligh <alex@alex.org.uk>
Message-Id: <1463006384-7734-6-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Improve server handling of bogus commands
Eric Blake [Wed, 11 May 2016 22:39:37 +0000 (16:39 -0600)]
nbd: Improve server handling of bogus commands

We have a few bugs in how we handle invalid client commands:

- A client can send an NBD_CMD_DISC where from + len overflows,
convincing us to reply with an error and stay connected, even
though the protocol requires us to silently disconnect. Fix by
hoisting the special case sooner.

- A client can send an NBD_CMD_WRITE where from + len overflows,
where we reply to the client with EINVAL without consuming the
payload; this will normally cause us to fail if the next thing
read is not the right magic, but in rare cases, could cause us
to interpret the data payload as valid commands and do things
not requested by the client. Fix by adding a complete flag to
track whether we are in sync or must disconnect.

Furthermore, we have split the checks for bogus from/len across
two functions, when it is easier to do it all at once.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-5-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Quit server after any write error
Eric Blake [Wed, 11 May 2016 22:39:36 +0000 (16:39 -0600)]
nbd: Quit server after any write error

We should never ignore failure from nbd_negotiate_send_rep(); if
we are unable to write to the client, then it is not worth trying
to continue the negotiation.  Fortunately, the problem is not
too severe - chances are that the errors being ignored here (mainly
inability to write the reply to the client) are indications of
a closed connection or something similar, which will also affect
the next attempt to interact with the client and eventually reach
a point where the errors are detected to end the loop.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-4-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: More debug typo fixes, use correct formats
Eric Blake [Wed, 11 May 2016 22:39:35 +0000 (16:39 -0600)]
nbd: More debug typo fixes, use correct formats

Clean up some debug message oddities missed earlier; this includes
some typos, and recognizing that %d is not necessarily compatible
with uint32_t. Also add a couple messages that I found useful
while debugging things.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-3-git-send-email-eblake@redhat.com>
[Do not use PRIx16, clang complains. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Use BDRV_REQ_FUA for better FUA where supported
Eric Blake [Wed, 11 May 2016 22:39:34 +0000 (16:39 -0600)]
nbd: Use BDRV_REQ_FUA for better FUA where supported

Rather than always flushing ourselves, let the block layer
forward the FUA on to the underlying device - where all
underlying layers also understand FUA, we are now more
efficient; and where any underlying layer doesn't understand
it, now the block layer takes care of the full flush fallback
on our behalf.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463006384-7734-2-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agovl.c: Add '-L help' which lists data dirs.
Richard W.M. Jones [Mon, 16 May 2016 16:34:35 +0000 (17:34 +0100)]
vl.c: Add '-L help' which lists data dirs.

QEMU compiles a list of data directories from various sources.  When
consuming a QEMU binary it's useful to be able to get this list of
data directories: a primary reason is so you can list what BIOSes or
keymaps ship with this version of QEMU.  However without reproducing
the method that QEMU uses internally, it's not possible to get the
list of data directories.

This commit adds a simple '-L help' option that just lists out the
data directories as qemu calculates them:

$ ./x86_64-softmmu/qemu-system-x86_64 -L help
/home/rjones/d/qemu/pc-bios
/usr/local/share/qemu

$ ./x86_64-softmmu/qemu-system-x86_64 -L /tmp -L help
/tmp
/home/rjones/d/qemu/pc-bios
/usr/local/share/qemu

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463416475-11728-2-git-send-email-rjones@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: use KVM_CAP_MAX_VCPU_ID
Greg Kurz [Thu, 26 May 2016 08:02:23 +0000 (10:02 +0200)]
KVM: use KVM_CAP_MAX_VCPU_ID

As stated in linux/Documentation/virtual/kvm/api.txt:

The maximum possible value for max_vcpu_id can be retrieved using the
KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.

If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that
max_vcpu_id is the same as the value returned from KVM_CAP_MAX_VCPUS.

Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Message-Id: <146424974323.5666.5471538288045048119.stgit@bahia.huguette.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoscsi-disk: Use (unsigned long) typecasts when using "%lu" format string
Thomas Huth [Mon, 13 Jun 2016 08:10:18 +0000 (10:10 +0200)]
scsi-disk: Use (unsigned long) typecasts when using "%lu" format string

Some source code analyzers like cppcheck spill out a warning if
the sign of the argument does not match the format string.

