qemu.git
5 years agomemory_region: Fix name comments
Dr. David Alan Gilbert [Thu, 9 Mar 2017 15:27:08 +0000 (15:27 +0000)] 
memory_region: Fix name comments

The 'name' parameter to memory_region_init_* had been marked as debug
only, however vmstate_region_ram uses it as a parameter to
qemu_ram_set_idstr to set RAMBlock names and these form part of the
migration stream.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20170309152708.30635-1-dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoMerge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.9-20170314' into staging
Peter Maydell [Tue, 14 Mar 2017 10:13:19 +0000 (10:13 +0000)] 
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.9-20170314' into staging

ppc patch queue for 2017-03-14

This set has a handful og bugfixes to go into qemu-2.9.  This includes
an update to the dtc/libfdt submodule which will fix the build errors
seen on some distributions.

# gpg: Signature made Tue 14 Mar 2017 04:00:41 GMT
# gpg:                using RSA key 0x6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>"
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-for-2.9-20170314:
  dtc: Update submodule to avoid build errors
  pseries: Don't expose PCIe extended config space on older machine types
  target/ppc: fix cpu_ov setting for 32-bit
  target/ppc: Fix wrong number of UAMR register

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agobuild: include sys/sysmacros.h for major() and minor()
Christopher Covington [Wed, 28 Dec 2016 20:04:33 +0000 (15:04 -0500)] 
build: include sys/sysmacros.h for major() and minor()

The definition of the major() and minor() macros are moving within glibc to
<sys/sysmacros.h>. Include this header when it is available to avoid the
following sorts of build-stopping messages:

qga/commands-posix.c: In function ‘dev_major_minor’:
qga/commands-posix.c:656:13: error: In the GNU C Library, "major" is defined
 by <sys/sysmacros.h>. For historical compatibility, it is
 currently defined by <sys/types.h> as well, but we plan to
 remove this soon. To use "major", include <sys/sysmacros.h>
 directly. If you did not intend to use a system-defined macro
 "major", you should undefine it after including <sys/types.h>. [-Werror]
         *devmajor = major(st.st_rdev);
             ^~~~~~~~~~~~~~~~~~~~~~~~~~

qga/commands-posix.c:657:13: error: In the GNU C Library, "minor" is defined
 by <sys/sysmacros.h>. For historical compatibility, it is
 currently defined by <sys/types.h> as well, but we plan to
 remove this soon. To use "minor", include <sys/sysmacros.h>
 directly. If you did not intend to use a system-defined macro
 "minor", you should undefine it after including <sys/types.h>. [-Werror]
         *devminor = minor(st.st_rdev);
             ^~~~~~~~~~~~~~~~~~~~~~~~~~

The additional include allows the build to complete on Fedora 26 (Rawhide)
with glibc version 2.24.90.

Signed-off-by: Christopher Covington <cov@codeaurora.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agodtc: Update submodule to avoid build errors
David Gibson [Tue, 14 Mar 2017 01:24:29 +0000 (12:24 +1100)] 
dtc: Update submodule to avoid build errors

The currently included version of the dtc/libfdt submodule has some build
errors on certain distributions (including RHEL7).  This is due to some
poorly named macros in libfdt.h; they're designed for use with the sparse
static checker, but use reserved names which conflict with some symbols in
the standard headers.

That's been corrected in upstream dtc, this updates the qemu submodule to
bring the fix to qemu.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
5 years agopseries: Don't expose PCIe extended config space on older machine types
David Gibson [Tue, 14 Mar 2017 00:54:17 +0000 (11:54 +1100)] 
pseries: Don't expose PCIe extended config space on older machine types

bb9986452 "spapr_pci: Advertise access to PCIe extended config space"
allowed guests to access the extended config space of PCI Express devices
via the PAPR interfaces, even though the paravirtualized bus mostly acts
like plain PCI.

However, that patch enabled access unconditionally, including for existing
machine types, which is an unwise change in behaviour.  This patch limits
the change to pseries-2.9 (and later) machine types.

Suggested-by: Andrea Bolognani <abologna@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
5 years agotarget/ppc: fix cpu_ov setting for 32-bit
Nikunj A Dadhania [Mon, 13 Mar 2017 09:31:04 +0000 (15:01 +0530)] 
target/ppc: fix cpu_ov setting for 32-bit

A bug was introduced in following commit:

    dc0ad84 target/ppc: update overflow flags for add/sub

As for 32-bit ppc target extracting bit 63 for overflow is not correct.
Made it dependent on TARGET_LOG_BITS. This had broken booting MacOS
9.2.1 image

Reported-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
5 years agotarget/ppc: Fix wrong number of UAMR register
Thomas Huth [Wed, 8 Mar 2017 19:58:43 +0000 (20:58 +0100)] 
target/ppc: Fix wrong number of UAMR register

The SPR UAMR has the number 13, and not 12. (Fortunately it seems like
Linux is not using this register yet - only the privileged version with
number 29 ... that's why nobody noticed this problem yet)

Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
5 years agoMerge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Peter Maydell [Mon, 13 Mar 2017 15:08:01 +0000 (15:08 +0000)] 
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer fixes for 2.9.0-rc1

# gpg: Signature made Mon 13 Mar 2017 11:53:16 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

* remotes/kevin/tags/for-upstream:
  commit: Implement .bdrv_refresh_filename
  mirror: Implement .bdrv_refresh_filename
  block: Refresh filename after changing backing file
  commit: Implement bdrv_commit_top.bdrv_co_get_block_status
  block: Request block status from *file for BDRV_BLOCK_RAW
  block: Remove check_new_perm from bdrv_replace_child()
  migration: Document handling of bdrv_is_allocated() errors
  vvfat: React to bdrv_is_allocated() errors
  backup: React to bdrv_is_allocated() errors
  block: Drop unmaintained 'archipelago' driver
  file-posix: Consider max_segments for BlockLimits.max_transfer
  backup: allow target without .bdrv_get_info

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agoMerge remote-tracking branch 'remotes/ehabkost/tags/x86-pull-request' into staging
Peter Maydell [Mon, 13 Mar 2017 13:16:35 +0000 (13:16 +0000)] 
Merge remote-tracking branch 'remotes/ehabkost/tags/x86-pull-request' into staging

x86: Haswell TSX blacklist fix for 2.9

# gpg: Signature made Fri 10 Mar 2017 18:45:08 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

* remotes/ehabkost/tags/x86-pull-request:
  i386: Change stepping of Haswell to non-blacklisted value
  i386/kvm: Blacklist TSX on known broken hosts
  i386: host_vendor_fms() helper function

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agocommit: Implement .bdrv_refresh_filename
Kevin Wolf [Thu, 9 Mar 2017 10:49:16 +0000 (11:49 +0100)] 
commit: Implement .bdrv_refresh_filename

