sdk/emulator/qemu.git
8 years agoblock: Include hbitmap.h in block.h
Fam Zheng [Tue, 8 Mar 2016 04:44:53 +0000 (12:44 +0800)]
block: Include hbitmap.h in block.h

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 1457412306-18940-3-git-send-email-famz@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
8 years agobackup: Use Bitmap to replace "s->bitmap"
Fam Zheng [Tue, 8 Mar 2016 04:44:52 +0000 (12:44 +0800)]
backup: Use Bitmap to replace "s->bitmap"

"s->bitmap" tracks done sectors, we only check bit states without using any
iterator which HBitmap is good for. Switch to "Bitmap" which is simpler and
more memory efficient.

Meanwhile, rename it to done_bitmap, to reflect the intention.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 1457412306-18940-2-git-send-email-famz@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
8 years agovpc: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
vpc: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovmdk: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
vmdk: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovhdx: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
vhdx: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovdi: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
vdi: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agosheepdog: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
sheepdog: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqed: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
qed: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqcow2: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
qcow2: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqcow: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
qcow: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoparallels: Use BB functions in .bdrv_create()
Kevin Wolf [Tue, 8 Mar 2016 14:57:05 +0000 (15:57 +0100)]
parallels: Use BB functions in .bdrv_create()

All users of the block layers are supposed to go through a BlockBackend.
The .bdrv_create() implementation is one such user, so this patch
converts it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock: Introduce blk_set_allow_write_beyond_eof()
Kevin Wolf [Tue, 8 Mar 2016 15:39:49 +0000 (16:39 +0100)]
block: Introduce blk_set_allow_write_beyond_eof()

We check that the guest can't write beyond the end of its disk, but for
other internal users it can make sense to allow growing a file.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock: Use writeback in .bdrv_create() implementations
Kevin Wolf [Fri, 4 Mar 2016 13:53:50 +0000 (14:53 +0100)]
block: Use writeback in .bdrv_create() implementations

There's no reason to use a writethrough cache mode while creating an
image.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agohmp: Extend drive_del to delete nodes without BB
Kevin Wolf [Tue, 23 Feb 2016 16:50:37 +0000 (17:50 +0100)]
hmp: Extend drive_del to delete nodes without BB

Now that we can use drive_add to create new nodes without a BB, we also
want to be able to delete such nodes again.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agohmp: 'drive_add -n' for creating a node without BB
Kevin Wolf [Tue, 23 Feb 2016 16:33:24 +0000 (17:33 +0100)]
hmp: 'drive_add -n' for creating a node without BB

This patch adds an option to the drive_add HMP command to create only a
BlockDriverState without a BlockBackend on top.

The motivation for this is that libvirt needs to specify options to a
migration target (specifically, detect-zeroes). drive-mirror doesn't
allow specifying options, and the proper way to do this is to create the
target BDS separately with blockdev-add (where you can specify options)
and then use blockdev-mirror to that BDS.

However, libvirt can't use blockdev-add as long as it is still
experimental, and we're expecting that it will still take some time, so
we need to resort to drive_add.

The problem with drive_add is that so far it always created a BB, and
BDSes with a BB can't be used as a mirroring target as long as we don't
support multiple BBs per BDS - and while we're working towards that
goal, it's another thing that will still take some time.

So to achieve the goal, the simplest solution to provide the
functionality now without adding one-off options to the mirror QMP
commands is to extend drive_add to create nodes without BBs.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovmdk: Switch to heap arrays for vmdk_parent_open
Fam Zheng [Tue, 8 Mar 2016 08:24:36 +0000 (16:24 +0800)]
vmdk: Switch to heap arrays for vmdk_parent_open

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovmdk: Switch to heap arrays for vmdk_read_cid
Fam Zheng [Tue, 8 Mar 2016 08:24:35 +0000 (16:24 +0800)]
vmdk: Switch to heap arrays for vmdk_read_cid

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agovmdk: Switch to heap arrays for vmdk_write_cid
Fam Zheng [Tue, 8 Mar 2016 08:24:34 +0000 (16:24 +0800)]
vmdk: Switch to heap arrays for vmdk_write_cid

It is only called once for each opened image, so we can do it the easy
way.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock: Fix cache mode defaults in bds_tree_init()
Kevin Wolf [Mon, 7 Mar 2016 13:23:04 +0000 (14:23 +0100)]
block: Fix cache mode defaults in bds_tree_init()

Without setting explicit defaults in the options, blockdev-add without
an ID ended up defaulting to writethrough. It should be writeback as
documented.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoblock: Fix snapshot=on cache modes
Kevin Wolf [Mon, 7 Mar 2016 12:02:15 +0000 (13:02 +0100)]
block: Fix snapshot=on cache modes

Since commit 91a097e, we end up with a somewhat weird cache mode
configuration with snapshot=on: The commit broke the cache mode
inheritance for the snapshot overlay so that it is opened as
writethrough instead of unsafe now. The following bdrv_append() call to
put it on top of the tree swaps the WCE flag with the snapshot's backing
file (i.e. the originally given file), so what we eventually get is
cache=writeback on the temporary overlay and
cache=writethrough,cache.no-flush=on on the real image file.

This patch changes things so that the temporary overlay gets
cache=unsafe again like it used to, and the real images get whatever the
user specified. This means that cache.direct is now respected even with
snapshot=on, and in the case of committing changes, the final flush is
no longer ignored except explicitly requested by the user.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblockdev: Snapshotting must not open second instance of old top
Kevin Wolf [Wed, 2 Mar 2016 11:16:44 +0000 (12:16 +0100)]
blockdev: Snapshotting must not open second instance of old top

Calling bdrv_img_create() with a size of -1 means that it determines the
size automatically by opening the backing file. However, in the case of
live snapshots, the backing file is already opened and we must avoid
opening the same image twice at the same time. Apart from that, just
getting the size from the already existing BDS is a lot less overhead
than opening a new instance.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
8 years agoquorum: modify vote rules for flush operation
Changlong Xie [Fri, 26 Feb 2016 01:39:02 +0000 (09:39 +0800)]
quorum: modify vote rules for flush operation

Keep flush interface the same logic as quorum read/write, Otherwise in
following scenario, we'll encounter unexpected errors.

