Michael Roth [Sun, 15 Nov 2015 15:46:06 +0000 (09:46 -0600)]
makefile: fix w32 install target for qemu-ga
fafcaf1 added a 'qemu-ga' install target on w32, which can be used
in place of the existing qemu-ga.exe target to also handle dealing
with other components such as DLLs for VSS/fsfreeze and generating
an MSI package if appropriate configure options are present.
As part of that, qemu-ga$(EXESUF) was removed from $TOOLS in favor
of this new qemu-ga target.
The install rule however relies on a direct mapping of the $TOOLS
entry to the actual resulting binary. In the case of w32, qemu-ga
is not identical to qemu-ga$(EXESUF), and the install recipe fails
to find the 'qemu-ga' binary.
Fix this by essentially remapping 'qemu-ga' back to 'qemu-ga.exe'
in the install recipe.
This raises the question of whether or not qemu-ga should continue
to live in TOOLS as opposed to its own special target, but as a
late fix for a regression in 2.5 this commit should be safer, since
we rely on qemu-ga's presence in $TOOLS in several places throughout
Makefile.
qga: fix for default env processing for guest-exec
envp == NULL must be passed inside gspawn() if it was not passed with
the command line. Original code inherits environment from the QGA,
which is wrong.
Peter Maydell [Tue, 17 Nov 2015 12:34:07 +0000 (12:34 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-vnc-20151116-1' into staging
vnc: buffer code improvements, bugfixes.
# gpg: Signature made Mon 16 Nov 2015 17:20:02 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <[email protected]>"
# gpg: aka "Gerd Hoffmann <[email protected]>"
# gpg: aka "Gerd Hoffmann (private) <[email protected]>"
* remotes/kraxel/tags/pull-vnc-20151116-1:
vnc: fix mismerge
buffer: allow a buffer to shrink gracefully
buffer: factor out buffer_adj_size
buffer: factor out buffer_req_size
vnc: recycle empty vs->output buffer
vnc: fix local state init
vnc: only alloc server surface with clients connected
vnc: use vnc_{width,height} in vnc_set_area_dirty
vnc: factor out vnc_update_server_surface
vnc: add vnc_width+vnc_height helpers
vnc: zap dead code
vnc-jobs: move buffer reset, use new buffer move
vnc: kill jobs queue buffer
vnc: attach names to buffers
buffer: add tracing
buffer: add buffer_shrink
buffer: add buffer_move
buffer: add buffer_move_empty
buffer: add buffer_init
buffer: make the Buffer capacity increase in powers of two
Peter Maydell [Tue, 17 Nov 2015 11:33:38 +0000 (11:33 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging
# gpg: Signature made Tue 17 Nov 2015 11:13:05 GMT using RSA key ID 81AB73C8
# gpg: Good signature from "Stefan Hajnoczi <[email protected]>"
# gpg: aka "Stefan Hajnoczi <[email protected]>"
* remotes/stefanha/tags/block-pull-request:
virtio-blk: Fix double completion for werror=stop
block: make 'stats-interval' an array of ints instead of a string
aio-epoll: Fix use-after-free of node
disas/arm: avoid clang shifting negative signed warning
tpm: avoid clang shifting negative signed warning
tests: Ignore recent test binaries
docs: update bitmaps.md
Fam Zheng [Tue, 17 Nov 2015 10:20:11 +0000 (18:20 +0800)]
virtio-blk: Fix double completion for werror=stop
When a request R is absorbed by request M, it is appended to the
"mr_next" queue led by M, and is completed together with the completion
of M, in virtio_blk_rw_complete.
During DMA restart in virtio_blk_dma_restart_bh, requests in s->rq are
parsed and submitted again, possibly with a stale req->mr_next. It could
be a problem if the request merging in virtio_blk_handle_request hasn't
refreshed every mr_next pointer, in which case, virtio_blk_rw_complete
could walk through unexpected requests following the stale pointers.
Fix this by unsetting the pointer in virtio_blk_rw_complete. It is safe
because this req is either completed and freed right away, or it will be
restarted and parsed from scratch out of the vq later.