We want query-block to return the right filename, even if a commit job
put a bdrv_commit_top on top of the actual image format driver. Let
bdrv_commit_top.bdrv_refresh_filename get the filename from its backing
file.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agomirror: Implement .bdrv_refresh_filename
Kevin Wolf [Thu, 9 Mar 2017 10:49:16 +0000 (11:49 +0100)] 
mirror: Implement .bdrv_refresh_filename

We want query-block to return the right filename, even if a mirror job
put a bdrv_mirror_top on top of the actual image format driver. Let
bdrv_mirror_top.bdrv_refresh_filename get the filename from its backing
file.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Refresh filename after changing backing file
Kevin Wolf [Thu, 9 Mar 2017 10:45:39 +0000 (11:45 +0100)] 
block: Refresh filename after changing backing file

In bdrv_open_inherit(), the filename is refreshed after opening the
backing file, but we neglected to do the same when the backing file
changes later.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agocommit: Implement bdrv_commit_top.bdrv_co_get_block_status
Kevin Wolf [Wed, 8 Mar 2017 14:07:12 +0000 (15:07 +0100)] 
commit: Implement bdrv_commit_top.bdrv_co_get_block_status

In some cases, bdrv_co_get_block_status() is called recursively for the
whole backing chain. The automatically inserted bdrv_commit_top filter
driver must not stop the recursion, so implement a callback that simply
forwards the request to bs->backing.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Request block status from *file for BDRV_BLOCK_RAW
Kevin Wolf [Wed, 8 Mar 2017 14:02:49 +0000 (15:02 +0100)] 
block: Request block status from *file for BDRV_BLOCK_RAW

This fixes bdrv_co_get_block_status() for the bdrv_mirror_top block
driver, which must fall through to bs->backing instead of bs->file.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Remove check_new_perm from bdrv_replace_child()
Kevin Wolf [Wed, 8 Mar 2017 13:44:07 +0000 (14:44 +0100)] 
block: Remove check_new_perm from bdrv_replace_child()

All callers pass false now, so the parameter can go away again.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agomigration: Document handling of bdrv_is_allocated() errors
Eric Blake [Wed, 8 Mar 2017 21:34:29 +0000 (15:34 -0600)] 
migration: Document handling of bdrv_is_allocated() errors

Migration is the only code left in the tree that does not react
to bdrv_is_allocated() failures.  But as there is no useful way
to react to the failure, and we are merely skipping unallocated
sectors on success, just document that our choice of handling
is intended.

Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agovvfat: React to bdrv_is_allocated() errors
Eric Blake [Wed, 8 Mar 2017 21:34:28 +0000 (15:34 -0600)] 
vvfat: React to bdrv_is_allocated() errors

If bdrv_is_allocated() fails, we should react to that failure.
For 2 of the 3 callers, reporting the error was easy.  But in
cluster_was_modified() and its lone caller
get_cluster_count_for_direntry(), it's rather invasive to update
the logic to pass the error back; so there, I went with merely
documenting the issue by changing the return type to bool (in
all likelihood, treating the cluster as modified will then
trigger a read which will also fail, and eventually get to an
error - but given the appalling number of abort() calls in this
code, I'm not making it any worse).

Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agobackup: React to bdrv_is_allocated() errors
Eric Blake [Wed, 8 Mar 2017 21:34:27 +0000 (15:34 -0600)] 
backup: React to bdrv_is_allocated() errors

If bdrv_is_allocated() fails, we should immediately do the backup
error action, rather than attempting backup_do_cow() (although
that will likely fail too).

Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoblock: Drop unmaintained 'archipelago' driver
Eric Blake [Wed, 8 Mar 2017 20:02:16 +0000 (14:02 -0600)] 
block: Drop unmaintained 'archipelago' driver

The driver has failed to build since commit da34e65, in qemu 2.6,
due to a missing include of qapi/error.h for error_setg().
Since no one has complained in three releases, it is easier to
remove the dead code than to keep it around, especially since it
is not being built by default and therefore prone to bitrot.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agofile-posix: Consider max_segments for BlockLimits.max_transfer
Fam Zheng [Wed, 8 Mar 2017 12:08:14 +0000 (20:08 +0800)] 
file-posix: Consider max_segments for BlockLimits.max_transfer

BlockLimits.max_transfer can be too high without this fix, guest will
encounter I/O error or even get paused with werror=stop or rerror=stop. The
cause is explained below.

Linux has a separate limit, /sys/block/.../queue/max_segments, which in
the worst case can be more restrictive than the BLKSECTGET which we
already consider (note that they are two different things). So, the
failure scenario before this patch is:

1) host device has max_sectors_kb = 4096 and max_segments = 64;
2) guest learns max_sectors_kb limit from QEMU, but doesn't know
   max_segments;
3) guest issues e.g. a 512KB request thinking it's okay, but actually
   it's not, because it will be passed through to host device as an
   SG_IO req that has niov > 64;
4) host kernel doesn't like the segmenting of the request, and returns
   -EINVAL;

This patch checks the max_segments sysfs entry for the host device and
calculates a "conservative" bytes limit using the page size, which is
then merged into the existing max_transfer limit. Guest will discover
this from the usual virtual block device interfaces. (In the case of
scsi-generic, it will be done in the INQUIRY reply interception in
device model.)

The other possibility is to actually propagate it as a separate limit,
but it's not better. On the one hand, there is a big complication: the
limit is per-LUN in QEMU PoV (because we can attach LUNs from different
host HBAs to the same virtio-scsi bus), but the channel to communicate
it in a per-LUN manner is missing down the stack; on the other hand,
two limits versus one doesn't change much about the valid size of I/O
(because guest has no control over host segmenting).

Also, the idea to fall back to bounce buffering in QEMU, upon -EINVAL,
was explored. Unfortunately there is no neat way to ensure the bounce
buffer is less segmented (in terms of DMA addr) than the guest buffer.

Practically, this bug is not very common. It is only reported on a
Emulex (lpfc), so it's okay to get it fixed in the easier way.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agobackup: allow target without .bdrv_get_info
Vladimir Sementsov-Ogievskiy [Tue, 28 Feb 2017 19:33:40 +0000 (22:33 +0300)] 
backup: allow target without .bdrv_get_info

Currently backup to nbd target is broken, as nbd doesn't have
.bdrv_get_info realization.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoMerge remote-tracking branch 'remotes/famz/tags/docker-pull-request' into staging
Peter Maydell [Mon, 13 Mar 2017 11:26:36 +0000 (11:26 +0000)] 
Merge remote-tracking branch 'remotes/famz/tags/docker-pull-request' into staging

# gpg: Signature made Fri 10 Mar 2017 07:15:38 GMT
# gpg:                using RSA key 0xCA35624C6A9171C6
# gpg: Good signature from "Fam Zheng <famz@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: 5003 7CB7 9706 0F76 F021  AD56 CA35 624C 6A91 71C6

* remotes/famz/tags/docker-pull-request:
  docker/dockerfiles/debian-s390-cross: include clang
  tests/docker: support proxy / corporate firewall

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agoi386: Change stepping of Haswell to non-blacklisted value
Eduardo Habkost [Thu, 9 Mar 2017 18:12:12 +0000 (15:12 -0300)] 
i386: Change stepping of Haswell to non-blacklisted value

glibc blacklists TSX on Haswell CPUs with model==60 and
stepping < 4. To make the Haswell CPU model more useful, make
those guests actually use TSX by changing CPU stepping to 4.