Quorum has two children(A, B). A do flush sucessfully, but B flush failed.
This cause the filesystem of guest become read-only with following errors:

end_request: I/O error, dev vda, sector 11159960
Aborting journal on device vda3-8
EXT4-fs error (device vda3): ext4_journal_start_sb:327: Detected abort journal
EXT4-fs (vda3): Remounting filesystem read-only

Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Changlong Xie <xiecl.fnst@cn.fujitsu.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoqmp event: Refactor QUORUM_REPORT_BAD
Changlong Xie [Fri, 26 Feb 2016 01:39:01 +0000 (09:39 +0800)]
qmp event: Refactor QUORUM_REPORT_BAD

Introduce QuorumOpType, and make QUORUM_REPORT_BAD compatible
with it.

Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Changlong Xie <xiecl.fnst@cn.fujitsu.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agodocs: fix invalid node name in qmp event
Changlong Xie [Fri, 26 Feb 2016 01:39:00 +0000 (09:39 +0800)]
docs: fix invalid node name in qmp event

Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Changlong Xie <xiecl.fnst@cn.fujitsu.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock/vpc: add tests for image creation force_size parameter
Jeff Cody [Thu, 25 Feb 2016 17:27:30 +0000 (12:27 -0500)]
block/vpc: add tests for image creation force_size parameter

Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock/vpc: give option to force the current_size field in .bdrv_create
Jeff Cody [Thu, 25 Feb 2016 17:27:29 +0000 (12:27 -0500)]
block/vpc: give option to force the current_size field in .bdrv_create

When QEMU creates a VHD image, it goes by the original spec,
calculating the current_size based on the nearest CHS geometry (with an
exception for disks > 127GB).

Apparently, Azure will only allow images that are sized to the nearest
MB, and the current_size as calculated from CHS cannot guarantee that.

Allow QEMU to create images similar to how Hyper-V creates images, by
setting current_size to the specified virtual disk size.  This
introduces an option, force_size, to be passed to the vpc format during
image creation, e.g.:

    qemu-img convert -f raw -o force_size -O vpc test.img test.vhd

When using the "force_size" option, the creator app field used by
QEMU will be "qem2" instead of "qemu", to indicate the difference.
In light of this, we also add parsing of the "qem2" field during
vpc_open.

Bug reference: https://bugs.launchpad.net/qemu/+bug/1490611

Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock/vpc: tests for auto-detecting VPC and Hyper-V VHD images
Jeff Cody [Thu, 25 Feb 2016 17:27:28 +0000 (12:27 -0500)]
block/vpc: tests for auto-detecting VPC and Hyper-V VHD images

This tests auto-detection, and overrides, of VHD image sizes created
by Virtual PC, Hyper-V, and Disk2vhd.

This adds three sample images:

hyperv2012r2-dynamic.vhd.bz2 - dynamic VHD image created with Hyper-V
virtualpc-dynamic.vhd.bz2    - dynamic VHD image created with Virtual PC
d2v-zerofilled.vhd.bz2       - dynamic VHD image created with Disk2vhd

Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock/vpc: choose size calculation method based on creator_app field
Jeff Cody [Thu, 25 Feb 2016 17:27:27 +0000 (12:27 -0500)]
block/vpc: choose size calculation method based on creator_app field

The VHD file format is used by both Virtual PC, and Hyper-V.  However,
how the virtual disk size is calculated varies between the two.

Virtual PC uses the CHS drive parameters to determine the drive size.
Hyper-V, on the other hand, uses the current_size field in the footer
when determining image size.

This is problematic for a few reasons:

* VHD images from Hyper-V, using CHS calculations, will likely be
  trunctated.

* If we just rely always on current_size, then QEMU may have data
  compatibility issues with Virtual PC (we may write too much data
  into a VHD file to be used by Virtual PC, for instance).

* Existing VHD images created by QEMU have used the CHS calculations,
  except for images exceeding the 127GB limit.  We want to remain
  compatible with our own generated images.

Luckily, the VHD specification defines a 'Creator App' field, that is
used to indicate what software created the VHD file.

This patch does two things:

    1. Uses the 'Creator App' field to help determine how to calculate
       size, and

    2. Adds a VPC format option 'force_size_calc', so that the user can
       override the 'Creator App' auto-detection, in case there exist
       VHD images with unknown or contradictory 'Creator App' entries.

N.B.: We currently use the maximum CHS value as an indication to use the
current_size field.  This patch does not change that, even with the
'force_size_calc' option.

Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoblock/qapi: Include empty drives in query-blockstats
Kevin Wolf [Fri, 26 Feb 2016 20:16:38 +0000 (21:16 +0100)]
block/qapi: Include empty drives in query-blockstats

Since commit 5ec18f8c, query-blockstats didn't return the statistics of
drives without media any more because such drives have only a BB now,
but not a BDS any more.