Stefan Hajnoczi [Tue, 10 Nov 2015 15:57:34 +0000 (15:57 +0000)]
tpm: avoid clang shifting negative signed warning
clang 3.7.0 on x86_64 warns about the following:
hw/tpm/tpm_tis.c:1000:36: warning: shifting a negative signed value is undefined [-Wshift-negative-value]
tis->loc[c].iface_id = TPM_TIS_IFACE_ID_SUPPORTED_FLAGS1_3;
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
hw/tpm/tpm_tis.c:144:10: note: expanded from macro 'TPM_TIS_IFACE_ID_SUPPORTED_FLAGS1_3'
(~0 << 4)/* all of it is don't care */)
~~ ^
Eric Blake [Fri, 13 Nov 2015 03:47:03 +0000 (20:47 -0700)]
tests: Ignore recent test binaries
Commits 6c6f312d and bd797fc1 added new tests (test-blockjob-txn
and test-timed-average, respectively), but did not mark them for
exclusion in .gitignore.
Eric Blake [Thu, 12 Nov 2015 18:50:43 +0000 (11:50 -0700)]
input: Document why x-input-send-event is still experimental
The x-input-send-event command was introduced in 2.2 with mention
that it is experimental, but now that several releases have elapsed
without any changes, it would be nice to document why that was done
and should still remain experimental in 2.5.
Meanwhile, our documentation states that we prefer 'lower-case',
rather than 'CamelCase', for qapi enum values. The InputButton and
InputAxis enums violate this convention. However, because they are
currently used primarily for generating code that is used internally;
and their only exposure through QMP is via the experimental
'x-input-send-event' command, we are free to change their spelling.
Of course, it would be nicer to delay such a change until the same
time we promote the command to non-experimental. Adding
documentation will help us remember to do that rename.
We have plans to tighten the qapi generator to flag instances of
inconsistent use of naming conventions; if that lands first, it
will just need to whitelist these exceptions until the time we
settle on the final interface.
We are not ready (and might never be ready) to declare
introspection stable between releases. Clients written to
control multiple versions of qemu, and desiring to know
whether a particular member is supported for a given
command, must be prepared to locate that member in spite
of qapi changes that may affect the member's location or
type within the overall object, even though such changes
did not break QMP wire back-compatibility.
Peter Maydell [Mon, 16 Nov 2015 12:09:47 +0000 (12:09 +0000)]
Merge remote-tracking branch 'remotes/otubo/tags/pull-seccomp-20151116' into staging
seccomp branch queue
# gpg: Signature made Mon 16 Nov 2015 08:50:28 GMT using RSA key ID 12F8BD2F
# gpg: Good signature from "Eduardo Otubo (Software Engineer @ ProfitBricks) <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 1C96 46B6 E1D1 C38A F2EC 3FDE FD0C FF5B 12F8 BD2F
* remotes/otubo/tags/pull-seccomp-20151116:
seccomp: loosen library version dependency
configure: arm/aarch64: allow enable-seccomp
seccomp: add cacheflush to whitelist
Peter Maydell [Mon, 16 Nov 2015 10:14:33 +0000 (10:14 +0000)]
Merge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging
# gpg: Signature made Fri 13 Nov 2015 20:16:21 GMT using RSA key ID AAFC390E
# gpg: Good signature from "John Snow (John Huston) <[email protected]>"
* remotes/jnsnow/tags/ide-pull-request:
qtest/ahci: use raw format when qemu-img is absent
libqos: add qemu-img presence check
qtest/ahci: always specify image format
ahci/qtest: don't use tcp sockets for migration tests
atapi: Prioritize unknown cmd error over BCL error
atapi: add byte_count_limit helper
dann frazier [Fri, 23 Oct 2015 21:34:22 +0000 (15:34 -0600)]
seccomp: loosen library version dependency
Drop the libseccomp required version back to 2.1.0, restoring the ability
to build w/ --enable-seccomp on Ubuntu 14.04.
Commit 4cc47f8b3cc4f32586ba2f7fce1dc267da774a69 tightened the dependency
on libseccomp from version 2.1.0 to 2.1.1. This broke building on Ubuntu
14.04, the current Ubuntu LTS release. The commit message didn't mention
any specific functional need for 2.1.1, just that it was the most recent
stable version at the time. I reviewed the changes between 2.1.0 and 2.1.1,
but it looks like that update just contained minor fixes and cleanups - no
obvious (to me) new interfaces or critical bug fixes.
