qemu.git
4 years agodisplay: virtio-gpu-3d: check virgl capabilities max_size
Prasad J Pandit [Wed, 14 Dec 2016 07:01:56 +0000 (12:31 +0530)] 
display: virtio-gpu-3d: check virgl capabilities max_size

Virtio GPU device while processing 'VIRTIO_GPU_CMD_GET_CAPSET'
command, retrieves the maximum capabilities size to fill in the
response object. It continues to fill in capabilities even if
retrieved 'max_size' is zero(0), thus resulting in OOB access.
Add check to avoid it.

Reported-by: Zhenhao Hong <zhenhaohong@gmail.com>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-id: 20161214070156.23368-1-ppandit@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
4 years agoUpdate version for v2.8.0-rc4 release v2.8.0-rc4
Stefan Hajnoczi [Thu, 15 Dec 2016 07:36:03 +0000 (07:36 +0000)] 
Update version for v2.8.0-rc4 release

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio-pci: Fix cross-version migration with older machines
Maxime Coquelin [Wed, 14 Dec 2016 16:30:35 +0000 (17:30 +0100)] 
virtio-pci: Fix cross-version migration with older machines

This patch fixes a cross-version migration regression introduced
by commit d1b4259f ("virtio-bus: Plug devices after features are
negotiated").

The problem is encountered when host's vhost backend does not support
VIRTIO_F_VERSION_1, and migration is initiated from a v2.7 or prior
machine with virtio-pci modern capabilities enabled to a v2.8 machine.

In this case, modern capabilities get exposed to the guest by the source,
whereas the target will detect version 1 is not supported so will only
expose legacy capabilities.

The problem is fixed by introducing a new "x-ignore-backend-features"
property, which is set in v2.7 and prior compatibility modes. Doing this,
v2.7 machine keeps its broken behaviour (enabling modern while version
is not supported), and newer machines will behave correctly.

Reported-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Message-id: 20161214163035.3297-1-maxime.coquelin@redhat.com
Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Marcel Apfelbaum <marcel@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agopo: add missing translations in de, fr, it, zh
Stefan Hajnoczi [Wed, 14 Dec 2016 14:47:13 +0000 (14:47 +0000)] 
po: add missing translations in de, fr, it, zh

There are missing translations for the new "Copy" menu item.

The following people provided them to me on IRC just in time for the
QEMU 2.8 release:

 * de_DE - Stefan Hajnoczi <stefanha@redhat.com>
 * fr_FR - Laurent Vivier <laurent@vivier.eu>
 * it    - Pino Toscano <ptoscano@redhat.com>
 * zh_CN - Fam Zheng <famz@redhat.com>

[Removed spurious space in zh_CN "Copy" translation that Fam Zheng
pointed out.
--Stefan]

Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20161214144713.11009-1-stefanha@redhat.com
Cc: Fam Zheng <famz@redhat.com>
Cc: Pino Toscano <ptoscano@redhat.com>
Cc: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agoui/gtk: fix "Copy" menu item segfault
Stefan Hajnoczi [Wed, 14 Dec 2016 14:25:18 +0000 (14:25 +0000)] 
ui/gtk: fix "Copy" menu item segfault

The "Copy" menu item copies VTE terminal text to the clipboard.  This
only works with VTE terminals, not with graphics consoles.

Disable the menu item when the current notebook page isn't a VTE
terminal.

This patch fixes a segfault.  Reproducer: Start QEMU and click the Copy
menu item when the guest display is visible.

Reported-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20161214142518.10504-1-stefanha@redhat.com
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agoUpdate language files for QEMU 2.8.0
Stefan Hajnoczi [Tue, 13 Dec 2016 21:49:17 +0000 (21:49 +0000)] 
Update language files for QEMU 2.8.0

Update translation files (change created via 'make -C po update').

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Message-id: 20161213214917.6436-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoUpdate version for v2.8.0-rc3 release v2.8.0-rc3
Stefan Hajnoczi [Tue, 6 Dec 2016 23:26:04 +0000 (23:26 +0000)] 
Update version for v2.8.0-rc3 release

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoqapi: Document introduction of gluster's 'debug' option
Eric Blake [Tue, 6 Dec 2016 18:20:20 +0000 (12:20 -0600)] 
qapi: Document introduction of gluster's 'debug' option

We intentionally renamed 'debug-level' to 'debug' in the QMP
schema for 'blockdev-add' related to gluster, in order to
match the command line (commit 1a417e46).  However, since
'debug-level' was visible in 2.7, that means that we should
document that 'debug' was not available until 2.8.

The change was intentional because 'blockdev-add' itself
underwent incompatible changes (such as commit 0153d2f) for
the same release; our intent is that after 2.8, these
interfaces will now be stable.  [In hindsight, we should have
used the name x-blockdev-add when we first introduced it]

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 20161206182020.25736-1-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoexec.c: Fix breakpoint invalidation race
Peter Maydell [Tue, 6 Dec 2016 18:07:09 +0000 (18:07 +0000)] 
exec.c: Fix breakpoint invalidation race

A bug (1647683) was reported showing a crash when removing
breakpoints.  The reproducer was bisected to 3359baad when tb_flush
was finally made thread safe.  While in MTTCG the locking in
breakpoint_invalidate would have prevented any problems, but
currently tb_lock() is a NOP for system emulation.

The race is between a tb_flush from the gdbstub and the
tb_invalidate_phys_addr() in breakpoint_invalidate().

Ideally we'd have actual locking here; for the moment the
simple fix is to do a full tb_flush() for a bp invalidate,
since that is thread-safe even if no lock is taken.

Reported-by: Julian Brown <julian@codesourcery.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1481047629-7763-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoqapi: Document DEVICE_TRAY_MOVED addition
Eric Blake [Tue, 6 Dec 2016 16:03:45 +0000 (10:03 -0600)] 
qapi: Document DEVICE_TRAY_MOVED addition

Commit 2d76e72 failed to add a versioning tag to 'id'.

I audited all qapi*.json files from v2.7.0 to the current
state of the tree, and didn't find any other additions where
we failed to use a version tag.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20161206160345.22425-1-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agotests/.gitignore: Ignore test-char
Changlong Xie [Mon, 7 Nov 2016 04:59:25 +0000 (12:59 +0800)] 
tests/.gitignore: Ignore test-char

[Lin Ma <lma@suse.com> notes that commit ea3af47d added test for chardev
unit tests, but didn't add the name of generated binary in .gitignore.
--Stefan]

Signed-off-by: Changlong Xie <xiecl.fnst@cn.fujitsu.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1478494765-13233-1-git-send-email-xiecl.fnst@cn.fujitsu.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'kwolf/tags/for-upstream' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 17:35:29 +0000 (17:35 +0000)] 
Merge remote-tracking branch 'kwolf/tags/for-upstream' into staging

Block layer patches for 2.8.0-rc3

# gpg: Signature made Tue 06 Dec 2016 02:44:39 PM GMT
# 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

* kwolf/tags/for-upstream:
  qcow2: Don't strand clusters near 2G intervals during commit

Message-id: 1481037418-10239-1-git-send-email-kwolf@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoqcow2: Don't strand clusters near 2G intervals during commit
Eric Blake [Mon, 5 Dec 2016 15:49:34 +0000 (09:49 -0600)] 
qcow2: Don't strand clusters near 2G intervals during commit

The qcow2_make_empty() function is reached during 'qemu-img commit',
in order to clear out ALL clusters of an image.  However, if the
image cannot use the fast code path (true if the image is format
0.10, or if the image contains a snapshot), the cluster size is
larger than 512, and the image is larger than 2G in size, then our
choice of sector_step causes problems.  Since it is not cluster
aligned, but qcow2_discard_clusters() silently ignores an unaligned
head or tail, we are leaving clusters allocated.