This patch fixes the regression so that query-blockstats iterates over
BBs by default and empty drives are displayed again.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock/qapi: Factor out bdrv_query_bds_stats()
Kevin Wolf [Fri, 26 Feb 2016 20:03:01 +0000 (21:03 +0100)]
block/qapi: Factor out bdrv_query_bds_stats()

The new functions handles the data that is taken from the
BlockDriverState.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoblock/qapi: Factor out bdrv_query_blk_stats()
Kevin Wolf [Fri, 26 Feb 2016 18:02:30 +0000 (19:02 +0100)]
block/qapi: Factor out bdrv_query_blk_stats()

The new functions handles the data that is taken from the BlockBackend.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
8 years agoqemu-img: eliminate memory leak
Paolo Bonzini [Thu, 25 Feb 2016 22:53:54 +0000 (23:53 +0100)]
qemu-img: eliminate memory leak

Not particularly important since qemu-img exits immediately after
calling img_rebase, but easily fixed.  Coverity says thanks.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.6-7' into...
Peter Maydell [Mon, 14 Mar 2016 13:51:21 +0000 (13:51 +0000)]
Merge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.6-7' into staging

migration:
 - postcopy is no longer experimental
 - fix a use-after-free in postcopy
 - fix a compile warning

# gpg: Signature made Fri 11 Mar 2016 12:29:33 GMT using RSA key ID 854083B6
# gpg: Good signature from "Amit Shah <amit@amitshah.net>"
# gpg:                 aka "Amit Shah <amit@kernel.org>"
# gpg:                 aka "Amit Shah <amitshah@gmx.net>"

* remotes/amit-migration/tags/migration-for-2.6-7:
  postcopy: Remove the x-
  postcopy: listen thread is never joined
  migration: fix use-after-free in loadvm_postcopy_handle_run_bh
  migration: fix warning for source_return_path_thread

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/berrange/tags/pull-io-win32-2016-03-11-1' into...
Peter Maydell [Mon, 14 Mar 2016 11:49:32 +0000 (11:49 +0000)]
Merge remote-tracking branch 'remotes/berrange/tags/pull-io-win32-2016-03-11-1' into staging

Merge I/O fixes for win32

# gpg: Signature made Fri 11 Mar 2016 10:03:20 GMT using RSA key ID 15104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"

* remotes/berrange/tags/pull-io-win32-2016-03-11-1:
  osdep: remove use of socket_error() from all code
  osdep: add wrappers for socket functions
  char: remove qemu_chr_open_socket_fd method
  char: remove socket_try_connect method
  char: remove qemu_chr_finish_socket_connection method
  io: implement socket watch for win32 using WSAEventSelect+select
  io: remove checking of EWOULDBLOCK
  io: use qemu_accept to ensure SOCK_CLOEXEC is set
  io: introduce qio_channel_create_socket_watch
  io: pass HANDLE to g_source_add_poll on Win32
  io: fix copy+paste mistake in socket error message
  io: assert errors before asserting content in I/O test
  io: set correct error object in background reader test thread
  io: wait for incoming client in socket test
  io: bind to socket before creating QIOChannelSocket
  io: initialize sockets in test program
  io: use bind() to check for IPv4/6 availability
  osdep: fix socket_error() to work with Mingw64

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/cohuck/tags/s390x-20160311' into staging
Peter Maydell [Mon, 14 Mar 2016 11:13:11 +0000 (11:13 +0000)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20160311' into staging

CPU hotplug via cpu-add for s390x, cleanup of the s390x machine
compat code and a bugfix in the s390-ccw bios.

# gpg: Signature made Fri 11 Mar 2016 09:48:02 GMT using RSA key ID C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"

* remotes/cohuck/tags/s390x-20160311:
  s390x/cpu: use g_new0
  s390x: Introduce S390MachineClass
  s390x: Introduce machine definition macros
  pc-bios/s390-ccw: fix old bug in ptr increment
  s390x/cpu: Allow hotplug of CPUs
  s390x/cpu: Add error handling to cpu creation
  s390x/cpu: Add CPU property links
  s390x/cpu: Tolerate max_cpus
  s390x/cpu: Get rid of side effects when creating a vcpu
  s390x/cpu: Set initial CPU state in common routine
  s390x/cpu: Cleanup init in preparation for hotplug

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agopostcopy: Remove the x-
Dr. David Alan Gilbert [Fri, 11 Mar 2016 09:53:36 +0000 (09:53 +0000)]
postcopy: Remove the x-

Postcopy seems to have survived a cycle with only a few fixes,
and Jiri has the current libvirt wired up and working
( https://www.redhat.com/archives/libvir-list/2016-March/msg00080.html )
so remove the experimental tag.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457690016-9070-3-git-send-email-dgilbert@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agopostcopy: listen thread is never joined
Dr. David Alan Gilbert [Fri, 11 Mar 2016 09:53:35 +0000 (09:53 +0000)]
postcopy: listen thread is never joined

We don't join the listen thread, it does its own cleanup.
Mark as detached not joinable.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457690016-9070-2-git-send-email-dgilbert@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: fix use-after-free in loadvm_postcopy_handle_run_bh
Denis V. Lunev [Wed, 9 Mar 2016 15:35:08 +0000 (18:35 +0300)]
migration: fix use-after-free in loadvm_postcopy_handle_run_bh

MigrationState is destroyed before we can come into bottom half.

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>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1457537708-8622-1-git-send-email-den@openvz.org>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agomigration: fix warning for source_return_path_thread
Peter Xu [Wed, 9 Mar 2016 06:12:12 +0000 (14:12 +0800)]
migration: fix warning for source_return_path_thread

max_len is not necessary, while it brings a warning during compilation
when specify "-Wstack-usage=1000000". Replacing using sizeof().

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1457503932-31763-1-git-send-email-peterx@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoosdep: remove use of socket_error() from all code
Daniel P. Berrange [Mon, 7 Mar 2016 20:36:03 +0000 (20:36 +0000)]
osdep: remove use of socket_error() from all code

Now that QEMU wraps the Win32 sockets methods to automatically
set errno upon failure, there is no reason for callers to use
the socket_error() method. They can rely on accessing errno
even on Win32. Remove all use of socket_error() from general
code, leaving it as a static method in oslib-win32.c only.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoosdep: add wrappers for socket functions
Daniel P. Berrange [Mon, 7 Mar 2016 20:25:19 +0000 (20:25 +0000)]
osdep: add wrappers for socket functions

The windows socket functions look identical to the normal POSIX
sockets functions, but instead of setting errno, the caller needs
to call WSAGetLastError(). QEMU has tried to deal with this
incompatibility by defining a socket_error() method that callers
must use that abstracts the difference between WSAGetLastError()
and errno.