Ticket: https://bugs.launchpad.net/qemu/+bug/1589564
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1465805418-15906-1-git-send-email-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agotarget-i386: kvm: cache KVM_GET_SUPPORTED_CPUID data
Chao Peng [Mon, 13 Jun 2016 02:21:27 +0000 (10:21 +0800)]
target-i386: kvm: cache KVM_GET_SUPPORTED_CPUID data

KVM_GET_SUPPORTED_CPUID ioctl is called frequently when initializing
CPU. Depends on CPU features and CPU count, the number of calls can be
extremely high which slows down QEMU booting significantly. In our
testing, we saw 5922 calls with switches:

    -cpu SandyBridge -smp 6,sockets=6,cores=1,threads=1

This ioctl takes more than 100ms, which is almost half of the total
QEMU startup time.

While for most cases the data returned from two different invocations
are not changed, that means, we can cache the data to avoid trapping
into kernel for the second time. To make sure the cache safe one
assumption is desirable: the ioctl is stateless. This is not true for
CPUID leaves in general (such as CPUID leaf 0xD, whose value depends
on guest XCR0 and IA32_XSS) but it is true of KVM_GET_SUPPORTED_CPUID,
which runs before there is a value for XCR0 and IA32_XSS.

Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Message-Id: <1465784487-23482-1-git-send-email-chao.p.peng@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: simplify the nbd_request and nbd_reply structs
Paolo Bonzini [Mon, 13 Jun 2016 09:42:40 +0000 (11:42 +0200)]
nbd: simplify the nbd_request and nbd_reply structs

These structs are never used to represent the bytes that go over the
network.  The big-endian network data is built into a uint8_t array
in nbd_{receive,send}_{request,reply}.  Remove the unused magic field,
reorder the struct to avoid holes, and remove the packed attribute.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Don't use cpu_to_*w() functions
Peter Maydell [Fri, 10 Jun 2016 16:15:42 +0000 (17:15 +0100)]
nbd: Don't use cpu_to_*w() functions

The cpu_to_*w() functions just compose a pointer dereference
with a byteswap. Instead use st*_p(), which handles potential
pointer misalignment and avoids the need to cast the pointer.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <1465575342-12146-1-git-send-email-peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agonbd: Don't use *_to_cpup() functions
Peter Maydell [Fri, 10 Jun 2016 15:00:36 +0000 (16:00 +0100)]
nbd: Don't use *_to_cpup() functions

The *_to_cpup() functions are not very useful, as they simply do
a pointer dereference and then a *_to_cpu(). Instead use either:
 * ld*_*_p(), if the data is at an address that might not be
   correctly aligned for the load
 * a local dereference and *_to_cpu(), if the pointer is
   the correct type and known to be correctly aligned

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <1465570836-22211-1-git-send-email-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoconfigure: Remove unused CONFIG_SIGEV_THREAD_ID switch
Thomas Huth [Fri, 10 Jun 2016 15:04:44 +0000 (17:04 +0200)]
configure: Remove unused CONFIG_SIGEV_THREAD_ID switch

The CONFIG_SIGEV_THREAD_ID switch is unused since the related code
has been removed by commit 6d327171551a12b937c5718073b9848d0274c74d
("aio / timers: Remove alarm timers"), so it can safely be removed
nowadays.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1465571084-19885-1-git-send-email-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoavx2 configure: Use primitives in test
Dr. David Alan Gilbert [Fri, 10 Jun 2016 11:16:18 +0000 (12:16 +0100)]
avx2 configure: Use primitives in test

Use the avx2 primitives during the test, thus making sure that the
compiler and assembler could actually use avx2.

This also detects the failure case on gcc 4.8.x with -save-temps
and avoids the need for the gcc version check in cutils.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1465557378-24105-3-git-send-email-dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMake avx2 configure test work with -O2
Dr. David Alan Gilbert [Fri, 10 Jun 2016 11:16:17 +0000 (12:16 +0100)]
Make avx2 configure test work with -O2

When configured with --extra-cflags=-O2 gcc optimised out the test
and the readelf failed the check leaving avx2 disabled.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1465557378-24105-2-git-send-email-dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMakefile: Fix tag file generation targets
Sergey Fedorov [Thu, 9 Jun 2016 17:58:35 +0000 (20:58 +0300)]
Makefile: Fix tag file generation targets

"ctags" produces a file named "tags", not "ctags". It doesn't look
reasonable to use phony target name as a file name to remove. Just use
exact file names to remove in "ctags" and "TAGS" target receipts.

Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <1465495115-24665-1-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoconfigure: Enable -Werror for MinGW builds, too
Thomas Huth [Wed, 8 Jun 2016 08:13:26 +0000 (10:13 +0200)]
configure: Enable -Werror for MinGW builds, too

MinGW seems to compile currently without warnings, so it should
be safe to enable -Werror now for this environment, too.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1465373606-18486-1-git-send-email-thuth@redhat.com>
Tested-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoclean-includes: run it once more
Paolo Bonzini [Mon, 6 Jun 2016 16:56:37 +0000 (18:56 +0200)]
clean-includes: run it once more

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoos-posix: include sys/mman.h
Paolo Bonzini [Mon, 6 Jun 2016 11:57:39 +0000 (13:57 +0200)]
os-posix: include sys/mman.h

qemu/osdep.h checks whether MAP_ANONYMOUS is defined, but this check
is bogus without a previous inclusion of sys/mman.h.  Include it in
sysemu/os-posix.h and remove it from everywhere else.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoconfigure: Remove unused CONFIG_ZERO_MALLOC setting
Thomas Huth [Wed, 8 Jun 2016 15:11:23 +0000 (17:11 +0200)]
configure: Remove unused CONFIG_ZERO_MALLOC setting

CONFIG_ZERO_MALLOC was only used in qemu-malloc.c and
this file has been removed with the following commit:

41a748265f4879b52b0e87ff9c93bed975163886
Remove qemu_malloc/qemu_free

So we don't need this configuration setting anymore.
This patch also removes the z_version variable, since
this is now also not needed anymore.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Message-Id: <1465398683-3152-1-git-send-email-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Peter Maydell [Thu, 16 Jun 2016 14:22:56 +0000 (15:22 +0100)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer patches

# gpg: Signature made Thu 16 Jun 2016 15:01:27 BST
# gpg:                using RSA key 0x7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>"
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74  56FE 7F09 B272 C88F 2FD6

* remotes/kevin/tags/for-upstream: (39 commits)
  hbitmap: add 'pos < size' asserts
  iotests: Add test for oVirt-like storage migration
  iotests: Add test for post-mirror backing chains
  block/null: Implement bdrv_refresh_filename()
  block/mirror: Fix target backing BDS
  block: Allow replacement of a BDS by its overlay
  rbd:change error_setg() to error_setg_errno()
  iotests: 095: Clean up QEMU before showing image info
  block: Create the commit block job before reopening any image
  block: Prevent sleeping jobs from resuming if they have been paused
  block: use the block job list in qmp_query_block_jobs()
  block: use the block job list in bdrv_drain_all()
  block: Fix snapshot=on with aio=native
  block: Remove bs->zero_beyond_eof
  qcow2: Let vmstate call qcow2_co_preadv/pwrite directly
  block: Make bdrv_load/save_vmstate coroutine_fns
  block: Allow .bdrv_load/save_vmstate() to return 0/-errno
  block: Make .bdrv_load_vmstate() vectored
  block: Introduce bdrv_preadv()
  doc: Fix mailing list address in tests/qemu-iotests/README
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'mreitz/tags/pull-block-for-kevin-2016-06-16' into queue...
Kevin Wolf [Thu, 16 Jun 2016 13:22:18 +0000 (15:22 +0200)]
Merge remote-tracking branch 'mreitz/tags/pull-block-for-kevin-2016-06-16' into queue-block

Block patches

# gpg: Signature made Thu Jun 16 15:21:35 2016 CEST
# gpg:                using RSA key 0x3BB14202E838ACAD
# gpg: Good signature from "Max Reitz <mreitz@redhat.com>"
# Primary key fingerprint: 91BE B60A 30DB 3E88 57D1  1829 F407 DB00 61D5 CF40
#      Subkey fingerprint: 58B3 81CE 2DC8 9CF9 9730  EE64 3BB1 4202 E838 ACAD

* mreitz/tags/pull-block-for-kevin-2016-06-16:
  hbitmap: add 'pos < size' asserts
  iotests: Add test for oVirt-like storage migration
  iotests: Add test for post-mirror backing chains
  block/null: Implement bdrv_refresh_filename()
  block/mirror: Fix target backing BDS
  block: Allow replacement of a BDS by its overlay
  rbd:change error_setg() to error_setg_errno()
  iotests: 095: Clean up QEMU before showing image info
  block: Create the commit block job before reopening any image
  block: Prevent sleeping jobs from resuming if they have been paused
  block: use the block job list in qmp_query_block_jobs()
  block: use the block job list in bdrv_drain_all()

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agohbitmap: add 'pos < size' asserts
Vladimir Sementsov-Ogievskiy [Tue, 14 Jun 2016 17:08:12 +0000 (20:08 +0300)]
hbitmap: add 'pos < size' asserts

For now, fail in hbitmap_set on start + count > size will come from
hbitmap_set
  hb_count_between
    hbitmap_iter_init
      assert(pos < hb->size)

This patch adds such checks to set/get/reset functions of hbitmap.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 1465924093-76875-2-git-send-email-vsementsov@virtuozzo.com
Signed-off-by: Max Reitz <mreitz@redhat.com>