Enhance the testsuite to expose the flaw, and patch the problem by
ensuring our step size is aligned.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoMerge remote-tracking branch 'jasowang/tags/net-pull-request' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 10:24:24 +0000 (10:24 +0000)] 
Merge remote-tracking branch 'jasowang/tags/net-pull-request' into staging

# gpg: Signature made Tue 06 Dec 2016 02:24:23 AM GMT
# gpg:                using RSA key 0xEF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>"
# Primary key fingerprint: 215D 46F4 8246 689E C77F  3562 EF04 965B 398D 6211

* jasowang/tags/net-pull-request:
  fsl_etsec: Fix various small problems in hexdump code
  fsl_etsec: Pad short payloads with zeros
  net: mcf: check receive buffer size register value

Message-id: 1480991552-14360-1-git-send-email-jasowang@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'jtc/tags/block-pull-request' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 09:55:57 +0000 (09:55 +0000)] 
Merge remote-tracking branch 'jtc/tags/block-pull-request' into staging

# gpg: Signature made Mon 05 Dec 2016 09:30:45 PM GMT
# gpg:                using RSA key 0xBDBE7B27C0DE3057
# gpg: Good signature from "Jeffrey Cody <jcody@redhat.com>"
# gpg:                 aka "Jeffrey Cody <jeff@codyprime.org>"
# gpg:                 aka "Jeffrey Cody <codyprime@gmail.com>"
# Primary key fingerprint: 9957 4B4D 3474 90E7 9D98  D624 BDBE 7B27 C0DE 3057

* jtc/tags/block-pull-request:
  qemu-doc: update gluster protocol usage guide
  block/nfs: fix QMP to match debug option
  block/gluster: fix QMP to match debug option

Message-id: 1480973521-28945-1-git-send-email-jcody@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'pm215/tags/pull-target-arm-20161205' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 09:51:41 +0000 (09:51 +0000)] 
Merge remote-tracking branch 'pm215/tags/pull-target-arm-20161205' into staging

target-arm queue:
 * fix gen_load_exclusive handling of ldaxp

# gpg: Signature made Mon 05 Dec 2016 05:57:51 PM GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* pm215/tags/pull-target-arm-20161205:
  target-arm/translate-a64: fix gen_load_exclusive

Message-id: 1480960775-5002-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'armbru/tags/pull-qapi-2016-12-05' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 09:49:51 +0000 (09:49 +0000)] 
Merge remote-tracking branch 'armbru/tags/pull-qapi-2016-12-05' into staging

QAPI patches for 2016-12-05

# gpg: Signature made Mon 05 Dec 2016 04:41:53 PM GMT
# gpg:                using RSA key 0x3870B400EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* armbru/tags/pull-qapi-2016-12-05:
  qapi: add missing colon-ending for section name
  qapi: use one symbol per line
  qapi: fix various symbols mismatch in documentation
  qapi: fix missing symbol @prefix
  qapi: fix schema symbol sections
  qga/schema: fix double-return in doc
  tests: Avoid qobject_from_jsonf("%"PRId64)
  test-qga: Avoid qobject_from_jsonv("%"PRId64)
  qmp-event: Avoid qobject_from_jsonf("%"PRId64)

Message-id: 1480956313-31322-1-git-send-email-armbru@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'kraxel/tags/pull-vga-20161205-1' into staging
Stefan Hajnoczi [Tue, 6 Dec 2016 09:38:39 +0000 (09:38 +0000)] 
Merge remote-tracking branch 'kraxel/tags/pull-vga-20161205-1' into staging

qxl: fix flickering.
cirrus: avoid devision by zero.
virtio-gpu: fix two leaks.

# gpg: Signature made Mon 05 Dec 2016 10:55:45 AM GMT
# gpg:                using RSA key 0x4CB6D8EED3E87138
# 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>"
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* kraxel/tags/pull-vga-20161205-1:
  display: cirrus: check vga bits per pixel(bpp) value
  virtio-gpu: fix memory leak in update_cursor_data_virgl
  virtio-gpu: fix information leak in getting capset info dispatch
  qxl: Only emit QXL_INTERRUPT_CLIENT_MONITORS_CONFIG on config changes

Message-id: 1480935840-3961-1-git-send-email-kraxel@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agofsl_etsec: Fix various small problems in hexdump code
Andrey Smirnov [Mon, 28 Nov 2016 18:13:15 +0000 (10:13 -0800)] 
fsl_etsec: Fix various small problems in hexdump code

Fix various small problems in hexdump code, such as:
    - Reference to non-existing field etsec->nic->nc.name is replaced
    with nc->name

    - Type mismatch warnings

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
5 years agoqemu-doc: update gluster protocol usage guide
Prasanna Kumar Kalever [Wed, 2 Nov 2016 16:50:38 +0000 (22:20 +0530)] 
qemu-doc: update gluster protocol usage guide

Document:
1. The new debug and logfile options with their usages
2. New json format and its usage and
3. update "GlusterFS, Device URL Syntax" section in "Invocation"

Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
5 years agoblock/nfs: fix QMP to match debug option
Prasanna Kumar Kalever [Wed, 2 Nov 2016 16:50:37 +0000 (22:20 +0530)] 
block/nfs: fix QMP to match debug option

The QMP definition of BlockdevOptionsNfs:
{ 'struct': 'BlockdevOptionsNfs',
  'data': { 'server': 'NFSServer',
            'path': 'str',
            '*user': 'int',
            '*group': 'int',
            '*tcp-syn-count': 'int',
            '*readahead-size': 'int',
            '*page-cache-size': 'int',
            '*debug-level': 'int' } }

To make this consistent with other block protocols like gluster, lets
change s/debug-level/debug/

Suggested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
5 years agoblock/gluster: fix QMP to match debug option
Prasanna Kumar Kalever [Wed, 2 Nov 2016 16:50:36 +0000 (22:20 +0530)] 
block/gluster: fix QMP to match debug option

The QMP definition of BlockdevOptionsGluster:
{ 'struct': 'BlockdevOptionsGluster',
  'data': { 'volume': 'str',
            'path': 'str',
            'server': ['GlusterServer'],
            '*debug-level': 'int',
            '*logfile': 'str' } }

But instead of 'debug-level we have exported 'debug' as the option for choosing
debug level of gluster protocol driver.

This patch fix QMP definition BlockdevOptionsGluster
s/debug-level/debug/

Suggested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Prasanna Kumar Kalever <prasanna.kalever@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
5 years agotarget-arm/translate-a64: fix gen_load_exclusive
Alex Bennée [Fri, 2 Dec 2016 17:34:54 +0000 (17:34 +0000)] 
target-arm/translate-a64: fix gen_load_exclusive

While testing rth's latest TCG patches with risu I found ldaxp was
broken. Investigating further I found it was broken by 1dd089d0 when
the cmpxchg atomic work was merged. As part of that change the code
attempted to be clever by doing a single 64 bit load and then shuffle
the data around to set the two 32 bit registers.

As I couldn't quite follow the endian magic I've simply partially
reverted the change to the original code gen_load_exclusive code. This
doesn't affect the cmpxchg functionality as that is all done on in
gen_store_exclusive part which is untouched.

I've also restored the comment that was removed (with a slight tweak
to mention cmpxchg).

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Richard Henderson <rth@twiddle.net>
Message-id: 20161202173454.19179-1-alex.bennee@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agoqapi: add missing colon-ending for section name
Marc-André Lureau [Thu, 17 Nov 2016 15:54:55 +0000 (19:54 +0400)] 
qapi: add missing colon-ending for section name

The documentation parser we are going to add expects a section name to
end with ':', otherwise the comment is treated as free-form text body.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20161117155504.21843-9-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqapi: use one symbol per line
Marc-André Lureau [Thu, 17 Nov 2016 15:54:54 +0000 (19:54 +0400)] 
qapi: use one symbol per line

The documentation parser we are going to add only handles a single
symbol per line.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20161117155504.21843-8-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqapi: fix various symbols mismatch in documentation
Marc-André Lureau [Thu, 17 Nov 2016 15:54:53 +0000 (19:54 +0400)] 
qapi: fix various symbols mismatch in documentation

There are various mismatch:
- invalid symbols
- section and member symbols mismatch
- enum or union values vs 'type'

The documentation parser catches all these cases.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20161117155504.21843-7-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqapi: fix missing symbol @prefix
Marc-André Lureau [Thu, 17 Nov 2016 15:54:52 +0000 (19:54 +0400)] 
qapi: fix missing symbol @prefix

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20161117155504.21843-6-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqapi: fix schema symbol sections
Marc-André Lureau [Thu, 17 Nov 2016 15:54:51 +0000 (19:54 +0400)] 
qapi: fix schema symbol sections

According to docs/qapi-code-gen.txt, there needs to be '##' to start a
and end a symbol section, that's also what the documentation parser
expects.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20161117155504.21843-5-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqga/schema: fix double-return in doc
Marc-André Lureau [Thu, 17 Nov 2016 15:54:49 +0000 (19:54 +0400)] 
qga/schema: fix double-return in doc

guest-get-memory-block-info documentation should have only one
"Returns:".

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20161117155504.21843-3-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agotests: Avoid qobject_from_jsonf("%"PRId64)
Eric Blake [Wed, 23 Nov 2016 17:36:56 +0000 (11:36 -0600)] 
tests: Avoid qobject_from_jsonf("%"PRId64)

The qobject_from_jsonf() function implements a pseudo-printf
language for creating a QObject; however, it is hard-coded to
only parse a subset of formats understood by -Wformat, and is
not a straight synonym to bare printf().  In particular, any
use of an int64_t integer works only if the system's
definition of PRId64 matches what the parser expects; which
works on glibc (%lld or %ld depending on 32- vs. 64-bit) and
mingw (%I64d), but not on Mac OS (%qd).  Rather than enhance
the parser, it is just as easy to force the use of int (where
the value is small enough) or long long instead of int64_t,
which we know always works.