This approach is somewhat error prone though - many callers of
the sockets functions are just using errno directly because it
is easy to forget the need use a QEMU specific wrapper. It is
not always immediately obvious that a particular function will
in fact call into Windows sockets functions, so the dev may not
even realize they need to use socket_error().

This introduces an alternative approach to portability inspired
by the way GNULIB fixes portability problems. We use a macro to
redefine the original socket function names to refer to a QEMU
wrapper function. The wrapper function calls the original Win32
sockets method and then sets errno from the WSAGetLastError()
value.

Thus all code can simply call the normal POSIX sockets APIs are
have standard errno reporting on error, even on Windows. This
makes the socket_error() method obsolete.

We also bring closesocket & ioctlsocket into this approach. Even
though they are non-standard Win32 names, we can't wrap the normal
close/ioctl methods since there's no reliable way to distinguish
between a file descriptor and HANDLE in Win32.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agochar: remove qemu_chr_open_socket_fd method
Daniel P. Berrange [Wed, 9 Mar 2016 16:49:56 +0000 (16:49 +0000)]
char: remove qemu_chr_open_socket_fd method

The qemu_chr_open_socket_fd method takes care of either doing a
synchronous socket connect, or creating a listener socket. Part
of the work when creating the listener socket is to register a
watch for incoming clients. The caller of qemu_chr_open_socket_fd
may not want this watch created, as it might be doing a synchronous
wait for the first client. Rather than passing yet more parameters
into qemu_chr_open_socket_fd to let it handle this, just remove
the qemu_chr_open_socket_fd method an inline its functionality
into the caller. This allows for a clearer control flow and shorter
code.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agochar: remove socket_try_connect method
Daniel P. Berrange [Wed, 9 Mar 2016 16:45:04 +0000 (16:45 +0000)]
char: remove socket_try_connect method

The qemu_chr_open_socket_fd() method multiplexes three different
actions into one method. The socket_try_connect() method is one
of its callers, but it only ever want one specific action
performed. By inlining that action into socket_try_connect()
we see that there is not in fact any failure scenario, so there
is not even any reason for socket_try_connect to exist. Just
inline the asynchronous connection attempts directly at the
places that need them. This shortens & clarifies the code.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agochar: remove qemu_chr_finish_socket_connection method
Daniel P. Berrange [Wed, 9 Mar 2016 16:42:28 +0000 (16:42 +0000)]
char: remove qemu_chr_finish_socket_connection method

The qemu_chr_finish_socket_connection method is multiplexing two
different actions into one method. Each caller of it though, only
wants one specific action. The code is shorter & clearer if we
thus remove the method and just inline the specific actions
where needed.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: implement socket watch for win32 using WSAEventSelect+select
Paolo Bonzini [Mon, 7 Mar 2016 11:12:36 +0000 (12:12 +0100)]
io: implement socket watch for win32 using WSAEventSelect+select

On Win32 we cannot directly poll on socket handles. Instead we
create a Win32 event object and associate the socket handle with
the event. When the event signals readyness we then have to
use select to determine which events are ready. Creating Win32
events is moderately heavyweight, so we don't want todo it
every time we create a GSource, so this associates a single
event with a QIOChannel.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: remove checking of EWOULDBLOCK
Daniel P. Berrange [Thu, 10 Mar 2016 17:17:33 +0000 (17:17 +0000)]
io: remove checking of EWOULDBLOCK

Since we now canonicalize WSAEWOULDBLOCK into EAGAIN there is
no longer any need to explicitly check EWOULDBLOCK for Win32.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: use qemu_accept to ensure SOCK_CLOEXEC is set
Daniel P. Berrange [Thu, 10 Mar 2016 17:07:27 +0000 (17:07 +0000)]
io: use qemu_accept to ensure SOCK_CLOEXEC is set

The QIOChannelSocket code mistakenly uses the bare accept()
function which does not set SOCK_CLOEXEC.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: introduce qio_channel_create_socket_watch
Paolo Bonzini [Mon, 7 Mar 2016 10:16:39 +0000 (11:16 +0100)]
io: introduce qio_channel_create_socket_watch

Sockets are not in the same namespace as file descriptors on Windows.
As an initial step, introduce separate APIs for file descriptor and
socket watches.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoio: pass HANDLE to g_source_add_poll on Win32
Paolo Bonzini [Mon, 7 Mar 2016 10:13:06 +0000 (11:13 +0100)]
io: pass HANDLE to g_source_add_poll on Win32

Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoio: fix copy+paste mistake in socket error message
Daniel P. Berrange [Tue, 8 Mar 2016 12:06:30 +0000 (12:06 +0000)]
io: fix copy+paste mistake in socket error message

s/write/read/ in the error message reported after
readmsg() fails

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: assert errors before asserting content in I/O test
Daniel P. Berrange [Tue, 8 Mar 2016 15:37:29 +0000 (15:37 +0000)]
io: assert errors before asserting content in I/O test

When checking the results of an I/O operation test, assert that
the error objects are NULL before asserting on the content. This
is found to give more useful indication of the problem when
diagnosing test failures.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: set correct error object in background reader test thread
Daniel P. Berrange [Tue, 8 Mar 2016 12:06:45 +0000 (12:06 +0000)]
io: set correct error object in background reader test thread

The reader thread was accidentally setting the error pointer
intended for the writer thread. If both threads set errors
this would result in QEMU abort'ing due to the error already
being set.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: wait for incoming client in socket test
Daniel P. Berrange [Tue, 8 Mar 2016 15:26:39 +0000 (15:26 +0000)]
io: wait for incoming client in socket test

Exercise the GSource code for server sockets by calling
qio_channel_wait() prior to accepting the incoming client.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: bind to socket before creating QIOChannelSocket
Daniel P. Berrange [Tue, 8 Mar 2016 15:27:30 +0000 (15:27 +0000)]
io: bind to socket before creating QIOChannelSocket

In the QIOChannelSocket test we create a socket file
descriptor and then try to create a QIOChannelSocket.
This works on Linux, but fails on Win32 because it is
not valid to call getsockname() on an unbound socket.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: initialize sockets in test program
Daniel P. Berrange [Tue, 8 Mar 2016 11:44:26 +0000 (11:44 +0000)]
io: initialize sockets in test program

The win32 sockets layer requires that socket_init() is called
otherwise nothing will work.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoio: use bind() to check for IPv4/6 availability
Daniel P. Berrange [Tue, 8 Mar 2016 11:39:36 +0000 (11:39 +0000)]
io: use bind() to check for IPv4/6 availability

Currently the test-io-channel-socket.c test uses getifaddrs
to see if an IPv4/6 address is present on any host NIC, as
a way to determine if IPv4/6 sockets can be used. This is
problematic because getifaddrs is not available on Win32.