Andrew Jones [Wed, 30 Sep 2015 15:59:18 +0000 (11:59 -0400)]
configure: arm/aarch64: allow enable-seccomp
This is a revert of ae6e8ef11e6cb, but with a bit of refactoring,
and also specifically adding arm/aarch64, rather than all
architectures. Currently, libseccomp code appears to also support
mips, ppc, and s390. We could therefore allow qemu to enable
seccomp for those platforms as well, with additional configure
patches, given they're tested and proven to work.
Andrew Jones [Mon, 2 Nov 2015 22:53:26 +0000 (23:53 +0100)]
seccomp: add cacheflush to whitelist
cacheflush is an arm-specific syscall that qemu built for arm
uses. Add it to the whitelist, but only if we're linking with
a recent enough libseccomp.
Guenter Roeck [Thu, 12 Nov 2015 17:54:55 +0000 (09:54 -0800)]
hw/misc: Add support for ADC controller in Xilinx Zynq 7000
Add support for the Xilinx XADC core used in Zynq 7000.
References:
- Zynq-7000 All Programmable SoC Technical Reference Manual
- 7 Series FPGAs and Zynq-7000 All Programmable SoC XADC
Dual 12-Bit 1 MSPS Analog-to-Digital Converter
Tested with Linux using QEMU machine xilinx-zynq-a9 with devicetree
files zynq-zc702.dtb and zynq-zc706.dtb, and kernel configuration
multi_v7_defconfig.
Reviewed-by: Alistair Francis <[email protected]> Signed-off-by: Guenter Roeck <[email protected]>
[ PC changes:
* Changed macro names to match TRM where possible
* Made programmers model macro scheme consistent
* Dropped XADC_ZYNQ_ prefix on local macros
* Fix ALM field width
* Update threshold-comparison interrupts in _update_ints()
* factored out DFIFO pushes into helper. Renamed to "push/pop"
* Changed xadc_reg to 10 bits and added OOB check.
* Reduced scope of MCTL reset to just stop channel coms.
* Added dummy read data to write commands
* Changed _ to - seperators in string names and filenames
* Dropped ------------ in header comment
* Catchall'ed _update_ints() in _write handler.
* Minor whitespace changes.
* Use ZYNQ_XADC_FIFO_DEPTH instead of ARRAY_SIZE()
] Signed-off-by: Peter Crosthwaite <[email protected]> Tested-by: Guenter Roeck <[email protected]> Signed-off-by: Peter Maydell <[email protected]>
# gpg: Signature made Thu 12 Nov 2015 16:56:44 GMT using RSA key ID 5872D723
# gpg: Good signature from "Juan Quintela <[email protected]>"
# gpg: aka "Juan Quintela <[email protected]>"
* remotes/juanquintela/tags/migration/20151112:
migration_init: Fix lock initialisation/make it explicit
migrate-start-postcopy: Improve text
Postcopy: Fix TP!=HP zero case
Finish non-postcopiable iterative devices before package
migration: Make 32bit linux compile with RDMA
migration: print ram_addr_t as RAM_ADDR_FMT not %zx
# gpg: Signature made Thu 12 Nov 2015 15:34:16 GMT using RSA key ID C88F2FD6
# gpg: Good signature from "Kevin Wolf <[email protected]>"
* remotes/kevin/tags/for-upstream: (43 commits)
block: Update copyright of the accounting code
scsi-disk: Account for failed operations
macio: Account for failed operations
ide: Account for failed and invalid operations
atapi: Account for failed and invalid operations
xen_disk: Account for failed and invalid operations
virtio-blk: Account for failed and invalid operations
nvme: Account for failed and invalid operations
iotests: Add test for the block device statistics
block: Use QEMU_CLOCK_VIRTUAL for the accounting code in qtest mode
qemu-io: Account for failed, invalid and flush operations
block: New option to define the intervals for collecting I/O statistics
block: Add average I/O queue depth to BlockDeviceTimedStats
block: Compute minimum, maximum and average I/O latencies
block: Allow configuring whether to account failed and invalid ops
block: Add statistics for failed and invalid I/O operations
block: Add idle_time_ns to BlockDeviceStats
util: Infrastructure for computing recent averages
block: define 'clock_type' for the accounting code
ide: Account for write operations correctly
...
migration_init: Fix lock initialisation/make it explicit
Peter reported a lock error on MacOS after my a82d593b
patch.
migrate_get_current does one-time initialisation of
a bunch of variables.
migrate_init does reinitialisation even on a 2nd
migrate after a cancel.