This should cover all remaining testsuite uses of
qobject_from_json[fv]() that were trying to rely on PRId64,
although my proof for that was done by adding in asserts and
checking that 'make check' still passed, where such asserts
are inappropriate during hard freeze.  A later series in 2.9
may remove all dynamic JSON parsing, but that's a bigger task.

Reported by: G 3 <programmingkidx@gmail.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1479922617-4400-4-git-send-email-eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Rename value64 to value_ll]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agotest-qga: Avoid qobject_from_jsonv("%"PRId64)
Eric Blake [Wed, 23 Nov 2016 17:36:55 +0000 (11:36 -0600)] 
test-qga: Avoid qobject_from_jsonv("%"PRId64)

The qobject_from_jsonv() function implements a pseudo-printf
language for creating a QObject; however, it is hard-coded to
only parse a subset of formats understood by -Wformat, and is
not a straight synonym to bare printf().  In particular, any
use of an int64_t integer works only if the system's
definition of PRId64 matches what the parser expects; which
works on glibc (%lld or %ld depending on 32- vs. 64-bit) and
mingw (%I64d), but not on Mac OS (%qd).  Rather than enhance
the parser, it is just as easy to use normal printf() for
this particular conversion, matching what is done elsewhere
in this file [1], which is safe in this instance because the
format does not contain any of the problematic differences
(bare '%' or the '%s' format).

The use of PRId64 for a variable named 'pid' is gross, but it
is a sad reality of the 64-bit mingw environment, which
mistakenly defines pid_t as a 64-bit type even though getpid()
returns 'int' on that platform [2].  Our definition of the
QGA GuestExec type defines 'pid' as a 64-bit entity, and we
can't tighten it to 'int32' unless the mingw header is fixed.
Using 'long long' instead of 'int64_t' just so that we can
stick with qobject_from_jsonv("%lld") instead of printf() is
not any prettier, since we may have later type churn anyways.

[1] see 'git grep -A2 strdup_printf tests/test-qga.c'
[2] https://bugzilla.redhat.com/show_bug.cgi?id=1397787

Reported by: G 3 <programmingkidx@gmail.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1479922617-4400-3-git-send-email-eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoqmp-event: Avoid qobject_from_jsonf("%"PRId64)
Eric Blake [Wed, 23 Nov 2016 17:36:54 +0000 (11:36 -0600)] 
qmp-event: Avoid qobject_from_jsonf("%"PRId64)

The qobject_from_jsonf() function implements a pseudo-printf
language for creating a QObject; however, it is hard-coded to
only parse a subset of formats understood by -Wformat, and is
not a straight synonym to bare printf().  In particular, any
use of an int64_t integer works only if the system's
definition of PRId64 matches what the parser expects; which
works on glibc (%lld or %ld depending on 32- vs. 64-bit) and
mingw (%I64d), but not on Mac OS (%qd).  Rather than enhance
the parser, it is just as easy to use 'long long', which we
know always works.  There are few enough callers of
qobject_from_json[fv]() that it is easy to audit that this is
the only non-testsuite caller that was actually relying on
this particular conversion.

Reported by: G 3 <programmingkidx@gmail.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1479922617-4400-2-git-send-email-eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Cast tv.tv_sec, tv.tv_usec to long long for type correctness]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
5 years agoMerge remote-tracking branch 'yongbok/tags/mips-20161204' into staging
Stefan Hajnoczi [Mon, 5 Dec 2016 10:56:45 +0000 (10:56 +0000)] 
Merge remote-tracking branch 'yongbok/tags/mips-20161204' into staging

MIPS patches 2016-12-04

Changes:
* Fix Loongson instructions
* Fix bad shifts in {dextp|dextpdp}

# gpg: Signature made Sun 04 Dec 2016 01:39:38 AM GMT
# gpg:                using RSA key 0x2238EB86D5F797C2
# gpg: Can't check signature: public key not found

* yongbok/tags/mips-20161204:
  target-mips: fix bad shifts in {dextp|dextpdp}
  target-mips: Fix Loongson multimedia instructions.
  target-mips: Fix Loongson multimedia 'or' instruction.
  target-mips: Fix Loongson pandn instruction.

Message-id: 1480816817-53245-1-git-send-email-yongbok.kim@imgtec.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agodisplay: cirrus: check vga bits per pixel(bpp) value
Prasad J Pandit [Tue, 18 Oct 2016 07:45:17 +0000 (13:15 +0530)] 
display: cirrus: check vga bits per pixel(bpp) value

In Cirrus CLGD 54xx VGA Emulator, if cirrus graphics mode is VGA,
'cirrus_get_bpp' returns zero(0), which could lead to a divide
by zero error in while copying pixel data. The same could occur
via blit pitch values. Add check to avoid it.

Reported-by: Huawei PSIRT <psirt@huawei.com>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-id: 1476776717-24807-1-git-send-email-ppandit@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
5 years agofsl_etsec: Pad short payloads with zeros
Andrey Smirnov [Mon, 28 Nov 2016 18:13:14 +0000 (10:13 -0800)] 
fsl_etsec: Pad short payloads with zeros

Depending on QEMU network setup it is possible for us to receive a
complete Ethernet packet that is less 64 bytes long. One such example is
when QEMU is configured to use a standalone TAP device (not set to be a
part of any bridge) receives and ARP packet. In cases like that we need
to add more than just 4-bytes of CRC padding and ensure that our payload
is at least 60 bytes long, such that, when combined with CRC padding
bytes the resulting size is at least 802.3 minimum MTU bytes
long (64). Failing to do that results in code in etsec_walk_rx_ring()
setting BD_RX_SH which, in turn, makes corresponding Linux driver of
emulated host to reject buffer as a runt packet

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
5 years agonet: mcf: check receive buffer size register value
Prasad J Pandit [Mon, 28 Nov 2016 19:08:39 +0000 (00:38 +0530)] 
net: mcf: check receive buffer size register value

ColdFire Fast Ethernet Controller uses a receive buffer size
register(EMRBR) to hold maximum size of all receive buffers.
It is set by a user before any operation. If it was set to be
zero, ColdFire emulator would go into an infinite loop while
receiving data in mcf_fec_receive. Add check to avoid it.

Reported-by: Wjjzhang <wjjzhang@tencent.com>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
5 years agovirtio-gpu: fix memory leak in update_cursor_data_virgl
Li Qiang [Tue, 1 Nov 2016 11:06:58 +0000 (04:06 -0700)] 
virtio-gpu: fix memory leak in update_cursor_data_virgl

In update_cursor_data_virgl function, if the 'width'/ 'height'
is not equal to current cursor's width/height it will return
without free the 'data' allocated previously. This will lead
a memory leak issue. This patch fix this issue.

Signed-off-by: Li Qiang <liqiang6-s@360.cn>
Message-id: 58187760.41d71c0a.cca75.4cb9@mx.google.com
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
5 years agovirtio-gpu: fix information leak in getting capset info dispatch
Li Qiang [Tue, 1 Nov 2016 09:53:11 +0000 (02:53 -0700)] 
virtio-gpu: fix information leak in getting capset info dispatch

In virgl_cmd_get_capset_info dispatch function, the 'resp' hasn't
been full initialized before writing to the guest. This will leak
the 'resp.padding' and 'resp.hdr.padding' fieds to the guest. This
patch fix this issue.

Signed-off-by: Li Qiang <liqiang6-s@360.cn>
Message-id: 5818661e.0860240a.77264.7a56@mx.google.com
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
5 years agoqxl: Only emit QXL_INTERRUPT_CLIENT_MONITORS_CONFIG on config changes
Christophe Fergeau [Fri, 28 Oct 2016 14:48:40 +0000 (16:48 +0200)] 
qxl: Only emit QXL_INTERRUPT_CLIENT_MONITORS_CONFIG on config changes

Currently if the client keeps sending the same monitor config to
QEMU/spice-server, QEMU will always raise
a QXL_INTERRUPT_CLIENT_MONITORS_CONFIG regardless of whether there was a
change or not.
Guest-side (with fedora 25), the kernel QXL KMS driver will also forward the
event to user-space without checking if there were actual changes.
Next in line are gnome-shell/mutter (on a default f25 install), which
will try to reconfigure everything without checking if there is anything
to do.
Where this gets ugly is that when applying the resolution changes,
gnome-shell/mutter will call drmModeRmFB, drmModeAddFB, and
drmModeSetCrtc, which will cause the primary surface to be destroyed and
recreated by the QXL KMS driver. This in turn will cause the client to
resend a client monitors config message, which will cause QEMU to reemit
an interrupt with an unchanged monitors configuration, ...
This causes https://bugzilla.redhat.com/show_bug.cgi?id=1266484

This commit makes sure that we only emit
QXL_INTERRUPT_CLIENT_MONITORS_CONFIG when there are actual configuration
changes the guest should act on.

Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
Message-id: 20161028144840.18326-1-cfergeau@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
5 years agotarget-mips: fix bad shifts in {dextp|dextpdp}
Yongbok Kim [Wed, 30 Nov 2016 15:25:04 +0000 (15:25 +0000)] 
target-mips: fix bad shifts in {dextp|dextpdp}

Fixed issues in the MIPSDSP64 instructions dextp and dextpdp.
Shifting can go out of 32 bit range.

https://bugs.launchpad.net/qemu/+bug/1631625

Reported-by: Thomas Huth <thuth@redhat.com>
Reported-by: Jia Liu <proljc@gmail.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
5 years agotarget-mips: Fix Loongson multimedia instructions.
Heiher [Thu, 13 Oct 2016 07:10:32 +0000 (15:10 +0800)] 
target-mips: Fix Loongson multimedia instructions.

Needed to emit FPU exception on Loongson multimedia instructions
executing if Status:CU1 is clear. or FPR changes may be missed
on Linux.

Signed-off-by: Heiher <wangr@lemote.com>
Signed-off-by: Fuxin Zhang <zhangfx@lemote.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
5 years agotarget-mips: Fix Loongson multimedia 'or' instruction.
Heiher [Thu, 13 Oct 2016 07:09:39 +0000 (15:09 +0800)] 
target-mips: Fix Loongson multimedia 'or' instruction.

Signed-off-by: Heiher <wangr@lemote.com>
Signed-off-by: Fuxin Zhang <zhangfx@lemote.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
5 years agotarget-mips: Fix Loongson pandn instruction.
Heiher [Fri, 14 Oct 2016 02:46:04 +0000 (10:46 +0800)] 
target-mips: Fix Loongson pandn instruction.

pandn FD, FS, FT
Operation: FD = ((NOT FS) AND FT)

Signed-off-by: Heiher <wangr@lemote.com>
Signed-off-by: Fuxin Zhang <zhangfx@lemote.com>
Reviewed-by: Yongbok Kim <yongbok.kim@imgtec.com>
Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
5 years agoMerge remote-tracking branch 'dgibson/tags/ppc-for-2.8-20161201' into staging
Stefan Hajnoczi [Thu, 1 Dec 2016 13:39:29 +0000 (13:39 +0000)] 
Merge remote-tracking branch 'dgibson/tags/ppc-for-2.8-20161201' into staging

ppc patch queue 2016-12-01

Just a single migration / hotplug fix in this set.  I believe it's
important enough to go in this late in the 2.8 release process.

# gpg: Signature made Thu 01 Dec 2016 04:43:49 AM GMT
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* dgibson/tags/ppc-for-2.8-20161201:
  spapr: fix default DRC state for coldplugged LMBs

Message-id: 20161201044441.14365-1-david@gibson.dropbear.id.au
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'mst/tags/for_upstream' into staging
Stefan Hajnoczi [Thu, 1 Dec 2016 13:38:36 +0000 (13:38 +0000)] 
Merge remote-tracking branch 'mst/tags/for_upstream' into staging

virtio, vhost, pc: fixes

Minor fixes since 2.8.0-rc2.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Wed 30 Nov 2016 02:25:12 AM GMT
# 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

* mst/tags/for_upstream:
  virtio-crypto: fix uninitialized variables
  loader: fix undefined behavior in rom_order_compare()
  loader: fix handling of custom address spaces when adding ROM blobs
  intel_iommu: fix incorrect device invalidate
  spec/vhost-user: fix the VHOST_USER prefix

Message-id: 1480564455-23933-1-git-send-email-mst@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agospapr: fix default DRC state for coldplugged LMBs
Michael Roth [Wed, 30 Nov 2016 23:05:34 +0000 (17:05 -0600)] 
spapr: fix default DRC state for coldplugged LMBs

Currently we set the initial isolation/allocation state for DRCs
associated with coldplugged LMBs to ISOLATED/UNUSABLE,
respectively, under the assumption that the guest will move this
state to UNISOLATED/USABLE.

In fact, this is only the case for LMBs added via hotplug. For
coldplugged LMBs, the guest actually assumes the initial state to
be UNISOLATED/USABLE.

In practice, this only becomes an issue when we attempt to unplug
one of these LMBs, where the guest kernel will issue an
rtas-get-sensor-state call to check that the corresponding DRC is
in an USABLE state before it will release the LMB back to
QEMU. If the returned state is otherwise, the guest will assume no
further action is needed, which bypasses the QEMU-side cleanup that
occurs during the USABLE->UNUSABLE transition. This results in
LMBs and their corresponding pc-dimm devices to stick around
indefinitely.

This patch fixes the issue by manually setting DRCs associated with
cold-plugged LMBs to UNISOLATED/ALLOCATED, but leaving the hotplug
state untouched. As it turns out, this is analogous to the handling
for cold-plugged CPUs in spapr_core_plug().

Cc: qemu-ppc@nongnu.org
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
5 years agovirtio-crypto: fix uninitialized variables
Gonglei [Sat, 26 Nov 2016 03:07:55 +0000 (11:07 +0800)] 
virtio-crypto: fix uninitialized variables

Though crypto_cfg.reserve is an unused field, let me
initialize the structure in order to make coverity happy.

*** CID 1365923:  Uninitialized variables  (UNINIT)
/hw/virtio/virtio-crypto.c: 851 in virtio_crypto_get_config()
845         stl_le_p(&crypto_cfg.mac_algo_h, c->conf.mac_algo_h);
846         stl_le_p(&crypto_cfg.aead_algo, c->conf.aead_algo);
847         stl_le_p(&crypto_cfg.max_cipher_key_len, c->conf.max_cipher_key_len);
848         stl_le_p(&crypto_cfg.max_auth_key_len, c->conf.max_auth_key_len);
849         stq_le_p(&crypto_cfg.max_size, c->conf.max_size);
850
>>>     CID 1365923:  Uninitialized variables  (UNINIT)
>>>     Using uninitialized value "crypto_cfg". Field "crypto_cfg.reserve"
       is uninitialized when calling "memcpy".
      [Note: The source code implementation of the function
       has been overridden by a builtin model.]
851         memcpy(config, &crypto_cfg, c->config_size);
852     }
853

Rported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
5 years agoloader: fix undefined behavior in rom_order_compare()
Laszlo Ersek [Tue, 29 Nov 2016 19:55:33 +0000 (20:55 +0100)] 
loader: fix undefined behavior in rom_order_compare()

According to ISO C99 / N1256 (referenced in HACKING):

> 6.5.8 Relational operators
>
> 4 For the purposes of these operators, a pointer to an object that is
>   not an element of an array behaves the same as a pointer to the first
>   element of an array of length one with the type of the object as its
>   element type.
>
> 5 When two pointers are compared, the result depends on the relative
>   locations in the address space of the objects pointed to. If two
>   pointers to object or incomplete types both point to the same object,
>   or both point one past the last element of the same array object, they
>   compare equal. If the objects pointed to are members of the same
>   aggregate object, pointers to structure members declared later compare
>   greater than pointers to members declared earlier in the structure,
>   and pointers to array elements with larger subscript values compare
>   greater than pointers to elements of the same array with lower
>   subscript values. All pointers to members of the same union object
>   compare equal. If the expression /P/ points to an element of an array
>   object and the expression /Q/ points to the last element of the same
>   array object, the pointer expression /Q+1/ compares greater than /P/.
>   In all other cases, the behavior is undefined.

Our AddressSpace objects are allocated generally individually, and kept in
the "address_spaces" linked list, so we mustn't compare their addresses
with relops.

Convert the pointers subjected to the relop in rom_order_compare() to
"uintptr_t":

> 7.18.1.4 Integer types capable of holding object pointers
>
> 1 [...]
>
>   The following type designates an unsigned integer type with the
>   property that any valid pointer to void can be converted to this type,
>   then converted back to pointer to void, and the result will compare
>   equal to the original pointer:
>
>   /uintptr_t/
>
>   These types are optional.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alistair Francis <alistair.francis@xilinx.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-devel@nongnu.org
Fixes: 3e76099aacb4dae0d37ebf95305369e03d1491e6
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
5 years agoloader: fix handling of custom address spaces when adding ROM blobs
Laszlo Ersek [Tue, 29 Nov 2016 19:55:32 +0000 (20:55 +0100)] 
loader: fix handling of custom address spaces when adding ROM blobs

* Commit 3e76099aacb4 ("loader: Allow a custom AddressSpace when loading
  ROMs") introduced the "Rom.as" field:

  (1) It modified the utility callers of rom_insert() to take "as" as a
      new parameter from *their* callers, and set "rom->as" from that
      parameter. The functions covered were rom_add_file() and
      rom_add_elf_program().

  (2) It also modified rom_insert() itself, to auto-assign
      "&address_space_memory", in case the external caller passed -- and
      the utility caller forwarded -- as=NULL.

  Except, commit 3e76099aacb4 forgot to update the third utility caller of
  rom_insert(), under point (1), namely rom_add_blob().