Rather than testing indirectly via getifaddrs, just create
a socket and try to bind() to the loopback address instead.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agoosdep: fix socket_error() to work with Mingw64
Daniel P. Berrange [Mon, 7 Mar 2016 11:19:18 +0000 (11:19 +0000)]
osdep: fix socket_error() to work with Mingw64

Historically QEMU has had a socket_error() macro that was
defined to map to WSASocketError(). The os-win32.h header
file would define errno constants that mapped to the
WSA error constants. This worked fine with Mingw32 since
its header files never defined any errno values, nor did
it even provide an errno.h.  So callers of socket_error()
could match on traditional Exxxx constants and it would
all "just work".

With Mingw64 though, things work rather differently. First
there is an errno.h file which defines all the traditional
errno constants you'd expect from a UNIX platform. There
is then a winerror.h which defined the WSA error constants.
Crucially the WSAExxxx errno values in winerror.h do not
match the Exxxx errno values in error.h.

If QEMU had only imported winerror.h it would still work,
but the qemu/osdep.h file unconditionally imports errno.h.
So callers of socket_error() will get now WSAExxxx values
back and compare them to the Exxx constants. This will
always fail silently at runtime.

To solve this QEMU needs to stop assuming the WSAExxxx
constant values match the Exxx constant values. Thus the
socket_error() macro is turned into a small function that
re-maps WSAExxxx values into Exxx.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
8 years agos390x/cpu: use g_new0
Cornelia Huck [Thu, 10 Mar 2016 09:19:46 +0000 (10:19 +0100)]
s390x/cpu: use g_new0

Let's use g_new0 to allocate cpu_states.

Suggested-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x: Introduce S390MachineClass
Janosch Frank [Mon, 7 Mar 2016 08:04:21 +0000 (09:04 +0100)]
s390x: Introduce S390MachineClass

As we now have the new machine definitions, that let us disable/enable
machine options more easily, we need a way to save them and make them
publicly available.

The new s390-virtio-ccw.h header exports the s390 ccw machine state
and class, so they can be easily used in other C files.

Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x: Introduce machine definition macros
Janosch Frank [Thu, 3 Mar 2016 11:48:34 +0000 (12:48 +0100)]
s390x: Introduce machine definition macros

Most of the machine definition code looks the same between different
machine versions. The new DEFINE_CCW_MACHINE macro makes defining a
new machine easier by inserting standard machine version
definitions. This also makes it possible to propagate values between
machine versions.

The patch is inspired by code from hw/ppc/spapr.c

Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agopc-bios/s390-ccw: fix old bug in ptr increment
Eugene (jno) Dvurechenski [Mon, 15 Feb 2016 14:30:25 +0000 (15:30 +0100)]
pc-bios/s390-ccw: fix old bug in ptr increment

We need to increment by the size of the structure, whereas 'ns' is 'uint8_t *'.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Eugene (jno) Dvurechenski <jno@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Allow hotplug of CPUs
Matthew Rosato [Fri, 4 Mar 2016 17:34:35 +0000 (12:34 -0500)]
s390x/cpu: Allow hotplug of CPUs

Implement cpu hotplug routine and add the machine hook.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-8-git-send-email-mjrosato@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Add error handling to cpu creation
Matthew Rosato [Fri, 4 Mar 2016 17:34:34 +0000 (12:34 -0500)]
s390x/cpu: Add error handling to cpu creation

Check for and propogate errors during s390 cpu creation.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-7-git-send-email-mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Add CPU property links
Matthew Rosato [Fri, 4 Mar 2016 17:34:33 +0000 (12:34 -0500)]
s390x/cpu: Add CPU property links

Link each CPUState as property machine/cpu[n] during initialization.
Add a hotplug handler to s390-virtio-ccw machine and set the
state during plug.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-6-git-send-email-mjrosato@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Tolerate max_cpus
Matthew Rosato [Fri, 4 Mar 2016 17:34:32 +0000 (12:34 -0500)]
s390x/cpu: Tolerate max_cpus

Once hotplug is enabled, interrupts may come in for CPUs
with an address > smp_cpus.  Allocate for this and allow
search routines to look beyond smp_cpus.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-5-git-send-email-mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Get rid of side effects when creating a vcpu
Matthew Rosato [Fri, 4 Mar 2016 17:34:31 +0000 (12:34 -0500)]
s390x/cpu: Get rid of side effects when creating a vcpu

In preparation for hotplug, defer some CPU initialization
until the device is actually being realized, including
cpu_exec_init.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-4-git-send-email-mjrosato@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Set initial CPU state in common routine
Matthew Rosato [Fri, 4 Mar 2016 17:34:30 +0000 (12:34 -0500)]
s390x/cpu: Set initial CPU state in common routine

Both initial and hotplugged CPUs need to set the same initial
state.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-3-git-send-email-mjrosato@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/cpu: Cleanup init in preparation for hotplug
Matthew Rosato [Fri, 4 Mar 2016 17:34:29 +0000 (12:34 -0500)]
s390x/cpu: Cleanup init in preparation for hotplug

Ensure a valid cpu_model is set upfront by setting the
default value directly into the MachineState when none is
specified.  This is needed to ensure hotplugged CPUs share
the same cpu_model.