The problem here was that I'd initialised the mutex
in migrate_get_current, and the memset in migrate_init
corrupted it.
Remove the memset and replace it by explicit initialisation
of fields that need initialising; this also turns out to be simpler
than the old code that had to preserve some fields.
Improve the text in both the qapi-schema and hmp help to point out
you need to set the postcopy-ram capability prior to issuing
migrate-start-postcopy.
Also fix the text of the migrate_start_postcopy error that
deals with capabilities.
Finish non-postcopiable iterative devices before package
Where we have iterable, but non-postcopiable devices (e.g. htab
or block migration), complete them before forming the 'package'
but with the CPUs stopped. This stops them filling up the package.
Sergey Fedorov [Thu, 12 Nov 2015 15:10:43 +0000 (15:10 +0000)]
target-arm: Update PC before calling gen_helper_check_breakpoints()
PC should be updated in the CPU state before calling check_breakpoints()
helper. Otherwise, the helper would not see the correct PC in the CPU
state if it is not at the start of a TB.
Peter Maydell [Thu, 12 Nov 2015 15:15:30 +0000 (15:15 +0000)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
virtio, vhost: fixes for 2.5
This fixes a performance regression with virtio 1,
and makes device stop/start more robust for vhost-user.
virtio devices on pcie bus now have pcie and pm
capability, as required by the PCI Express spec.
migration now works better with virtio 9p.
Signed-off-by: Michael S. Tsirkin <[email protected]>
# gpg: Signature made Thu 12 Nov 2015 14:40:42 GMT using RSA key ID D28D5469
# gpg: Good signature from "Michael S. Tsirkin <[email protected]>"
# gpg: aka "Michael S. Tsirkin <[email protected]>"
* remotes/mst/tags/for_upstream:
virtio-9p: add savem handlers
hw/virtio: Add PCIe capability to virtio devices
vhost: send SET_VRING_ENABLE at start/stop
vhost: rename RESET_DEVICE backto RESET_OWNER
vhost-user: modify SET_LOG_BASE to pass mmap size and offset
virtio-pci: unbreak queue_enable read
virtio-pci: introduce pio notification capability for modern device
virtio-pci: use zero length mmio eventfd for 1.0 notification cap when possible
KVM: add support for any length io eventfd
memory: don't try to adjust endianness for zero length eventfd
virtio-pci: fix 1.0 virtqueue migration
Conflicts:
include/hw/compat.h
[Fixed a trivial merge conflict in compat.h] Signed-off-by: Peter Maydell <[email protected]>
Alberto Garcia [Wed, 28 Oct 2015 15:33:09 +0000 (17:33 +0200)]
block: Use QEMU_CLOCK_VIRTUAL for the accounting code in qtest mode
This patch switches to QEMU_CLOCK_VIRTUAL for the accounting code in
qtest mode, and makes the latency of the operation constant. This way we
can perform tests on the accounting code with reproducible results.
Alberto Garcia [Wed, 28 Oct 2015 15:33:07 +0000 (17:33 +0200)]
block: New option to define the intervals for collecting I/O statistics
The BlockAcctStats structure contains a list of BlockAcctTimedStats.
Each one of these collects statistics about the minimum, maximum and
average latencies of all I/O operations in a certain interval of time.
This patch adds a new "stats-intervals" option that allows defining
these intervals.
Alberto Garcia [Wed, 28 Oct 2015 15:33:04 +0000 (17:33 +0200)]
block: Allow configuring whether to account failed and invalid ops
This patch adds two options, "stats-account-invalid" and
"stats-account-failed", that can be used to decide whether invalid and
failed I/O operations must be used when collecting statistics for
latency and last access time.
Alberto Garcia [Wed, 28 Oct 2015 15:33:02 +0000 (17:33 +0200)]
block: Add idle_time_ns to BlockDeviceStats
This patch adds the new field 'idle_time_ns' to the BlockDeviceStats
structure, indicating the time that has passed since the previous I/O
operation.