References:
* glibc commit 2702856bf45c82cf8e69f2064f5aa15c0ceb6359
  https://sourceware.org/git/?p=glibc.git;a=commit;h=2702856bf45c82cf8e69f2064f5aa15c0ceb6359

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170309181212.18864-4-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
5 years agoi386/kvm: Blacklist TSX on known broken hosts
Eduardo Habkost [Thu, 9 Mar 2017 18:12:11 +0000 (15:12 -0300)] 
i386/kvm: Blacklist TSX on known broken hosts

Some Intel CPUs are known to have a broken TSX implementation. A
microcode update from Intel disabled TSX on those CPUs, but
GET_SUPPORTED_CPUID might be reporting it as supported if the
hosts were not updated yet.

Manually fixup the GET_SUPPORTED_CPUID data to ensure we will
never enable TSX when running on those hosts.

Reference:
* glibc commit 2702856bf45c82cf8e69f2064f5aa15c0ceb6359:
  https://sourceware.org/git/?p=glibc.git;a=commit;h=2702856bf45c82cf8e69f2064f5aa15c0ceb6359

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170309181212.18864-3-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
5 years agoi386: host_vendor_fms() helper function
Eduardo Habkost [Thu, 9 Mar 2017 18:12:10 +0000 (15:12 -0300)] 
i386: host_vendor_fms() helper function

Helper function for code that needs to check the host CPU
vendor/family/model/stepping values.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170309181212.18864-2-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
5 years agodocker/dockerfiles/debian-s390-cross: include clang
Alex Bennée [Mon, 6 Mar 2017 11:28:48 +0000 (11:28 +0000)] 
docker/dockerfiles/debian-s390-cross: include clang

It's a silly little limitation on Shippable that is looks for clang
in the container even though we won't use it. The arm/aarch64 cross
builds inherit this from debian.docker but as we needed to use
debian-testing for this we add it here. We also collapse the update
step into one RUN line to remove and intermediate layer of the docker
build.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20170306112848.659-1-alex.bennee@linaro.org>
Signed-off-by: Fam Zheng <famz@redhat.com>
5 years agoMerge remote-tracking branch 'remotes/stsquad/tags/pull-mttcg-fixups-090317-1' into...
Peter Maydell [Thu, 9 Mar 2017 18:53:55 +0000 (18:53 +0000)] 
Merge remote-tracking branch 'remotes/stsquad/tags/pull-mttcg-fixups-090317-1' into staging

Fix-ups for MTTCG regressions for 2.9

This is the same as v3 posted a few days ago except with a few extra
Reviewed-by tags added.

# gpg: Signature made Thu 09 Mar 2017 10:45:18 GMT
# gpg:                using RSA key 0xFBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>"
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* remotes/stsquad/tags/pull-mttcg-fixups-090317-1:
  hw/intc/arm_gic: modernise the DPRINTF
  target/arm/helper: make it clear the EC field is also in hex
  target-i386: defer VMEXIT to do_interrupt
  target/mips: hold BQL for timer interrupts
  translate-all: exit cpu_restore_state early if translating
  target/xtensa: hold BQL for interrupt processing
  s390x/misc_helper.c: wrap IO instructions in BQL
  sparc/sparc64: grab BQL before calling cpu_check_irqs
  cpus.c: add additional error_report when !TARGET_SUPPORT_MTTCG
  target/i386/cpu.h: declare TCG_GUEST_DEFAULT_MO
  vl/cpus: be smarter with icount and MTTCG

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-fixes-20170309-1' into staging
Peter Maydell [Thu, 9 Mar 2017 13:16:05 +0000 (13:16 +0000)] 
Merge remote-tracking branch 'remotes/kraxel/tags/pull-fixes-20170309-1' into staging

2.9 bugfixes for ohci and qxl

# gpg: Signature made Thu 09 Mar 2017 09:09:44 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

* remotes/kraxel/tags/pull-fixes-20170309-1:
  qxl: clear guest_cursor on QXL_CURSOR_HIDE
  ohci: relax link check

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agohw/intc/arm_gic: modernise the DPRINTF
Alex Bennée [Thu, 2 Mar 2017 19:08:52 +0000 (19:08 +0000)] 
hw/intc/arm_gic: modernise the DPRINTF

While I was debugging the icount issues I realised a bunch of the
messages look quite similar. I've fixed this by including __func__ in
the debug print. At the same time I move the a modern if (GATE) style
printf which ensures the compiler can check for format string errors
even if the code gets optimised away in the non-DEBUG_GIC case.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
5 years agotarget/arm/helper: make it clear the EC field is also in hex
Alex Bennée [Thu, 2 Mar 2017 14:07:16 +0000 (14:07 +0000)] 
target/arm/helper: make it clear the EC field is also in hex

..just like the rest of the displayed ESR register. Otherwise people
might scratch their heads if a not obviously hex number is displayed
for the EC field.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: KONRAD Frederic <fred.konrad@greensocs.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
5 years agotarget-i386: defer VMEXIT to do_interrupt
Paolo Bonzini [Tue, 7 Mar 2017 11:37:36 +0000 (12:37 +0100)] 
target-i386: defer VMEXIT to do_interrupt

Paths through the softmmu code during code generation now need to be audited
to check for double locking of tb_lock.  In particular, VMEXIT can take tb_lock
through cpu_vmexit -> cpu_x86_update_cr4 -> tlb_flush.

To avoid this, split VMEXIT delivery in two parts, similar to what is done with
exceptions.  cpu_vmexit only records the VMEXIT exit code and information, and
cc->do_interrupt can then deliver it when it is safe to take the lock.

Reported-by: Alexander Boettcher <alexander.boettcher@genode-labs.com>
Suggested-by: Richard Henderson <rth@twiddle.net>
Tested-by: Alexander Boettcher <alexander.boettcher@genode-labs.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agotarget/mips: hold BQL for timer interrupts
Yongbok Kim [Fri, 3 Mar 2017 11:20:21 +0000 (11:20 +0000)] 
target/mips: hold BQL for timer interrupts

Hold BQL when accessing timer which can cause interrupts

Signed-off-by: Yongbok Kim <yongbok.kim@imgtec.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
5 years agotranslate-all: exit cpu_restore_state early if translating
Alex Bennée [Thu, 2 Mar 2017 10:31:32 +0000 (10:31 +0000)] 
translate-all: exit cpu_restore_state early if translating

The translation code uses cpu_ld*_code which can trigger a tlb_fill
which if it fails will erroneously attempts a fault resolution. This
never works during translation as the TB being generated hasn't been
added yet. The target should have checked retaddr before calling
cpu_restore_state but for those that have yet to be fixed we do it
here to avoid a recursive tb_lock() under MTTCG's new locking regime.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agotarget/xtensa: hold BQL for interrupt processing
Alex Bennée [Wed, 1 Mar 2017 20:29:10 +0000 (20:29 +0000)] 
target/xtensa: hold BQL for interrupt processing

Make sure we have the BQL held when processing interrupts.

Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Max Filippov <jcmvbkbc@gmail.com>
5 years agos390x/misc_helper.c: wrap IO instructions in BQL
Alex Bennée [Wed, 1 Mar 2017 16:22:40 +0000 (16:22 +0000)] 
s390x/misc_helper.c: wrap IO instructions in BQL

Helpers that can trigger IO events (including interrupts) need to be
protected by the BQL. I've updated all the helpers that call into an
ioinst_handle_* functions.

Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
5 years agosparc/sparc64: grab BQL before calling cpu_check_irqs
Alex Bennée [Wed, 1 Mar 2017 13:20:02 +0000 (13:20 +0000)] 
sparc/sparc64: grab BQL before calling cpu_check_irqs

IRQ modification is part of device emulation and should be done while
the BQL is held to prevent races when MTTCG is enabled. This adds
assertions in the hw emulation layer and wraps the calls from helpers
in the BQL.

Reported-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
5 years agocpus.c: add additional error_report when !TARGET_SUPPORT_MTTCG
Alex Bennée [Tue, 28 Feb 2017 14:40:17 +0000 (14:40 +0000)] 
cpus.c: add additional error_report when !TARGET_SUPPORT_MTTCG

While we may fail the memory ordering check later that can be
confusing. So in cases where TARGET_SUPPORT_MTTCG has yet to be
defined we should say so specifically.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
5 years agotarget/i386/cpu.h: declare TCG_GUEST_DEFAULT_MO
Alex Bennée [Tue, 28 Feb 2017 14:35:14 +0000 (14:35 +0000)] 
target/i386/cpu.h: declare TCG_GUEST_DEFAULT_MO

This suppresses the incorrect warning when forcing MTTCG for x86
guests on x86 hosts. A future patch will still warn when
TARGET_SUPPORT_MTTCG hasn't been defined for the guest (which is still
pending for x86).

Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Acked-by: Eduardo Habkost <ehabkost@redhat.com>
5 years agovl/cpus: be smarter with icount and MTTCG
Alex Bennée [Mon, 27 Feb 2017 17:09:01 +0000 (17:09 +0000)] 
vl/cpus: be smarter with icount and MTTCG

The sense of the test was inverted. Make it simple, if icount is
enabled then we disabled MTTCG by default. If the user tries to force
MTTCG upon us then we tell them "no".

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
5 years agoqxl: clear guest_cursor on QXL_CURSOR_HIDE
Gerd Hoffmann [Mon, 6 Mar 2017 08:31:51 +0000 (09:31 +0100)] 
qxl: clear guest_cursor on QXL_CURSOR_HIDE

Make sure we don't leave guest_cursor pointing into nowhere.  This might
lead to (rare) live migration failures, due to target trying to restore
the cursor from the stale pointer.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1421788
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 1488789111-27340-1-git-send-email-kraxel@redhat.com

5 years agoohci: relax link check
Gerd Hoffmann [Tue, 7 Mar 2017 08:40:18 +0000 (09:40 +0100)] 
ohci: relax link check

The strict td link limit added by commit "95ed569 usb: ohci: limit the
number of link eds" causes problems with macos guests.  Lets raise the
limit.

Reported-by: Programmingkid <programmingkidx@gmail.com>
Reported-by: Howard Spoelstra <hsp.cat7@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: John Arbuckle <programmingkidx@gmail.com>
Message-id: 1488876018-31576-1-git-send-email-kraxel@redhat.com

5 years agoMerge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Peter Maydell [Wed, 8 Mar 2017 09:47:52 +0000 (09:47 +0000)] 
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer fixes for 2.9.0-rc0

# gpg: Signature made Tue 07 Mar 2017 14:59:18 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

* remotes/kevin/tags/for-upstream: (27 commits)
  commit: Don't use error_abort in commit_start
  block: Don't use error_abort in blk_new_open
  sheepdog: Support blockdev-add
  qapi-schema: Rename SocketAddressFlat's variant tcp to inet
  qapi-schema: Rename GlusterServer to SocketAddressFlat
  gluster: Plug memory leaks in qemu_gluster_parse_json()
  gluster: Don't duplicate qapi-util.c's qapi_enum_parse()
  gluster: Drop assumptions on SocketTransport names
  sheepdog: Implement bdrv_parse_filename()
  sheepdog: Use SocketAddress and socket_connect()
  sheepdog: Report errors in pseudo-filename more usefully
  sheepdog: Don't truncate long VDI name in _open(), _create()
  sheepdog: Fix snapshot ID parsing in _open(), _create, _goto()
  sheepdog: Mark sd_snapshot_delete() lossage FIXME
  sheepdog: Fix error handling sd_create()
  sheepdog: Fix error handling in sd_snapshot_delete()
  sheepdog: Defuse time bomb in sd_open() error handling
  block: Fix error handling in bdrv_replace_in_backing_chain()
  block: Handle permission errors in change_parent_backing_link()
  block: Ignore multiple children in bdrv_check_update_perm()
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-block-2017-02-28-v4' into...
Peter Maydell [Tue, 7 Mar 2017 17:06:48 +0000 (17:06 +0000)] 
Merge remote-tracking branch 'remotes/armbru/tags/pull-block-2017-02-28-v4' into staging

block: Command line option -blockdev

# gpg: Signature made Tue 07 Mar 2017 15:07:59 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

* remotes/armbru/tags/pull-block-2017-02-28-v4: (24 commits)
  keyval: Support lists
  docs/qapi-code-gen.txt: Clarify naming rules
  qapi: Improve how keyval input visitor reports unexpected dicts
  block: Initial implementation of -blockdev
  qapi: New qobject_input_visitor_new_str() for convenience
  keyval: Restrict key components to valid QAPI names
  qapi: New parse_qapi_name()
  test-qapi-util: New, covering qapi/qapi-util.c
  monitor: Assert qmp_schema_json[] is sane
  test-visitor-serialization: Pass &error_abort to qobject_from_json()
  check-qjson: Test errors from qobject_from_json()
  block: More detailed syntax error reporting for JSON filenames
  qobject: Propagate parse errors through qobject_from_json()
  test-qobject-input-visitor: Abort earlier on bad test input
  qjson: Abort earlier on qobject_from_jsonf() misuse
  libqtest: Fix qmp() & friends to abort on JSON parse errors
  qobject: Propagate parse errors through qobject_from_jsonv()
  qapi: Factor out common qobject_input_get_keyval()
  qapi: Factor out common part of qobject input visitor creation
  test-keyval: Cover use with qobject input visitor
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5 years agokeyval: Support lists
Markus Armbruster [Tue, 28 Feb 2017 21:27:10 +0000 (22:27 +0100)] 
keyval: Support lists

Additionally permit non-negative integers as key components.  A
dictionary's keys must either be all integers or none.  If all keys
are integers, convert the dictionary to a list.  The set of keys must
be [0,N].