Signed-off-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1457112875-5209-2-git-send-email-mjrosato@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-ui-20160309-1' into staging
Peter Maydell [Thu, 10 Mar 2016 02:51:14 +0000 (02:51 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-ui-20160309-1' into staging

add linux evdev support, vnc and console fixes.

# gpg: Signature made Wed 09 Mar 2016 09:02:47 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-ui-20160309-1:
  ui/console: add escape sequence \e[5, 6n
  input-linux: add switch to enable auto-repeat events
  input-linux: add option to toggle grab on all devices
  input: linux evdev support
  vnc: send cursor when a new client is connecting

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoui/console: add escape sequence \e[5, 6n
Ren Kimura [Tue, 8 Mar 2016 19:51:21 +0000 (04:51 +0900)]
ui/console: add escape sequence \e[5, 6n

Add support of escape sequence "\e[5n" and "\e[6n" to console.
"\e[5n" reports status of console and it always succeed
in virtual console.
"\e[6n" reports now cursor position in console.

Signed-off-by: Ren Kimura <rkx1209dev@gmail.com>
Message-id: 1457466681-7714-2-git-send-email-rkx1209dev@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/thibault/tags/samuel-thibault' into staging
Peter Maydell [Wed, 9 Mar 2016 05:14:55 +0000 (05:14 +0000)]
Merge remote-tracking branch 'remotes/thibault/tags/samuel-thibault' into staging

Add Samuel Thibault as slirp maintainer

# gpg: Signature made Tue 08 Mar 2016 20:43:01 GMT using RSA key ID FB6B2F1D
# gpg: Good signature from "Samuel Thibault <samuel.thibault@gnu.org>"
# gpg:                 aka "Samuel Thibault <sthibault@debian.org>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@inria.fr>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@labri.fr>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@ens-lyon.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: 900C B024 B679 31D4 0F82  304B D017 8C76 7D06 9EE6
#      Subkey fingerprint: F632 74CD C630 0873 CB3D  29D9 E3E5 1CE8 FB6B 2F1D

* remotes/thibault/tags/samuel-thibault:
  MAINTAINERS: Add Samuel Thibault as slirp maintainer

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.6-6' into...
Peter Maydell [Wed, 9 Mar 2016 01:07:16 +0000 (01:07 +0000)]
Merge remote-tracking branch 'remotes/amit-migration/tags/migration-for-2.6-6' into staging

migration:
* add avx2 instruction optimization, speeds up zero-page checking on
  compatible architectures and compilers (gcc 4.9+)
* add additional postcopy stats to 'info migrate' output

# gpg: Signature made Tue 08 Mar 2016 11:29:48 GMT using RSA key ID 854083B6
# gpg: Good signature from "Amit Shah <amit@amitshah.net>"
# gpg:                 aka "Amit Shah <amit@kernel.org>"
# gpg:                 aka "Amit Shah <amitshah@gmx.net>"

* remotes/amit-migration/tags/migration-for-2.6-6:
  cutils: add avx2 instruction optimization
  configure: detect ifunc and avx2 attribute
  Postcopy: Fix sync count in info migrate

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-fw-cfg-20160308-1' into staging
Peter Maydell [Wed, 9 Mar 2016 00:44:43 +0000 (00:44 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-fw-cfg-20160308-1' into staging

acpi: add fw_cfg device node to dsdt

# gpg: Signature made Tue 08 Mar 2016 11:15:42 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-fw-cfg-20160308-1:
  tests: update acpi test data
  fw_cfg: document ACPI device node information
  acpi: arm: add fw_cfg device node to dsdt
  acpi: pc: add fw_cfg device node to dsdt
  pc: fw_cfg: move ioport base constant to pc.h
  fw_cfg: expose control register size in fw_cfg.h

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/amit-virtio-rng/tags/rng-for-2.6-2' into staging
Peter Maydell [Wed, 9 Mar 2016 00:21:17 +0000 (00:21 +0000)]
Merge remote-tracking branch 'remotes/amit-virtio-rng/tags/rng-for-2.6-2' into staging

rng: use simpleq instead of gslist

# gpg: Signature made Tue 08 Mar 2016 10:51:23 GMT using RSA key ID 854083B6
# gpg: Good signature from "Amit Shah <amit@amitshah.net>"
# gpg:                 aka "Amit Shah <amit@kernel.org>"
# gpg:                 aka "Amit Shah <amitshah@gmx.net>"

* remotes/amit-virtio-rng/tags/rng-for-2.6-2:
  rng: switch request queue to QSIMPLEQ

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMAINTAINERS: Add Samuel Thibault as slirp maintainer
Samuel Thibault [Thu, 6 Aug 2015 11:10:14 +0000 (13:10 +0200)]
MAINTAINERS: Add Samuel Thibault as slirp maintainer

Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Acked-by: Jan Kiszka <jan.kiszka@siemens.com>
8 years agocutils: add avx2 instruction optimization
Liang Li [Tue, 8 Mar 2016 05:53:17 +0000 (13:53 +0800)]
cutils: add avx2 instruction optimization

buffer_find_nonzero_offset() is a hot function during live migration.
Now it use SSE2 instructions for optimization. For platform supports
AVX2 instructions, use AVX2 instructions for optimization can help
to improve the performance of buffer_find_nonzero_offset() about 30%
comparing to SSE2.

Live migration can be faster with this optimization, the test result
shows that for an 8GiB RAM idle guest just boots, this patch can help
to shorten the total live migration time about 6%.

This patch use the ifunc mechanism to select the proper function when
running, for platform supports AVX2, execute the AVX2 instructions,
else, execute the original instructions.