* Later, commit 5e774eb3bd264 ("loader: Add AddressSpace loading support
  to uImages") added the load_uimage_as() function, and the
  rom_add_blob_fixed_as() function-like macro, with the necessary changes
  elsewhere to propagate the new "as" parameter to rom_add_blob():

    load_uimage_as()
      load_uboot_image()
        rom_add_blob_fixed_as()
          rom_add_blob()

  At this point, the signature (and workings) of rom_add_blob() had been
  broken already, and the rom_add_blob_fixed_as() macro passed its "_as"
  parameter to rom_add_blob() as "callback_opaque". Given that the
  "fw_callback" parameter itself was set to NULL (correctly), this did no
  additional damage (the opaque arg would never be used), but ultimately
  it broke the new functionality of load_uimage_as().

* The load_uimage_as() function would be put to use in one of the later
  patches, commit e481a1f63c93 ("generic-loader: Add a generic loader").

* We can fix this only in a unified patch now. Append "AddressSpace *as"
  to the signature of rom_add_blob(), and handle the new parameter. Pass
  NULL from all current callers, except from rom_add_blob_fixed_as(),
  where "_as" has to be bumped to the proper position.

* Note that rom_add_file() rejects the case when both "mr" and "as" are
  passed in as non-NULL. The action that this is apparently supposed to
  prevent is the

    rom->mr = mr;

  assignment (that's the only place where the "mr" parameter is used in
  rom_add_file()). In rom_add_blob() though, we have no "mr" parameter,
  and the actions done on the fw_cfg branch:

    if (fw_file_name && fw_cfg) {
        if (mc->rom_file_has_mr) {
            data = rom_set_mr(rom, OBJECT(fw_cfg), devpath);
            mr = rom->mr;
        } else {
            data = rom->data;
        }

  reflect those that are performed by rom_add_file() too (with mr==NULL):

    if (rom->fw_file && fw_cfg) {
        if ((!option_rom || mc->option_rom_has_mr) &&
            mc->rom_file_has_mr) {
            data = rom_set_mr(rom, OBJECT(fw_cfg), devpath);
        } else {
            data = rom->data;
        }

  Hence we need no additional restrictions in rom_add_blob().

* Stable is not affected as both problematic commits appeared first in
  v2.8.0-rc0.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alistair Francis <alistair.francis@xilinx.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Michael Walle <michael@walle.cc>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Shannon Zhao <zhaoshenglong@huawei.com>
Cc: qemu-arm@nongnu.org
Cc: qemu-devel@nongnu.org
Fixes: 3e76099aacb4dae0d37ebf95305369e03d1491e6
Fixes: 5e774eb3bd264c76484906f4bd0fb38e00b8090e
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
5 years agointel_iommu: fix incorrect device invalidate
Peter Xu [Tue, 29 Nov 2016 05:43:40 +0000 (13:43 +0800)] 
intel_iommu: fix incorrect device invalidate

"mask" needs to be inverted before use.

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>
5 years agospec/vhost-user: fix the VHOST_USER prefix
Wei Wang [Thu, 24 Nov 2016 03:20:56 +0000 (22:20 -0500)] 
spec/vhost-user: fix the VHOST_USER prefix

Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
5 years agoUpdate version for v2.8.0-rc2 release v2.8.0-rc2
Stefan Hajnoczi [Tue, 29 Nov 2016 22:26:25 +0000 (22:26 +0000)] 
Update version for v2.8.0-rc2 release

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'kwolf/tags/for-upstream' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 17:06:38 +0000 (17:06 +0000)] 
Merge remote-tracking branch 'kwolf/tags/for-upstream' into staging

Block layer patches for 2.8.0-rc2

# gpg: Signature made Tue 29 Nov 2016 03:16:10 PM GMT
# 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

* kwolf/tags/for-upstream:
  docs: Specify that cache-clean-interval is only supported in Linux
  qcow2: Remove stale comment
  qcow2: Allow 'cache-clean-interval' in Linux only
  qcow2: Make qcow2_cache_table_release() work only in Linux

Message-id: 1480436227-2211-1-git-send-email-kwolf@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agorules.mak: Also try -r to build modules
Paolo Bonzini [Tue, 29 Nov 2016 15:37:20 +0000 (16:37 +0100)] 
rules.mak: Also try -r to build modules

Building qemu fails in distributions where gcc enables PIE by default
(e.g. Debian unstable) with:

/usr/bin/ld: -r and -pie may not be used together

You have to use -r instead of -Wl,-r to avoid gcc passing -pie to the linker
when PIE is enabled and a relocatable object is passed.  However, clang
does not know about -r, so try -Wl,-r first.

[This is a fix for commit c96f0ee6a67ca6277366e78ce5d84d5c20dd596f
("rules.mak: Use -r instead of -Wl, -r to fix building when PIE is
default") which mostly worked but broke the ./configure --enable-modules
build with clang.
--Stefan]

Reported-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20161129153720.29747-1-pbonzini@redhat.com
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'fam/tags/for-upstream' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 14:15:14 +0000 (14:15 +0000)] 
Merge remote-tracking branch 'fam/tags/for-upstream' into staging

# gpg: Signature made Tue 29 Nov 2016 10:33:34 AM GMT
# gpg:                using RSA key 0xCA35624C6A9171C6
# gpg: Good signature from "Fam Zheng <famz@redhat.com>"
# Primary key fingerprint: 5003 7CB7 9706 0F76 F021  AD56 CA35 624C 6A91 71C6

* fam/tags/for-upstream:
  hbitmap: Fix shifts of constants by granularity

Message-id: 20161129103438.15955-1-famz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'public/tags/tracing-pull-request' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 10:21:33 +0000 (10:21 +0000)] 
Merge remote-tracking branch 'public/tags/tracing-pull-request' into staging

# gpg: Signature made Tue 29 Nov 2016 10:07:16 AM GMT
# gpg:                using RSA key 0x9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* public/tags/tracing-pull-request:
  configure: fix LTTng UST tracing backend detection

Message-id: 20161129100724.15207-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'bonzini/tags/for-upstream' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 10:05:07 +0000 (10:05 +0000)] 
Merge remote-tracking branch 'bonzini/tags/for-upstream' into staging

Small fixes for rc2.

# gpg: Signature made Mon 28 Nov 2016 03:45:20 PM GMT
# 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

* bonzini/tags/for-upstream:
  rules.mak: Use -r instead of -Wl, -r to fix building when PIE is default
  migration/pcspk: Turn migration of pcspk off for 2.7 and older
  migration/pcspk: Add a property to state if pcspk is migrated
  pci-assign: sync MSI/MSI-X cap and table with PCIDevice
  megasas: clean up and fix request completion/cancellation
  megasas: do not call pci_dma_unmap after having freed the frame once

Message-id: 1480372837-109736-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agohbitmap: Fix shifts of constants by granularity
Max Reitz [Tue, 15 Nov 2016 22:47:32 +0000 (23:47 +0100)] 
hbitmap: Fix shifts of constants by granularity

An hbitmap's granularity may be anything from 0 to 63, so when shifting
constants by its value, they should not be plain ints.

Even having changed the types, hbitmap_serialization_granularity() still
tries to shift 64 to the right by the granularity. This operation is
undefined if the granularity is greater than 57. Adding an assertion is
fine for now, because serializing is done only in tests so far, but this
means that only bitmaps with a granularity below 58 can be serialized
and we should thus add a hbitmap_is_serializable() function later.

One of the two places touched in this patch uses
QEMU_ALIGN_UP(x, 1 << y). We can use ROUND_UP() there, since the second
parameter is obviously a power of two.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20161115224732.1334-1-mreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
5 years agoconfigure: fix LTTng UST tracing backend detection
Francis Deslauriers [Mon, 28 Nov 2016 15:52:17 +0000 (10:52 -0500)] 
configure: fix LTTng UST tracing backend detection

The detection program needs to be linked with -ldl to build succesfully
with recent versions of LTTng-UST.

We also need to add -ldl to the libs required to build the LTTng-UST
backend (lttng_ust_libs).