Examples:

* list.1=goner,list.0=null,list.1=eins,list.2=zwei
  is equivalent to JSON [ "null", "eins", "zwei" ]

* a.b.c=1,a.b.0=2
  is inconsistent: a.b.c clashes with a.b.0

* list.0=null,list.2=eins,list.2=zwei
  has a hole: list.1 is missing

Similar design flaw as for objects: there is no way to denote an empty
list.  While interpreting "key absent" as empty list seems natural
(removing a list member from the input string works when there are
multiple ones, so why not when there's just one), it doesn't work:
"key absent" already means "optional list absent", which isn't the
same as "empty list present".

Update the keyval object visitor to use this a.0 syntax in error
messages rather than the usual a[0].

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1488317230-26248-25-git-send-email-armbru@redhat.com>
[Off-by-one fix squashed in, as per Kevin's review]
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
5 years agodocs/qapi-code-gen.txt: Clarify naming rules
Markus Armbruster [Tue, 28 Feb 2017 21:27:09 +0000 (22:27 +0100)] 
docs/qapi-code-gen.txt: Clarify naming rules

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-24-git-send-email-armbru@redhat.com>

5 years agoqapi: Improve how keyval input visitor reports unexpected dicts
Markus Armbruster [Tue, 28 Feb 2017 21:27:08 +0000 (22:27 +0100)] 
qapi: Improve how keyval input visitor reports unexpected dicts

Incorrect option

    -blockdev node-name=foo,driver=file,filename=foo.img,aio.unmap=on

is rejected with "Invalid parameter type for 'aio', expected: string".
To make sense of this, you almost have to translate it into the
equivalent QMP command

    { "execute": "blockdev-add", "arguments": { "node-name": "foo", "driver": "file", "filename": "foo.img", "aio": { "unmap": true } } }

Improve the error message to "Parameters 'aio.*' are unexpected".
Take care not to confuse the case "unexpected nested parameters"
(i.e. the object is a QDict or QList) with the case "non-string scalar
parameter".  The latter is a misuse of the visitor, and should perhaps
be an assertion.  Note that test-qobject-input-visitor exercises this
misuse in test_visitor_in_int_keyval(), test_visitor_in_bool_keyval()
and test_visitor_in_number_keyval().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-23-git-send-email-armbru@redhat.com>

5 years agoblock: Initial implementation of -blockdev
Markus Armbruster [Tue, 28 Feb 2017 21:27:07 +0000 (22:27 +0100)] 
block: Initial implementation of -blockdev

The new command line option -blockdev works like QMP command
blockdev-add.

The option argument may be given in JSON syntax, exactly as in QMP.
Example usage:

    -blockdev '{"node-name": "foo", "driver": "raw", "file": {"driver": "file", "filename": "foo.img"} }'

The JSON argument doesn't exactly blend into the existing option
syntax, so the traditional KEY=VALUE,... syntax is also supported,
using dotted keys to do the nesting:

    -blockdev node-name=foo,driver=raw,file.driver=file,file.filename=foo.img

This does not yet support lists, but that will be addressed shortly.

Note that calling qmp_blockdev_add() (say via qmp_marshal_block_add())
right away would crash.  We need to stash the configuration for later
instead.  This is crudely done, and bypasses QemuOpts, even though
storing configuration is what QemuOpts is for.  Need to revamp option
infrastructure to support QAPI types like BlockdevOptions.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1488317230-26248-22-git-send-email-armbru@redhat.com>

5 years agoqapi: New qobject_input_visitor_new_str() for convenience
Markus Armbruster [Tue, 28 Feb 2017 21:27:06 +0000 (22:27 +0100)] 
qapi: New qobject_input_visitor_new_str() for convenience

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1488317230-26248-21-git-send-email-armbru@redhat.com>

5 years agokeyval: Restrict key components to valid QAPI names
Markus Armbruster [Tue, 28 Feb 2017 21:27:05 +0000 (22:27 +0100)] 
keyval: Restrict key components to valid QAPI names

Until now, key components are separated by '.'.  This leaves little
room for evolving the syntax, and is incompatible with the __RFQDN_
prefix convention for downstream extensions.

Since key components will be commonly used as QAPI member names by the
QObject input visitor, we can just as well borrow the QAPI naming
rules here: letters, digits, hyphen and period starting with a letter,
with an optional __RFQDN_ prefix for downstream extensions.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-20-git-send-email-armbru@redhat.com>

5 years agoqapi: New parse_qapi_name()
Markus Armbruster [Tue, 28 Feb 2017 21:27:04 +0000 (22:27 +0100)] 
qapi: New parse_qapi_name()

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1488317230-26248-19-git-send-email-armbru@redhat.com>

5 years agotest-qapi-util: New, covering qapi/qapi-util.c
Markus Armbruster [Tue, 28 Feb 2017 21:27:03 +0000 (22:27 +0100)] 
test-qapi-util: New, covering qapi/qapi-util.c

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-18-git-send-email-armbru@redhat.com>

5 years agomonitor: Assert qmp_schema_json[] is sane
Markus Armbruster [Tue, 28 Feb 2017 21:27:02 +0000 (22:27 +0100)] 
monitor: Assert qmp_schema_json[] is sane

qmp_query_qmp_schema() parses qmp_schema_json[] with
qobject_from_json().  This must not fail, so pass &error_abort.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-17-git-send-email-armbru@redhat.com>

5 years agotest-visitor-serialization: Pass &error_abort to qobject_from_json()
Markus Armbruster [Tue, 28 Feb 2017 21:27:01 +0000 (22:27 +0100)] 
test-visitor-serialization: Pass &error_abort to qobject_from_json()

qmp_deserialize() calls qobject_from_json() ignoring errors.  It
passes the result to qobject_input_visitor_new(), which asserts it's
not null.  Therefore, we can just as well pass &error_abort to
qobject_from_json().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-16-git-send-email-armbru@redhat.com>

5 years agocheck-qjson: Test errors from qobject_from_json()
Markus Armbruster [Tue, 28 Feb 2017 21:27:00 +0000 (22:27 +0100)] 
check-qjson: Test errors from qobject_from_json()

Pass &error_abort with known-good input.  Else pass &err and check
what comes back.  This demonstrates that the parser fails silently for
many errors.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-15-git-send-email-armbru@redhat.com>

5 years agoblock: More detailed syntax error reporting for JSON filenames
Markus Armbruster [Tue, 28 Feb 2017 21:26:59 +0000 (22:26 +0100)] 
block: More detailed syntax error reporting for JSON filenames