Signed-off-by: Liang Li <liang.z.li@intel.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Suggested-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1457416397-26671-3-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoconfigure: detect ifunc and avx2 attribute
Liang Li [Tue, 8 Mar 2016 05:53:16 +0000 (13:53 +0800)]
configure: detect ifunc and avx2 attribute

Detect if the compiler can support the ifun and avx2, if so, set
CONFIG_AVX2_OPT which will be used to turn on the avx2 instruction
optimization.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Liang Li <liang.z.li@intel.com>
Message-Id: <1457416397-26671-2-git-send-email-liang.z.li@intel.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoPostcopy: Fix sync count in info migrate
Dr. David Alan Gilbert [Thu, 25 Feb 2016 10:03:51 +0000 (10:03 +0000)]
Postcopy: Fix sync count in info migrate

I'd missed the sync count off in the postcopy case.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Message-id: 1456394631-18010-1-git-send-email-dgilbert@redhat.com
Message-Id: <1456394631-18010-1-git-send-email-dgilbert@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoinput-linux: add switch to enable auto-repeat events
Gerd Hoffmann [Fri, 4 Mar 2016 10:25:16 +0000 (11:25 +0100)]
input-linux: add switch to enable auto-repeat events

Enable with "-input-linux /dev/input/${device},repeat=on".

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1457087116-4379-4-git-send-email-kraxel@redhat.com

8 years agoinput-linux: add option to toggle grab on all devices
Gerd Hoffmann [Fri, 4 Mar 2016 10:25:15 +0000 (11:25 +0100)]
input-linux: add option to toggle grab on all devices

Maintain a list of all input devices.  Add an option to make grab
work across all devices (so toggling grab on the keybard can switch
over the mouse too).

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1457087116-4379-3-git-send-email-kraxel@redhat.com

8 years agoinput: linux evdev support
Gerd Hoffmann [Fri, 4 Mar 2016 10:25:14 +0000 (11:25 +0100)]
input: linux evdev support

This patch adds support for reading input events directly from linux
evdev devices and forward them to the guest.  Unlike virtio-input-host
which simply passes on all events to the guest without looking at them
this will interpret the events and feed them into the qemu input
subsystem.

Therefore this is limited to what the qemu input subsystem and the
emulated input devices are able to handle.  Also there is no support for
absolute coordinates (tablet/touchscreen).  So we are talking here about
basic mouse and keyboard support.

The advantage is that it'll work without virtio-input drivers in the
guest, the events are delivered to the usual ps/2 or usb input devices
(depending on what the machine happens to have).  And for keyboards
qemu is able to switch the keyboard between guest and host on hotkey.
The hotkey is hard-coded for now (both control keys), initialy the
guest owns the keyboard.

Probably most useful when assigning vga devices with vfio and using a
physical monitor instead of vnc/spice/gtk as guest display.

Usage:  Add '-input-linux /dev/input/event<nr>' to the qemu command
line.  Note that udev has rules which populate /dev/input/by-{id,path}
with static names, which might be more convinient to use.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 1457087116-4379-2-git-send-email-kraxel@redhat.com

8 years agotests: update acpi test data
Gerd Hoffmann [Tue, 8 Mar 2016 10:04:36 +0000 (11:04 +0100)]
tests: update acpi test data

using tests/acpi-test-data/rebuild-expected-aml.sh

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agofw_cfg: document ACPI device node information
Gabriel L. Somlo [Fri, 19 Feb 2016 18:20:29 +0000 (13:20 -0500)]
fw_cfg: document ACPI device node information

Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marc Marí <markmb@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1455906029-25565-6-git-send-email-somlo@cmu.edu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoacpi: arm: add fw_cfg device node to dsdt
Gabriel L. Somlo [Fri, 19 Feb 2016 18:20:28 +0000 (13:20 -0500)]
acpi: arm: add fw_cfg device node to dsdt

Add a fw_cfg device node to the ACPI DSDT. This is mostly
informational, as the authoritative fw_cfg MMIO region(s)
are listed in the Device Tree. However, since we are building
ACPI tables, we might as well be thorough while at it...

Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marc Marí <markmb@redhat.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1455906029-25565-5-git-send-email-somlo@cmu.edu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoacpi: pc: add fw_cfg device node to dsdt
Gabriel L. Somlo [Fri, 19 Feb 2016 18:20:27 +0000 (13:20 -0500)]
acpi: pc: add fw_cfg device node to dsdt

Add a fw_cfg device node to the ACPI DSDT. While the guest-side
firmware can't utilize this information (since it has to access
the hard-coded fw_cfg device to extract ACPI tables to begin with),
having fw_cfg listed in ACPI will help the guest kernel keep a more
accurate inventory of in-use IO port regions.

Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marc Marí <markmb@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1455906029-25565-4-git-send-email-somlo@cmu.edu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agopc: fw_cfg: move ioport base constant to pc.h
Gabriel L. Somlo [Fri, 19 Feb 2016 18:20:26 +0000 (13:20 -0500)]
pc: fw_cfg: move ioport base constant to pc.h

Move BIOS_CFG_IOPORT define from pc.c to pc.h, and rename
it to FW_CFG_IO_BASE.

Cc: Marc Marí <markmb@redhat.com>
Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marc Marí <markmb@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-id: 1455906029-25565-3-git-send-email-somlo@cmu.edu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
Peter Maydell [Tue, 8 Mar 2016 10:25:50 +0000 (10:25 +0000)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging

# gpg: Signature made Tue 08 Mar 2016 07:46:08 GMT using RSA key ID 398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>"
# 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: check packet payload length
  filter-buffer: Add status_changed callback processing
  filter: Add 'status' property for filter object
  rocker: allow user to specify rocker world by property
  rocker: add name field into WorldOps ale let world specify its name
  rocker: return -ENOMEM in case of some world alloc fails
  rocker: forbid to change world type
  net: netmap: probe netmap interface for virtio-net header
  net: simplify net_init_tap_one logic
  MAINTAINERS: Add entries for include/net/ files
  net: filter: correctly remove filter from the list during finalization
  net: ne2000: check ring buffer control registers

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agofw_cfg: expose control register size in fw_cfg.h
Gabriel L. Somlo [Fri, 19 Feb 2016 18:20:25 +0000 (13:20 -0500)]
fw_cfg: expose control register size in fw_cfg.h

Expose the size of the control register (FW_CFG_CTL_SIZE) in fw_cfg.h.
Add comment to fw_cfg_io_realize() pointing out that since the
8-bit data register is always subsumed by the 16-bit control
register in the port I/O case, we use the control register width
as the *total* width of the (classic, non-DMA) port I/O region reserved
for the device.