It also adds the block_acct_idle_time_ns() call, to ensure that all
references to the clock type used for accounting are in the same
place. This will later allow us to use a different clock for iotests.
Alberto Garcia [Wed, 28 Oct 2015 15:33:01 +0000 (17:33 +0200)]
util: Infrastructure for computing recent averages
This module computes the average of a set of values within a time
window, keeping also track of the minimum and maximum values.
In order to produce more accurate results it works internally by
creating two time windows of the same period, offsetted by half of
that period. Values are accounted on both windows and the data is
always returned from the oldest one.
[Add missing util/replay.o to test-timed-average dependencies to fix the
build.
--Stefan]
Stefan Hajnoczi [Thu, 5 Nov 2015 23:13:20 +0000 (18:13 -0500)]
tests: add BlockJobTxn unit test
The BlockJobTxn unit test verifies that both single jobs and pairs of
jobs behave as a transaction group. Either all jobs complete
successfully or the group is cancelled.
John Snow [Thu, 5 Nov 2015 23:13:19 +0000 (18:13 -0500)]
iotests: 124 - transactional failure test
Use a transaction to request an incremental backup across two drives.
Coerce one of the jobs to fail, and then re-run the transaction.
Verify that no bitmap data was lost due to the partial transaction
failure.
To support the 'err-cancel' QMP argument name it's necessary for
transaction_action() to convert underscores in Python argument names
to hyphens for QMP argument names.
John Snow [Thu, 5 Nov 2015 23:13:18 +0000 (18:13 -0500)]
block: add transactional properties
Add both transactional properties to the QMP transactional interface,
and add the BlockJobTxn that we create as a result of the err-cancel
property to the BlkActionState structure.
John Snow [Thu, 5 Nov 2015 23:13:17 +0000 (18:13 -0500)]
block: Add BlockJobTxn support to backup_run
Allow a BlockJobTxn to be passed into backup_run, which
will allow the job to join a transactional group if present.
Propagate this new parameter outward into new QMP helper
functions in blockdev.c to allow transaction commands to
pass forward their BlockJobTxn object in a forthcoming patch.
Fam Zheng [Thu, 5 Nov 2015 23:13:15 +0000 (18:13 -0500)]
block: Add block job transactions
Sometimes block jobs must execute as a transaction group. Finishing
jobs wait until all other jobs are ready to complete successfully.
Failure or cancellation of one job cancels the other jobs in the group.
Fam Zheng [Thu, 5 Nov 2015 23:13:11 +0000 (18:13 -0500)]
blockjob: Introduce reference count and fix reference to job->bs
Add reference count to block job, meanwhile move the ownership of the
reference to job->bs from the caller (which is released in two
completion callbacks) to the block job itself. It is necessary for
block_job_complete_sync to work, because block job shouldn't live longer
than its bs, as asserted in bdrv_delete.
John Snow [Thu, 5 Nov 2015 23:13:09 +0000 (18:13 -0500)]
block: rename BlkTransactionState and BdrvActionOps
These structures are misnomers, somewhat.
(1) BlockTransactionState is not state for a transaction,
but is rather state for a single transaction action.
Rename it "BlkActionState" to be more accurate.
(2) The BdrvActionOps describes operations for the BlkActionState,
above. This name might imply a 'BdrvAction' or a 'BdrvActionState',
which there isn't.
Rename this to 'BlkActionOps' to match 'BlkActionState'.
Lastly, update the surrounding in-line documentation and comments
to reflect the current nature of how Transactions operate.
This patch changes only comments and names, and should not affect
behavior in any way.
Fam Zheng [Mon, 9 Nov 2015 10:16:54 +0000 (18:16 +0800)]
qed: Implement .bdrv_drain
The "need_check_timer" is used to clear the "NEED_CHECK" flag in the
image header after a grace period once metadata update has finished. In
compliance to the bdrv_drain semantics we should make sure it remains
deleted once .bdrv_drain is called.
We cannot reuse qed_need_check_timer_cb because here it doesn't satisfy
the assertion. Do the "plug" and "flush" calls manually.