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-14-git-send-email-armbru@redhat.com>

5 years agoqobject: Propagate parse errors through qobject_from_json()
Markus Armbruster [Tue, 28 Feb 2017 21:26:58 +0000 (22:26 +0100)] 
qobject: Propagate parse errors through qobject_from_json()

The next few commits will put the errors to use where appropriate.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1488317230-26248-13-git-send-email-armbru@redhat.com>

5 years agotest-qobject-input-visitor: Abort earlier on bad test input
Markus Armbruster [Tue, 28 Feb 2017 21:26:57 +0000 (22:26 +0100)] 
test-qobject-input-visitor: Abort earlier on bad test input

visitor_input_test_init_internal() parses test input with
qobject_from_jsonv(), and asserts it succeeds.  Pass &error_abort for
good measure.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-12-git-send-email-armbru@redhat.com>

5 years agoqjson: Abort earlier on qobject_from_jsonf() misuse
Markus Armbruster [Tue, 28 Feb 2017 21:26:56 +0000 (22:26 +0100)] 
qjson: Abort earlier on qobject_from_jsonf() misuse

Ignoring errors first, then asserting success is suboptimal.  Pass
&error_abort instead, so we abort earlier, and hopefully get more
useful clues on what's wrong.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-11-git-send-email-armbru@redhat.com>

5 years agolibqtest: Fix qmp() & friends to abort on JSON parse errors
Markus Armbruster [Tue, 28 Feb 2017 21:26:55 +0000 (22:26 +0100)] 
libqtest: Fix qmp() & friends to abort on JSON parse errors

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-10-git-send-email-armbru@redhat.com>

5 years agoqobject: Propagate parse errors through qobject_from_jsonv()
Markus Armbruster [Tue, 28 Feb 2017 21:26:54 +0000 (22:26 +0100)] 
qobject: Propagate parse errors through qobject_from_jsonv()

The next few commits will put the errors to use where appropriate.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-9-git-send-email-armbru@redhat.com>

5 years agoqapi: Factor out common qobject_input_get_keyval()
Markus Armbruster [Tue, 28 Feb 2017 21:26:53 +0000 (22:26 +0100)] 
qapi: Factor out common qobject_input_get_keyval()

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <1488317230-26248-8-git-send-email-armbru@redhat.com>

5 years agoqapi: Factor out common part of qobject input visitor creation
Markus Armbruster [Tue, 28 Feb 2017 21:26:52 +0000 (22:26 +0100)] 
qapi: Factor out common part of qobject input visitor creation

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-7-git-send-email-armbru@redhat.com>

5 years agotest-keyval: Cover use with qobject input visitor
Markus Armbruster [Tue, 28 Feb 2017 21:26:51 +0000 (22:26 +0100)] 
test-keyval: Cover use with qobject input visitor

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-6-git-send-email-armbru@redhat.com>

5 years agoqapi: qobject input visitor variant for use with keyval_parse()
Daniel P. Berrange [Tue, 28 Feb 2017 21:26:50 +0000 (22:26 +0100)] 
qapi: qobject input visitor variant for use with keyval_parse()

Currently the QObjectInputVisitor assumes that all scalar values are
directly represented as the final types declared by the thing being
visited. i.e. it assumes an 'int' is using QInt, and a 'bool' is using
QBool, etc.  This is good when QObjectInputVisitor is fed a QObject
that came from a JSON document on the QMP monitor, as it will strictly
validate correctness.

To allow QObjectInputVisitor to be reused for visiting a QObject
originating from keyval_parse(), an alternative mode is needed where
all the scalars types are represented as QString and converted on the
fly to the final desired type.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1475246744-29302-8-git-send-email-berrange@redhat.com>

Rebased, conflicts resolved, commit message updated to refer to
keyval_parse().  autocast replaced by keyval in identifiers,
noautocast replaced by fail in tests.

Fix qobject_input_type_uint64_keyval() not to reject '-', for QemuOpts
compatibility: replace parse_uint_full() by open-coded
parse_option_number().  The next commit will add suitable tests.
Leave out the fancy ERANGE error reporting for now, but add a TODO
comment.  Add it qobject_input_type_int64_keyval() and
qobject_input_type_number_keyval(), too.

Open code parse_option_bool() and parse_option_size() so we have to
call qobject_input_get_name() only when actually needed.  Again, leave
out ERANGE error reporting for now.

QAPI/QMP downstream extension prefixes __RFQDN_ don't work, because
keyval_parse() splits them at '.'.  This will be addressed later in
the series.

qobject_input_type_int64_keyval(), qobject_input_type_uint64_keyval(),
qobject_input_type_number_keyval() tweaked for style.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-5-git-send-email-armbru@redhat.com>

5 years agokeyval: New keyval_parse()
Markus Armbruster [Tue, 28 Feb 2017 21:26:49 +0000 (22:26 +0100)] 
keyval: New keyval_parse()

keyval_parse() parses KEY=VALUE,... into a QDict.  Works like
qemu_opts_parse(), except:

* Returns a QDict instead of a QemuOpts (d'oh).

* Supports nesting, unlike QemuOpts: a KEY is split into key
  fragments at '.' (dotted key convention; the block layer does
  something similar on top of QemuOpts).  The key fragments are QDict
  keys, and the last one's value is updated to VALUE.

* Each key fragment may be up to 127 bytes long.  qemu_opts_parse()
  limits the entire key to 127 bytes.

* Overlong key fragments are rejected.  qemu_opts_parse() silently
  truncates them.

* Empty key fragments are rejected.  qemu_opts_parse() happily
  accepts empty keys.

* It does not store the returned value.  qemu_opts_parse() stores it
  in the QemuOptsList.

* It does not treat parameter "id" specially.  qemu_opts_parse()
  ignores all but the first "id", and fails when its value isn't
  id_wellformed(), or duplicate (a QemuOpts with the same ID is
  already stored).  It also screws up when a value contains ",id=".

* Implied value is not supported.  qemu_opts_parse() desugars "foo" to
  "foo=on", and "nofoo" to "foo=off".

* An implied key's value can't be empty, and can't contain ','.

I intend to grow this into a saner replacement for QemuOpts.  It'll
take time, though.