Cc: Marc Marí <markmb@redhat.com>
Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Marc Marí <markmb@redhat.com>
Message-id: 1455906029-25565-2-git-send-email-somlo@cmu.edu
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agovnc: send cursor when a new client is connecting
Frediano Ziglio [Wed, 2 Mar 2016 14:32:22 +0000 (14:32 +0000)]
vnc: send cursor when a new client is connecting

If you have hardware cursor and you are reconnecting the VNC client
you need to send the cursor. Failing to do so make the cursor invisible
till is changed.

Signed-off-by: Frediano Ziglio <fziglio@redhat.com>
Message-id: 1456929142-14033-1-git-send-email-fziglio@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agonet: check packet payload length
Prasad J Pandit [Wed, 2 Mar 2016 11:59:58 +0000 (17:29 +0530)]
net: check packet payload length

While computing IP checksum, 'net_checksum_calculate' reads
payload length from the packet. It could exceed the given 'data'
buffer size. Add a check to avoid it.

Reported-by: Liu Ling <liuling-it@360.cn>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agofilter-buffer: Add status_changed callback processing
zhanghailiang [Tue, 1 Mar 2016 05:37:03 +0000 (13:37 +0800)]
filter-buffer: Add status_changed callback processing

While the status of filter-buffer changing from 'on' to 'off',
it need to release all the buffered packets, and delete the related
timer, while switch from 'off' to 'on', it need to resume the release
packets timer.

Here, we extract the process of setup timer into a new helper,
which will be used in the new status_changed callback.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Yang Hongyang <hongyang.yang@easystack.cn>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agofilter: Add 'status' property for filter object
zhanghailiang [Tue, 1 Mar 2016 05:37:02 +0000 (13:37 +0800)]
filter: Add 'status' property for filter object

With this property, users can control if this filter is 'on'
or 'off'. The default behavior for filter is 'on'.

For some types of filters, they may need to react to status changing,
So here, we introduced status changing callback/notifier for filter class.

We will skip the disabled ('off') filter when delivering packets in net layer.

Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Yang Hongyang <hongyang.yang@easystack.cn>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agorocker: allow user to specify rocker world by property
Jiri Pirko [Thu, 25 Feb 2016 14:31:58 +0000 (15:31 +0100)]
rocker: allow user to specify rocker world by property

Add property to specify rocker world. All ports will be assigned to this
world.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agorocker: add name field into WorldOps ale let world specify its name
Jiri Pirko [Thu, 25 Feb 2016 14:31:57 +0000 (15:31 +0100)]
rocker: add name field into WorldOps ale let world specify its name

Also use this in world_name getter function.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agorocker: return -ENOMEM in case of some world alloc fails
Jiri Pirko [Thu, 25 Feb 2016 14:31:56 +0000 (15:31 +0100)]
rocker: return -ENOMEM in case of some world alloc fails

Until now, 0 is returned in this error case. Fix it ro return -ENOMEM.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agorocker: forbid to change world type
Jiri Pirko [Thu, 25 Feb 2016 14:31:55 +0000 (15:31 +0100)]
rocker: forbid to change world type

Port to world assignment should be permitted only by qemu user. Driver
should not be able to do it, so forbid that possibility.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agonet: netmap: probe netmap interface for virtio-net header
Vincenzo Maffione [Wed, 24 Feb 2016 10:30:41 +0000 (11:30 +0100)]
net: netmap: probe netmap interface for virtio-net header

Previous implementation of has_ufo, has_vnet_hdr, has_vnet_hdr_len, etc.
did not really probe for virtio-net header support for the netmap
interface attached to the backend. These callbacks were correct for
VALE ports, but incorrect for hardware NICs, pipes, monitors, etc.

This patch fixes the implementation to work properly with all kinds
of netmap ports.

Signed-off-by: Vincenzo Maffione <v.maffione@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agonet: simplify net_init_tap_one logic
Paolo Bonzini [Thu, 25 Feb 2016 23:05:57 +0000 (00:05 +0100)]
net: simplify net_init_tap_one logic

net_init_tap_one receives in vhostfdname a fd name from vhostfd= or
vhostfds=, or NULL if there is no vhostfd=/vhostfds=.  It is simpler
to just check vhostfdname, than it is to check for vhostfd= or
vhostfds=.  This also calms down Coverity, which otherwise thinks
that monitor_fd_param could dereference a NULL vhostfdname.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agoMAINTAINERS: Add entries for include/net/ files
Thomas Huth [Thu, 25 Feb 2016 11:56:24 +0000 (12:56 +0100)]
MAINTAINERS: Add entries for include/net/ files

The include/net/ files correspond to the files in the net/ directory,
thus there should be corresponding entries in the MAINTAINERS file.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
8 years agonet: filter: correctly remove filter from the list during finalization
Jason Wang [Mon, 15 Feb 2016 04:52:34 +0000 (12:52 +0800)]
net: filter: correctly remove filter from the list during finalization

Qemu may crash when we want to add two filters on the same netdev but
the initialization of second fails (e.g missing parameters):

./qemu-system-x86_64 -netdev user,id=un0 \
 -object filter-buffer,id=f0,netdev=un0,interval=10 \
 -object filter-buffer,id=f1,netdev=un0
Segmentation fault (core dumped)

This is because we don't check whether or not the filter was in the
list of netdev. This patch fixes this.

Cc: Yang Hongyang <hongyang.yang@easystack.cn>
Reviewed-by: Yang Hongyang <hongyang.yang@easystack.cn>
Signed-off-by: Jason Wang <jasowang@redhat.com>