Fam Zheng [Mon, 9 Nov 2015 10:16:53 +0000 (18:16 +0800)]
block: Introduce BlockDriver.bdrv_drain callback
Drivers can have internal request sources that generate IO, like the
need_check_timer in QED. Since we want quiesced periods that contain
nested event loops in block layer, we need to have a way to disable such
event sources.
Block drivers must implement the "bdrv_drain" callback if it has any
internal sources that can generate I/O activity, like a timer or a
worker thread (even in a library) that can schedule QEMUBH in an
asynchronous callback.
Update the comments of bdrv_drain and bdrv_drained_begin accordingly.
Like bdrv_requests_pending(), we should consider all the children of bs.
Before, the while loop just works, as bdrv_requests_pending() already
tracks its children; now we mustn't miss the callback, so recurse down
explicitly.
Fam Zheng [Mon, 9 Nov 2015 10:16:51 +0000 (18:16 +0800)]
block: Emulate bdrv_ioctl with bdrv_aio_ioctl and track both
Currently all drivers that support .bdrv_aio_ioctl also implement
.bdrv_ioctl redundantly. To track ioctl requests in block layer it is
easier if we unify the two paths, because we'll need to run it in a
coroutine, as required by tracked_request_begin. While we're at it, use
.bdrv_aio_ioctl plus aio_poll() to emulate bdrv_ioctl().
Fam Zheng [Mon, 9 Nov 2015 10:16:50 +0000 (18:16 +0800)]
block: Add ioctl parameter fields to BlockRequest
The two fields that will be used by ioctl handling code later are added
as union, because it's used exclusively by ioctl code which dosn't need
the four fields in the other struct of the union.
Fam Zheng [Mon, 9 Nov 2015 10:16:49 +0000 (18:16 +0800)]
iscsi: Emulate commands in iscsi_aio_ioctl as iscsi_ioctl
iscsi_ioctl emulates SG_GET_VERSION_NUM and SG_GET_SCSI_ID. Now that
bdrv_ioctl() will be emulated with .bdrv_aio_ioctl, replicate the logic
into iscsi_aio_ioctl to make them consistent.
Fam Zheng [Mon, 9 Nov 2015 10:16:47 +0000 (18:16 +0800)]
block: Track flush requests
Both bdrv_flush and bdrv_aio_flush eventually call bdrv_co_flush, add
tracked_request_begin and tracked_request_end pair in that function so
that all flush requests are now tracked.
Greg Kurz [Thu, 22 Oct 2015 17:38:42 +0000 (19:38 +0200)]
virtio-9p: add savem handlers
We don't support migration of mounted 9p shares. This is handled by a
migration blocker.
One would expect, however, to be able to migrate if the share is unmounted.
Unfortunately virtio-9p-device does not register savevm handlers at all !
Migration succeeds and leaves the guest with a dangling device...
This patch simply registers migration handlers for virtio-9p-device. Whether
migration is possible or not still depends on the migration blocker.
Peter Maydell [Thu, 12 Nov 2015 14:15:32 +0000 (14:15 +0000)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
# gpg: Signature made Thu 12 Nov 2015 08:01:55 GMT using RSA key ID 398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <[email protected]>"
# 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: 215D 46F4 8246 689E C77F 3562 EF04 965B 398D 6211
* remotes/jasowang/tags/net-pull-request:
net: netmap: use error_setg() helpers in place of error_report()
net: netmap: Fix compilation issue
e1000: Introducing backward compatibility command line parameter
e1000: Implementing various counters
e1000: Fixing the packet address filtering procedure
e1000: Fixing the received/transmitted octets' counters
e1000: Fixing the received/transmitted packets' counters
e1000: Trivial implementation of various MAC registers
e1000: Introduced an array to control the access to the MAC registers
e1000: Add support for migrating the entire MAC registers' array
e1000: Cosmetic and alignment fixes
slirp: Fix type casts and format strings in debug code
Victor Kaplansky [Wed, 11 Nov 2015 14:26:02 +0000 (16:26 +0200)]
vhost-user: modify SET_LOG_BASE to pass mmap size and offset
Unlike the kernel, vhost-user application accesses log table by
mmaping it to its user space. This change adds two new fields to
VhostUserMsg payload: mmap_size, and mmap_offset and make QEMU to
pass the to vhost-user application in VHOST_USER_SET_LOG_BASE
request.