Note: keyval_parse() provides no way to do lists, and its key syntax
is incompatible with the __RFQDN_ prefix convention for downstream
extensions, because it blindly splits at '.', even in __RFQDN_.  Both
issues will be addressed later in the series.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1488317230-26248-4-git-send-email-armbru@redhat.com>

5 years agotests: Fix gcov-files-test-qemu-opts-y, gcov-files-test-logging-y
Markus Armbruster [Tue, 28 Feb 2017 21:26:48 +0000 (22:26 +0100)] 
tests: Fix gcov-files-test-qemu-opts-y, gcov-files-test-logging-y

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-3-git-send-email-armbru@redhat.com>

5 years agotest-qemu-opts: Cover qemu_opts_parse() of "no"
Markus Armbruster [Tue, 28 Feb 2017 21:26:47 +0000 (22:26 +0100)] 
test-qemu-opts: Cover qemu_opts_parse() of "no"

qemu_opts_parse() interprets "no" as negated empty key.  Consistent
with its acceptance of empty keys elsewhere, whatever that's worth.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1488317230-26248-2-git-send-email-armbru@redhat.com>

5 years agodisas/arm: Avoid unintended sign extension
Peter Maydell [Fri, 3 Mar 2017 15:50:33 +0000 (15:50 +0000)] 
disas/arm: Avoid unintended sign extension

When assembling 'given' from the instruction bytes, C's integer
promotion rules mean we may promote an unsigned char to a signed
integer before shifting it, and then sign extend to a 64-bit long,
which can set the high bits of the long.  The code doesn't in fact
care about the high bits if the long is 64 bits, but this is
surprising, so don't do it.

(Spotted by Coverity, CID 1005404.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1488556233-31246-7-git-send-email-peter.maydell@linaro.org

5 years agodisas/cris: Avoid unintended sign extension
Peter Maydell [Fri, 3 Mar 2017 15:50:32 +0000 (15:50 +0000)] 
disas/cris: Avoid unintended sign extension

In the cris disassembler we were using 'unsigned long' to calculate
addresses which are supposed to be 32 bits.  This meant that we might
accidentally sign extend or calculate a value that was outside the 32
bit range of the guest CPU.  Use 'uint32_t' instead so we give the
right answers on 64-bit hosts.

(Spotted by Coverity, CID 1005402, 1005403.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 1488556233-31246-6-git-send-email-peter.maydell@linaro.org

5 years agodisas/microblaze: Avoid unintended sign extension
Peter Maydell [Fri, 3 Mar 2017 15:50:31 +0000 (15:50 +0000)] 
disas/microblaze: Avoid unintended sign extension

In read_insn_microblaze() we assemble 4 bytes into an 'unsigned
long'.  If 'unsigned long' is 64 bits and the high byte has its top
bit set, then C's implicit conversion from 'unsigned char' to 'int'
for the shift will result in an unintended sign extension which sets
the top 32 bits in 'inst'.  Add casts to prevent this.  (Spotted by
Coverity, CID 1005401.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 1488556233-31246-5-git-send-email-peter.maydell@linaro.org

5 years agodisas/m68k: Avoid unintended sign extension in get_field()
Peter Maydell [Fri, 3 Mar 2017 15:50:30 +0000 (15:50 +0000)] 
disas/m68k: Avoid unintended sign extension in get_field()

In get_field(), we take an 'unsigned char' value and shift it left,
which implicitly promotes it to 'signed int', before ORing it into an
'unsigned long' type.  If 'unsigned long' is 64 bits then this will
result in a sign extension and the top 32 bits of the result will be
1s.  Add explicit casts to unsigned long before shifting to prevent
this.

(Spotted by Coverity, CID 715697.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-id: 1488556233-31246-4-git-send-email-peter.maydell@linaro.org

5 years agodisas/i386: Avoid NULL pointer dereference in error case
Peter Maydell [Fri, 3 Mar 2017 15:50:29 +0000 (15:50 +0000)] 
disas/i386: Avoid NULL pointer dereference in error case

In a code path where we hit an internal disassembler error, execution
would subsequently attempt to dereference a NULL pointer.  This
should never happen, but avoid the crash.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1488556233-31246-3-git-send-email-peter.maydell@linaro.org

5 years agodisas/hppa: Remove dead code
Peter Maydell [Fri, 3 Mar 2017 15:50:28 +0000 (15:50 +0000)] 
disas/hppa: Remove dead code

Coverity complains (CID 1302705) that the "fr0" part of the ?: in
fput_fp_reg_r() is dead.  This looks like cut-n-paste error from
fput_fp_reg(); delete the dead code.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 1488556233-31246-2-git-send-email-peter.maydell@linaro.org

5 years agocommit: Don't use error_abort in commit_start
Fam Zheng [Tue, 7 Mar 2017 11:07:22 +0000 (19:07 +0800)] 
commit: Don't use error_abort in commit_start

bdrv_set_backing_hd failure needn't be abort. Since we already have
error parameter, use it.

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoblock: Don't use error_abort in blk_new_open
Fam Zheng [Tue, 7 Mar 2017 11:07:21 +0000 (19:07 +0800)] 
block: Don't use error_abort in blk_new_open

We have an errp and bdrv_root_attach_child can fail permission check,
error_abort is not the best choice here.

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Support blockdev-add
Markus Armbruster [Mon, 6 Mar 2017 19:00:49 +0000 (20:00 +0100)] 
sheepdog: Support blockdev-add

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoqapi-schema: Rename SocketAddressFlat's variant tcp to inet
Markus Armbruster [Mon, 6 Mar 2017 19:00:48 +0000 (20:00 +0100)] 
qapi-schema: Rename SocketAddressFlat's variant tcp to inet

QAPI type SocketAddressFlat differs from SocketAddress pointlessly:
the discriminator value for variant InetSocketAddress is 'tcp' instead
of 'inet'.  Rename.

The type is so far only used by the Gluster block drivers.  Take care
to keep 'tcp' working in things like -drive's file.server.0.type=tcp.
The "gluster+tcp" URI scheme in pseudo-filenames stays the same.
blockdev-add changes, but it has changed incompatibly since 2.8
already.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoqapi-schema: Rename GlusterServer to SocketAddressFlat
Markus Armbruster [Mon, 6 Mar 2017 19:00:47 +0000 (20:00 +0100)] 
qapi-schema: Rename GlusterServer to SocketAddressFlat

As its documentation says, it's not specific to Gluster.  Rename it,
as I'm going to use it for something else.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agogluster: Plug memory leaks in qemu_gluster_parse_json()
Markus Armbruster [Mon, 6 Mar 2017 19:00:46 +0000 (20:00 +0100)] 
gluster: Plug memory leaks in qemu_gluster_parse_json()

To reproduce, run

    $ valgrind qemu-system-x86_64 --nodefaults -S --drive driver=gluster,volume=testvol,path=/a/b/c,server.0.type=xxx

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agogluster: Don't duplicate qapi-util.c's qapi_enum_parse()
Markus Armbruster [Mon, 6 Mar 2017 19:00:45 +0000 (20:00 +0100)] 
gluster: Don't duplicate qapi-util.c's qapi_enum_parse()

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agogluster: Drop assumptions on SocketTransport names
Markus Armbruster [Mon, 6 Mar 2017 19:00:44 +0000 (20:00 +0100)] 
gluster: Drop assumptions on SocketTransport names

qemu_gluster_glfs_init() passes the names of QAPI enumeration type
SocketTransport to glfs_set_volfile_server().  Works, because they
were chosen to match.  But the coupling is artificial.  Use the
appropriate literal strings instead.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Implement bdrv_parse_filename()
Markus Armbruster [Mon, 6 Mar 2017 19:00:43 +0000 (20:00 +0100)] 
sheepdog: Implement bdrv_parse_filename()

This permits configuration with driver-specific options in addition to
pseudo-filename parsed as URI.  For instance,

    --drive driver=sheepdog,host=fido,vdi=dolly

instead of

    --drive driver=sheepdog,file=sheepdog://fido/dolly

It's also a first step towards supporting blockdev-add.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Use SocketAddress and socket_connect()
Markus Armbruster [Mon, 6 Mar 2017 19:00:42 +0000 (20:00 +0100)] 
sheepdog: Use SocketAddress and socket_connect()

sd_parse_uri() builds a string from host and port parts for
inet_connect().  inet_connect() parses it into host, port and options.
Whether this gets exactly the same host, port and no options for all
inputs is not obvious.