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Message-id: 1480348337-24271-1-git-send-email-francis.deslauriers@efficios.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'sstabellini/tags/xen-20161128-tag' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 09:28:51 +0000 (09:28 +0000)] 
Merge remote-tracking branch 'sstabellini/tags/xen-20161128-tag' into staging

Xen 2016/11/28

# gpg: Signature made Mon 28 Nov 2016 07:37:33 PM GMT
# gpg:                using RSA key 0x894F8F4870E1AE90
# gpg: Good signature from "Stefano Stabellini <sstabellini@kernel.org>"
# gpg:                 aka "Stefano Stabellini <stefano.stabellini@eu.citrix.com>"
# Primary key fingerprint: D04E 33AB A51F 67BA 07D3  0AEA 894F 8F48 70E1 AE90

* sstabellini/tags/xen-20161128-tag:
  xen: ignore direction in bufioreq handling
  xen: slightly simplify bufioreq handling
  xen: fix quad word bufioreq handling
  xen_disk: split discard input to match internal representation

Message-id: alpine.DEB.2.10.1611281137590.21858@sstabellini-ThinkPad-X260
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'pm215/tags/pull-target-arm-20161128' into staging
Stefan Hajnoczi [Tue, 29 Nov 2016 09:28:24 +0000 (09:28 +0000)] 
Merge remote-tracking branch 'pm215/tags/pull-target-arm-20161128' into staging

target-arm queue:
 * hw/arm/boot: fix crash handling device trees with no /chosen
   or /memory nodes
 * generic-loader: only set PC if a CPU is specified

# gpg: Signature made Mon 28 Nov 2016 01:47:21 PM GMT
# gpg:                using RSA key 0x3C2525ED14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* pm215/tags/pull-target-arm-20161128:
  arm: Create /chosen and /memory devicetree nodes if necessary
  generic-loader: file: Only set a PC if a CPU is specified

Message-id: 1480341071-5367-1-git-send-email-peter.maydell@linaro.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoxen: ignore direction in bufioreq handling
Jan Beulich [Fri, 25 Nov 2016 10:06:58 +0000 (03:06 -0700)] 
xen: ignore direction in bufioreq handling

There's no way to communicate back read data, so only writes can ever
be usefully specified. Ignore the field, paving the road for eventually
re-using the bit for something else in a few (many?) years time.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen: slightly simplify bufioreq handling
Jan Beulich [Fri, 25 Nov 2016 10:06:33 +0000 (03:06 -0700)] 
xen: slightly simplify bufioreq handling

There's no point setting fields always receiving the same value on each
iteration, as handle_ioreq() doesn't alter them anyway. Set state and
count once ahead of the loop, drop the redundant clearing of
data_is_ptr, and avoid the meaningless (because count is 1) setting of
df altogether.

Also avoid doing an unsigned long calculation of size when the field to
be initialized is only 32 bits wide (and the shift value in the range
0...3).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen: fix quad word bufioreq handling
Jan Beulich [Fri, 25 Nov 2016 10:05:57 +0000 (03:05 -0700)] 
xen: fix quad word bufioreq handling

We should not consume the second slot if it didn't get written yet.
Normal writers - i.e. Xen - would not update write_pointer between the
two writes, but the page may get fiddled with by the guest itself, and
we're better off avoiding to enter an infinite loop in that case.

Reported-by: yanghongke <yanghongke@huawei.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Signed-off-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoMerge remote-tracking branch 'ehabkost/tags/x86-pull-request' into staging
Stefan Hajnoczi [Mon, 28 Nov 2016 16:33:41 +0000 (16:33 +0000)] 
Merge remote-tracking branch 'ehabkost/tags/x86-pull-request' into staging

Coverity fix on target-i386

# gpg: Signature made Fri 25 Nov 2016 09:29:11 PM GMT
# gpg:                using RSA key 0x2807936F984DC5A6
# gpg: Good signature from "Eduardo Habkost <ehabkost@redhat.com>"
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF  D1AA 2807 936F 984D C5A6

* ehabkost/tags/x86-pull-request:
  target-i386: Remove unused local_err variable

Message-id: 1480109401-19470-1-git-send-email-ehabkost@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agorules.mak: Use -r instead of -Wl, -r to fix building when PIE is default
Adrian Bunk [Sun, 27 Nov 2016 16:28:17 +0000 (18:28 +0200)] 
rules.mak: Use -r instead of -Wl, -r to fix building when PIE is default

Building qemu fails in distributions where gcc enables PIE by default
(e.g. Debian unstable) with:

/usr/bin/ld: -r and -pie may not be used together

Use -r instead of -Wl,-r to avoid gcc passing -pie to the linker
when PIE is enabled and a relocatable object is passed.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Message-Id: <20161127162817.15144-1-bunk@stusta.de>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agomigration/pcspk: Turn migration of pcspk off for 2.7 and older
Dr. David Alan Gilbert [Mon, 28 Nov 2016 13:32:01 +0000 (13:32 +0000)] 
migration/pcspk: Turn migration of pcspk off for 2.7 and older

To keep backwards migration compatibility allow us to turn pcspk
migration off.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20161128133201.16104-3-dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agomigration/pcspk: Add a property to state if pcspk is migrated
Dr. David Alan Gilbert [Mon, 28 Nov 2016 13:32:00 +0000 (13:32 +0000)] 
migration/pcspk: Add a property to state if pcspk is migrated

Allow us to turn migration of pcspk off for compatibility.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20161128133201.16104-2-dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agopci-assign: sync MSI/MSI-X cap and table with PCIDevice
Peter Xu [Fri, 25 Nov 2016 02:55:22 +0000 (10:55 +0800)] 
pci-assign: sync MSI/MSI-X cap and table with PCIDevice

Since commit e1d4fb2d ("kvm-irqchip: x86: add msi route notify fn"),
kvm_irqchip_add_msi_route() starts to use pci_get_msi_message() to fetch
MSI info. This requires that we setup MSI related fields in PCIDevice.
For most devices, that won't be a problem, as long as we are using
general interfaces like msi_init()/msix_init().

However, for pci-assign devices, MSI/MSI-X is treated differently - PCI
assign devices are maintaining its own MSI table and cap information in
AssignedDevice struct. however that's not synced up with PCIDevice's
fields. That will leads to pci_get_msi_message() failed to find correct
MSI capability, even with an NULL msix_table.

A quick fix is to sync up the two places: both the capability bits and
table address for MSI/MSI-X.

Reported-by: Changlimin <changlimin@h3c.com>
Tested-by: Changlimin <changlimin@h3c.com>
Cc: qemu-stable@nongnu.org
Fixes: e1d4fb2d ("kvm-irqchip: x86: add msi route notify fn")
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1480042522-16551-1-git-send-email-peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agomegasas: clean up and fix request completion/cancellation
Paolo Bonzini [Thu, 10 Nov 2016 15:27:51 +0000 (16:27 +0100)] 
megasas: clean up and fix request completion/cancellation

megasas_command_cancel is a callback; it should report the abort in
the frame, not try another abort!  Compare for instance with
mptsas_request_cancelled.

So extract the common bits for request completion in a new function
megasas_complete_command, call it from both the .complete and .cancel
callbacks, and remove duplicate pieces from the DCMD path.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20161110152751.4267-2-pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agomegasas: do not call pci_dma_unmap after having freed the frame once
Paolo Bonzini [Thu, 15 Sep 2016 22:36:58 +0000 (00:36 +0200)] 
megasas: do not call pci_dma_unmap after having freed the frame once

Commit 8cc4678 ("megasas: remove useless check for cmd->frame", 2016-07-17) was
wrong because I trusted Coverity too much.  It turns out that there _is_ a
path through which cmd->frame can become NULL.  After megasas_handle_frame's
switch (md->frame->header.frame_cmd), megasas_init_firmware can be called.
From there, megasas_reset_frames will call megasas_unmap_frame which resets
cmd->frame = NULL.

However, there is another bug to fix in there, because megasas_unmap_frame
is called again after setting the command status.  In this case QEMU should
not do anything, instead it calls pci_dma_unmap again.  Harmless, but
better fix it.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoarm: Create /chosen and /memory devicetree nodes if necessary
Guenter Roeck [Thu, 17 Nov 2016 01:30:21 +0000 (17:30 -0800)] 
arm: Create /chosen and /memory devicetree nodes if necessary

While customary, the /chosen and /memory devicetree nodes do not have to
exist. Create if necessary. Also create the /memory/device_type property
if needed.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Message-id: 1479346221-18474-1-git-send-email-linux@roeck-us.net
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agogeneric-loader: file: Only set a PC if a CPU is specified
Alistair Francis [Sat, 12 Nov 2016 02:51:20 +0000 (18:51 -0800)] 
generic-loader: file: Only set a PC if a CPU is specified

This patch fixes the generic-loader file loading to only set the program
counter if a CPU is specified. This follows what is written in the
documentation and was always part of the original intention.

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 537bf4d08be7acf7a89b590cff69e19db7f0a6cd.1478908712.git.alistair.francis@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agotarget-i386: Remove unused local_err variable
Eduardo Habkost [Fri, 25 Nov 2016 15:21:53 +0000 (13:21 -0200)] 
target-i386: Remove unused local_err variable

local_err can never be set to non-NULL. Remove the variable.