Jason Wang [Fri, 6 Nov 2015 08:02:49 +0000 (16:02 +0800)]
virtio-pci: unbreak queue_enable read
Guest always get zero when reading queue_enable. This violates
spec. Fixing this by setting the queue_enable to true during any guest
writing and setting it to zero during reset.
Jason Wang [Fri, 6 Nov 2015 08:02:48 +0000 (16:02 +0800)]
virtio-pci: introduce pio notification capability for modern device
We used to use mmio for notification. This could be slow on some arch
(e.g on x86 without EPT). So this patch introduces pio bar and a pio
notification cap for modern device. This ability is enabled through
property "modern-pio-notify" for virtio pci devices and was disabled
by default. Management can enable when it thinks it was needed.
Benchmarks shows almost no obvious difference compared to legacy
device on machines without ept. Thanks Wenli Quan <[email protected]>
for the benchmarking.
Jason Wang [Fri, 6 Nov 2015 08:02:47 +0000 (16:02 +0800)]
virtio-pci: use zero length mmio eventfd for 1.0 notification cap when possible
We use data match eventfd for 1.0 notification currently. This could
be slow since software decoding is needed for mmio exit. To speed this
up, we can switch to use zero length mmio eventfd for 1.0 notification
since we can examine the queue index directly from the writing
address. KVM kernel module can utilize this by registering it to fast
mmio bus which could be as fast as pio on ept capable machine when
fast mmio is supported by host kernel.
Lots of improvements were seen on a ept capable machine:
Jason Wang [Fri, 6 Nov 2015 08:02:45 +0000 (16:02 +0800)]
memory: don't try to adjust endianness for zero length eventfd
There's no need to adjust endianness for zero length eventfd since the
data wrote was actually ignored by kernel. So skip the adjust in this
case to fix a possible crash when trying to use wildcard mmio eventfd
in ppc.
This will confuse driver if migrating during initialization. Solves
this issue by:
- introduce transport specific callbacks to load and store extra
virtqueue states.
- add a new subsection for virtio to migrate transport specific modern
device state.
- implement pci specific callbacks.
- add a new property for virtio-pci for whether or not to migrate
extra state.
- compat the migration for 2.4 and elder machine types
Peter Maydell [Thu, 12 Nov 2015 13:41:44 +0000 (13:41 +0000)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-next-20151112' into staging
ppc patch queue -2015-11-12
Highlights:
- A number of fixes for MacOS 9 compatibility based on the old MOL
(Mac-On-Linux) code and a GSoC project.
- Cleaner and more general way of handling register access from the
monitor
# gpg: Signature made Thu 12 Nov 2015 04:33:26 GMT using RSA key ID 20D9B392
# gpg: Good signature from "David Gibson <[email protected]>"
# gpg: aka "David Gibson (Red Hat) <[email protected]>"
# gpg: aka "David Gibson (ozlabs.org) <[email protected]>"
# 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-next-20151112:
monitor/target-ppc: Define target_get_monitor_def
cuda.c: add delay to setting of SR_INT bit
cuda.c: fix T2 timer and enable its interrupt
cuda.c: rename get_counter() state variable from s to ti for consistency
cuda.c: refactor get_tb() so that the time can be passed in
cuda.c: add defines for CUDA registers
cuda.c: fix CUDA SR interrupt clearing
cuda.c: implement dummy IIC access commands
cuda.c: implement simple CUDA_GET_6805_ADDR command
cuda.c: fix CUDA_PACKET response packet format
cuda.c: fix CUDA ADB error packet format
PPC: mac99: Always add USB controller
PPC: Fix lswx bounds checks
PPC: Allow Rc bit to be set on mtspr
Some tests may take long to run, move them under g_test_slow()
condition.
The 5s timeout for the "server" test will have to be adjusted to the worst
known time (for the records, it takes ~0.2s on my host). The "pair"
test takes ~1.7, a quickest version could be implemented.
Reorganization of struct NetClientOptions (commit e4ba22b) caused a
compilation failure of the netmap backend. This patch fixes the issue
by properly accessing the union field.
projects / qemu.git / log