Detected by Coverity:

    *** CID 1365201:  Possible Control flow issues  (DEADCODE)
    /target-i386/cpu.c: 2050 in x86_cpu_parse_featurestr()
    2044             prop->value = g_strdup(val);
    2045             prop->errp = &error_fatal;
    2046             qdev_prop_register_global(prop);
    2047         }
    2048
    2049         if (local_err) {
    >>>     CID 1365201:  Possible Control flow issues  (DEADCODE)
    >>>     Execution cannot reach this statement: "error_propagate(errp, local...".
    2050             error_propagate(errp, local_err);
    2051         }
    2052     }
    2053
    2054     static void x86_cpu_load_features(X86CPU *cpu, Error **errp);
    2055     static int x86_cpu_filter_features(X86CPU *cpu);

Reported-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1480087313-15102-1-git-send-email-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
5 years agodocs: Specify that cache-clean-interval is only supported in Linux
Alberto Garcia [Fri, 25 Nov 2016 11:27:46 +0000 (13:27 +0200)] 
docs: Specify that cache-clean-interval is only supported in Linux

Make it clear that having Linux is a hard requirement for this
feature.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoqcow2: Remove stale comment
Alberto Garcia [Fri, 25 Nov 2016 11:27:45 +0000 (13:27 +0200)] 
qcow2: Remove stale comment

We haven't been using CONFIG_MADVISE since 02d0e095031b7fda77de8b

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoqcow2: Allow 'cache-clean-interval' in Linux only
Alberto Garcia [Fri, 25 Nov 2016 11:27:44 +0000 (13:27 +0200)] 
qcow2: Allow 'cache-clean-interval' in Linux only

The cache-clean-interval option of qcow2 only works on Linux. However
we allow setting it in other systems regardless of whether it works or
not.

In those systems this option is not simply a no-op: it actually
invalidates perfectly valid cache tables for no good reason without
freeing their memory.

This patch forbids using that option in non-Linux systems.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoqcow2: Make qcow2_cache_table_release() work only in Linux
Alberto Garcia [Fri, 25 Nov 2016 11:27:43 +0000 (13:27 +0200)] 
qcow2: Make qcow2_cache_table_release() work only in Linux

We are using QEMU_MADV_DONTNEED to discard the memory of individual L2
cache tables. The problem with this is that those semantics are
specific to the Linux madvise() system call. Other implementations of
madvise() (including the very Linux implementation of posix_madvise())
don't do that, so we cannot use them for the same purpose.

This patch makes the code Linux-specific and uses madvise() directly
since there's no point in going through qemu_madvise() for this.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoMerge remote-tracking branch 'vivier-m68k/tags/m68k-for-2.8-pull-request' into staging
Stefan Hajnoczi [Fri, 25 Nov 2016 12:03:36 +0000 (12:03 +0000)] 
Merge remote-tracking branch 'vivier-m68k/tags/m68k-for-2.8-pull-request' into staging

# gpg: Signature made Thu 24 Nov 2016 03:25:39 PM GMT
# gpg:                using RSA key 0xF30C38BD3F2FBE3C
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>"
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>"
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>"
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* vivier-m68k/tags/m68k-for-2.8-pull-request:
  target-m68k: fix muluw/mulsw
  target-m68k: Fix cmpa operand size
  target-m68k: fix EXG instruction

Message-id: 1480001287-17515-1-git-send-email-laurent@vivier.eu
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'mcayland/tags/qemu-openbios-signed' into staging
Stefan Hajnoczi [Fri, 25 Nov 2016 10:33:26 +0000 (10:33 +0000)] 
Merge remote-tracking branch 'mcayland/tags/qemu-openbios-signed' into staging

Update OpenBIOS images

 # gpg: Signature made Thu 24 Nov 2016 09:29:40 PM GMT
 # gpg:                using RSA key 0x5BC2C56FAE0F321F
 # gpg: Good signature from "Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>"
 # Primary key fingerprint: CC62 1AB9 8E82 200D 915C  C9C4 5BC2 C56F AE0F 321F

* mcayland/tags/qemu-openbios-signed:
  Update OpenBIOS images to ef8a14e built from submodule.

Message-id: 20161124214109.GB29463@kentang.lan
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoUpdate OpenBIOS images to ef8a14e built from submodule.
Mark Cave-Ayland [Thu, 24 Nov 2016 21:26:00 +0000 (21:26 +0000)] 
Update OpenBIOS images to ef8a14e built from submodule.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
5 years agotarget-m68k: fix muluw/mulsw
Laurent Vivier [Thu, 24 Nov 2016 08:31:32 +0000 (09:31 +0100)] 
target-m68k: fix muluw/mulsw

"The multiplier and multiplicand are both word operands, and the result
is a long-word operand."

So compute flags on a long-word result, not on a word result.

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agoMerge remote-tracking branch 'gkurz/tags/for-upstream' into staging
Stefan Hajnoczi [Thu, 24 Nov 2016 10:16:23 +0000 (10:16 +0000)] 
Merge remote-tracking branch 'gkurz/tags/for-upstream' into staging

This pull request fixes some leaks (memory, fd) in the handle and proxy
backends.

 # gpg: Signature made Wed 23 Nov 2016 12:53:41 PM GMT
 # gpg:                using DSA key 0x02FC3AEB0101DBC2
 # gpg: Good signature from "Greg Kurz <groug@kaod.org>"
 # gpg:                 aka "Greg Kurz <groug@free.fr>"
 # gpg:                 aka "Greg Kurz <gkurz@fr.ibm.com>"
 # gpg:                 aka "Greg Kurz <gkurz@linux.vnet.ibm.com>"
 # gpg:                 aka "Gregory Kurz (Groug) <groug@free.fr>"
 # gpg:                 aka "Gregory Kurz (Cimai Technology) <gkurz@cimai.com>"
 # gpg:                 aka "Gregory Kurz (Meiosys Technology) <gkurz@meiosys.com>"
 # 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: 2BD4 3B44 535E C0A7 9894  DBA2 02FC 3AEB 0101 DBC2

* gkurz/tags/for-upstream:
  9pfs: add cleanup operation for proxy backend driver
  9pfs: add cleanup operation for handle backend driver
  9pfs: add cleanup operation in FileOperations
  9pfs: adjust the order of resource cleanup in device unrealize

Message-id: 1479920298-24983-1-git-send-email-groug@kaod.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'rth/tags/pull-axp-20161123' into staging
Stefan Hajnoczi [Thu, 24 Nov 2016 09:50:32 +0000 (09:50 +0000)] 
Merge remote-tracking branch 'rth/tags/pull-axp-20161123' into staging

Fix alpha smp interrupt masking

# gpg: Signature made Wed 23 Nov 2016 12:42:45 PM GMT
# gpg:                using RSA key 0xAD1270CC4DD0279B
# gpg: Good signature from "Richard Henderson <rth7680@gmail.com>"
# gpg:                 aka "Richard Henderson <rth@redhat.com>"
# gpg:                 aka "Richard Henderson <rth@twiddle.net>"
# Primary key fingerprint: 9CB1 8DDA F8E8 49AD 2AFC  16A4 AD12 70CC 4DD0 279B

* rth/tags/pull-axp-20161123:
  target-alpha: Fix interrupt mask for cpu1

Message-id: 1479905195-7424-1-git-send-email-rth@twiddle.net
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agotarget-m68k: Fix cmpa operand size
Laurent Vivier [Wed, 23 Nov 2016 20:45:56 +0000 (21:45 +0100)] 
target-m68k: Fix cmpa operand size

"The size of the operation can be specified as word or long.
Word length source operands are sign-extended to 32 bits for
comparison."

So comparison is always done using OS_LONG.

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agotarget-m68k: fix EXG instruction
Laurent Vivier [Wed, 23 Nov 2016 12:57:56 +0000 (13:57 +0100)] 
target-m68k: fix EXG instruction

opcodes of "EXG Ax,Ay" and "EXG Dx,Dy" have been swapped

Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agoxen_disk: split discard input to match internal representation
Olaf Hering [Wed, 23 Nov 2016 10:39:12 +0000 (10:39 +0000)] 
xen_disk: split discard input to match internal representation

The guest sends discard requests as u64 sector/count pairs, but the
block layer operates internally with s64/s32 pairs. The conversion
leads to IO errors in the guest, the discard request is not processed.

  domU.cfg:
  'vdev=xvda, format=qcow2, backendtype=qdisk, target=/x.qcow2'
  domU:
  mkfs.ext4 -F /dev/xvda
  Discarding device blocks: failed - Input/output error

Fix this by splitting the request into chunks of BDRV_REQUEST_MAX_SECTORS.
Add input range checking to avoid overflow.

Fixes f313520 ("xen_disk: add discard support")

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years ago9pfs: add cleanup operation for proxy backend driver
Li Qiang [Wed, 23 Nov 2016 12:53:34 +0000 (13:53 +0100)] 
9pfs: add cleanup operation for proxy backend driver

In the init operation of proxy backend dirver, it allocates a
V9fsProxy struct and some other resources. We should free these
resources when the 9pfs device is unrealized. This is what this
patch does.

Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
5 years ago9pfs: add cleanup operation for handle backend driver
Li Qiang [Wed, 23 Nov 2016 12:53:34 +0000 (13:53 +0100)] 
9pfs: add cleanup operation for handle backend driver

In the init operation of handle backend dirver, it allocates a
handle_data struct and opens a mount file. We should free these
resources when the 9pfs device is unrealized. This is what this
patch does.

Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
5 years ago9pfs: add cleanup operation in FileOperations
Li Qiang [Wed, 23 Nov 2016 12:53:34 +0000 (13:53 +0100)] 
9pfs: add cleanup operation in FileOperations

Currently, the backend of VirtFS doesn't have a cleanup
function. This will lead resource leak issues if the backed
driver allocates resources. This patch addresses this issue.

Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
5 years ago9pfs: adjust the order of resource cleanup in device unrealize
Li Qiang [Wed, 23 Nov 2016 12:53:34 +0000 (13:53 +0100)] 
9pfs: adjust the order of resource cleanup in device unrealize

Unrealize should undo things that were set during realize in
reverse order. So should do in the error path in realize.

Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
5 years agoMerge remote-tracking branch 'dgibson/tags/ppc-for-2.8-20161123' into staging
Stefan Hajnoczi [Wed, 23 Nov 2016 11:46:15 +0000 (11:46 +0000)] 
Merge remote-tracking branch 'dgibson/tags/ppc-for-2.8-20161123' into staging

ppc patch queue 2016-11-23

Here's the first set of 2.8 hard freeze bugfixes for ppc.

The biggest thing here is a batch of fixes for migration breakages in
both 2.7 and current 2.8.  Alas, there is at least one more migration
problem, which prevents memory unplug after a migration.  I hoped to
include a fix for that here, but it turned out to have some problems
bigger than those it was solving.  So, I expect at least one more hard
freeze pull request.

There are also a few other assorted bug fixes.

# gpg: Signature made Wed 23 Nov 2016 02:25:42 AM GMT
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* dgibson/tags/ppc-for-2.8-20161123:
  spapr: Fix 2.7<->2.8 migration of PCI host bridge
  Revert "spapr: Fix migration of PCI host bridges from qemu-2.7"
  target-ppc: Allow eventual removal of old migration mistakes
  migration: Add VMSTATE_UINTTL_TEST()
  target-ppc: Fix CPU migration from qemu-2.6 <-> later versions
  ppc: Make uninorth interrupt swizzling identical to Grackle
  target-ppc: fix index array of national digits
  hw/char/spapr_vty: Return amount of free buffer entries in vty_can_receive()
  ppc: BOOK3E: nothing should be done when MSR:PR is set
  spapr: migration support for CAS-negotiated option vectors
  tests/postcopy: Use KVM on ppc64 only if it is KVM-HV

Message-id: 1479869383-16162-1-git-send-email-david@gibson.dropbear.id.au
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agoMerge remote-tracking branch 'bonzini/tags/for-upstream' into staging
Stefan Hajnoczi [Wed, 23 Nov 2016 11:44:29 +0000 (11:44 +0000)] 
Merge remote-tracking branch 'bonzini/tags/for-upstream' into staging

Small fixes for rc1.

# gpg: Signature made Tue 22 Nov 2016 10:26:56 PM GMT
# 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

* bonzini/tags/for-upstream:
  scsi/esp: do not raise an interrupt when reading the FIFO register
  nbd: Allow unmap and fua during write zeroes
  cpu_ldst.h: use correct guest address parameter

Message-id: 1479853676-35995-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
5 years agospapr: Fix 2.7<->2.8 migration of PCI host bridge
David Gibson [Tue, 22 Nov 2016 23:26:38 +0000 (10:26 +1100)] 
spapr: Fix 2.7<->2.8 migration of PCI host bridge

daa2369 "spapr_pci: Add a 64-bit MMIO window" subtly broke migration
from qemu-2.7 to the current version.  It split the device's MMIO
window into two pieces for 32-bit and 64-bit MMIO.

The patch included backwards compatibility code to convert the old
property into the new format.  However, the property value was also
transferred in the migration stream and compared with a (probably
unwise) VMSTATE_EQUAL.  So, the "raw" value from 2.7 is compared to
the new style converted value from (pre-)2.8 giving a mismatch and
migration failure.

Along with the actual field that caused the breakage, there are
several other ill-advised VMSTATE_EQUAL()s.  To fix forwards
migration, we read the values in the stream into scratch variables and
ignore them, instead of comparing for equality.  To fix backwards
migration, we populate those scratch variables in pre_save() with
adjusted values to match the old behaviour.

To permit the eventual possibility of removing this cruft from the
stream, we only include these compatibility fields if a new
'pre-2.8-migration' property is set.  We clear it on the pseries-2.8
machine type, which obviously can't be migrated backwards, but set it
on earlier machine type versions.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
5 years agoRevert "spapr: Fix migration of PCI host bridges from qemu-2.7"
David Gibson [Mon, 21 Nov 2016 01:12:10 +0000 (12:12 +1100)] 
Revert "spapr: Fix migration of PCI host bridges from qemu-2.7"

This reverts commit 9b54ca0ba781012eeea4237b7c4832ba2ea81d89.

The commit above corrected a migration breakage between qemu-2.7 and
qemu-2.8.  However it did so by advancing the migration version for
the PCI host bridge, which obviously breaks migration backwards to
earlier qemu versions.

Although it's not totally essential, we'd like to maintain the
possibility for backwards migration, so revert the change in
preparation for a better fix.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
5 years agotarget-ppc: Allow eventual removal of old migration mistakes
David Gibson [Mon, 21 Nov 2016 05:29:30 +0000 (16:29 +1100)] 
target-ppc: Allow eventual removal of old migration mistakes

Until very recently, the vmstate for ppc cpus included some poorly
thought out VMSTATE_EQUAL() components, that can easily break
migration compatibility, and did so between qemu-2.6 and later
versions.  A hack was recently added which fixes this migration
breakage, but it leaves the unhelpful cruft of these fields in the
migration stream.

This patch adds a new cpu property allowing these fields to be removed
from the stream entirely.  For the pseries-2.8 machine type - which
comes after the fix - and for all non-pseries machine types - which
aren't mature enough to care about cross-version migration - we remove
the fields from the stream.

For pseries-2.7 and earlier, The migration hack remains in place,
allowing backwards and forwards migration with the older machine
types.

This restricts the migration compatibility cruft to older machine
types, and at least opens the possibility of eventually deprecating
and removing it entirely.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
5 years agomigration: Add VMSTATE_UINTTL_TEST()
David Gibson [Sun, 20 Nov 2016 23:50:02 +0000 (10:50 +1100)] 
migration: Add VMSTATE_UINTTL_TEST()

include/migration/cpu.h defines VMSTATE_UINTTL() and several variants
for migrating target_ulong fields.  It's defined in terms of
VMSTATE_UINT32() or VMSTATE_UINT64() as appropriate.

It doesn't, however, include a VMSTATE_UINTTL_TEST() variant, which
I'm going to need shortly.  So, add it.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
5 years agotarget-ppc: Fix CPU migration from qemu-2.6 <-> later versions
David Gibson [Mon, 21 Nov 2016 05:28:12 +0000 (16:28 +1100)] 
target-ppc: Fix CPU migration from qemu-2.6 <-> later versions

When migration for target-ppc was converted to vmstate, several
VMSTATE_EQUAL() checks were foolishly included of things that really
should be internal state.  Specifically we verified equality of the
insns_flags and insns_flags2 fields, which are used within TCG to
determine which groups of instructions are available on this cpu
model.  Between qemu-2.6 and qemu-2.7 we made some changes to these
classes which broke migration.

This path fixes migration both forwards and backwards.  On migration
from 2.6 to later versions we import the fields into teporary
variables, which we then ignore.  In migration backwards, we populate
the temporary fields from the runtime fields, but mask out the bits
which were added after qemu-2.6, allowing the VMSTATE_EQUAL in
qemu-2.6 to accept the stream.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
5 years agoppc: Make uninorth interrupt swizzling identical to Grackle
Benjamin Herrenschmidt [Sun, 20 Nov 2016 14:12:31 +0000 (00:12 +1000)] 
ppc: Make uninorth interrupt swizzling identical to Grackle

It's currently broken as it uses an incorrect shift, it tries
to use the slot number but uses the top bits of the bus number
instead.

Note: Neither implementation matches what OpenBIOS ends up putting
in the device-tree either, which will have to be fixed separately.

This is not quite correct for modelling a real Mac since Apple
tend to tie all 4 interrupt lines of a slot together and have
separate interrupts for every slot and every motherboard devices
going straight to the PIC but we'll sort that out later.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
5 years agotarget-ppc: fix index array of national digits
Jose Ricardo Ziviani [Mon, 21 Nov 2016 14:55:13 +0000 (12:55 -0200)] 
target-ppc: fix index array of national digits

Fixes the big endian array access of national digits, from commits
b815587 and e2106d7.

Signed-off-by: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>