Cut out the string middleman and build a SocketAddress for
socket_connect() instead.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Report errors in pseudo-filename more usefully
Markus Armbruster [Mon, 6 Mar 2017 19:00:41 +0000 (20:00 +0100)] 
sheepdog: Report errors in pseudo-filename more usefully

Errors in the pseudo-filename are all reported with the same laconic
"Can't parse filename" message.

Add real error reporting, such as:

    $ qemu-system-x86_64 --drive driver=sheepdog,filename=sheepdog:///
    qemu-system-x86_64: --drive driver=sheepdog,filename=sheepdog:///: missing file path in URI
    $ qemu-system-x86_64 --drive driver=sheepdog,filename=sheepgod:///vdi
    qemu-system-x86_64: --drive driver=sheepdog,filename=sheepgod:///vdi: URI scheme must be 'sheepdog', 'sheepdog+tcp', or 'sheepdog+unix'
    $ qemu-system-x86_64 --drive driver=sheepdog,filename=sheepdog+unix:///vdi?socke=sheepdog.sock
    qemu-system-x86_64: --drive driver=sheepdog,filename=sheepdog+unix:///vdi?socke=sheepdog.sock: unexpected query parameters

The code to translate legacy syntax to URI fails to escape URI
meta-characters.  The new error messages are misleading then.  Replace
them by the old "Can't parse filename" message.  "Internal error"
would be more honest.  Anyway, no worse than before.  Also add a FIXME
comment.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Don't truncate long VDI name in _open(), _create()
Markus Armbruster [Mon, 6 Mar 2017 19:00:40 +0000 (20:00 +0100)] 
sheepdog: Don't truncate long VDI name in _open(), _create()

sd_parse_uri() truncates long VDI names silently.  Reject them
instead.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Fix snapshot ID parsing in _open(), _create, _goto()
Markus Armbruster [Mon, 6 Mar 2017 19:00:39 +0000 (20:00 +0100)] 
sheepdog: Fix snapshot ID parsing in _open(), _create, _goto()

sd_parse_uri() and sd_snapshot_goto() screw up error checking after
strtoul(), and truncate long tag names silently.  Fix by replacing
those parts by new sd_parse_snapid_or_tag(), which checks more
carefully.

sd_snapshot_delete() also parses snapshot IDs, but is currently too
broken for me to touch.  Mark TODO.

Two calls of strtol() without error checking remain in
parse_redundancy().  Mark them FIXME.

More silent truncation of configuration strings remains elsewhere.
Not marked.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Mark sd_snapshot_delete() lossage FIXME
Markus Armbruster [Mon, 6 Mar 2017 19:00:38 +0000 (20:00 +0100)] 
sheepdog: Mark sd_snapshot_delete() lossage FIXME

sd_snapshot_delete() should delete the snapshot whose ID matches
@snapshot_id and whose name matches @name.  But that's not what it
does.  If @snapshot_id is a valid ID, it deletes the snapshot with
that ID, else it deletes the snapshot with that name.  It doesn't use
@name at all.  Add suitable FIXME comments, so someone who actually
knows Sheepdog can fix it.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Fix error handling sd_create()
Markus Armbruster [Mon, 6 Mar 2017 19:00:37 +0000 (20:00 +0100)] 
sheepdog: Fix error handling sd_create()

As a bdrv_create() method, sd_create() must set an error and return
negative errno on failure.  It prints the error instead of setting it
when connect_to_sdog() fails.  Fix that.

While there, return the value of connect_to_sdog() like we do
elsewhere, instead of -EIO.  No functional change, as
connect_to_sdog() returns no other error code.

Many more suspicious uses of error_report() and error_report_err()
remain in other functions.  Left for another day.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Fix error handling in sd_snapshot_delete()
Markus Armbruster [Mon, 6 Mar 2017 19:00:36 +0000 (20:00 +0100)] 
sheepdog: Fix error handling in sd_snapshot_delete()

As a bdrv_snapshot_delete() method, sd_snapshot_delete() must set an
error and return negative errno on failure.  It sometimes returns -1,
and sometimes neglects to set an error.  It also prints error messages
with error_report().  Fix all that.

Moreover, its handling of an attempt to delete a nonexistent snapshot
is wrong: it error_report()s and succeeds.  Fix it to set an error and
return -ENOENT instead.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agosheepdog: Defuse time bomb in sd_open() error handling
Markus Armbruster [Mon, 6 Mar 2017 19:00:35 +0000 (20:00 +0100)] 
sheepdog: Defuse time bomb in sd_open() error handling

When qemu_opts_absorb_qdict() fails, sd_open() closes stdin, because
sd->fd is still zero.  Fortunately, qemu_opts_absorb_qdict() can't
fail, because:

1. it only fails when qemu_opt_parse() fails, and
2. the only member of runtime_opts.desc[] is a QEMU_OPT_STRING, and
3. qemu_opt_parse() can't fail for QEMU_OPT_STRING.

Defuse this ticking time bomb by jumping behind the file descriptor
cleanup on error.

Also do that for the error paths where sd->fd is still -1.  The file
descriptor cleanup happens to do nothing then, but let's not rely on
that here.

While there, rename label out to err, because it's on the error path,
not the normal path out of the function.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
5 years agoblock: Fix error handling in bdrv_replace_in_backing_chain()
Kevin Wolf [Mon, 6 Mar 2017 15:20:51 +0000 (16:20 +0100)] 
block: Fix error handling in bdrv_replace_in_backing_chain()

When adding an Error parameter, bdrv_replace_in_backing_chain() would
become nothing more than a wrapper around change_parent_backing_link().
So make the latter public, renamed as bdrv_replace_node(), and remove
bdrv_replace_in_backing_chain().

Most of the callers just remove a node from the graph that they just
inserted, so they can use &error_abort, but completion of a mirror job
with 'replaces' set can actually fail.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Handle permission errors in change_parent_backing_link()
Kevin Wolf [Thu, 2 Mar 2017 17:43:00 +0000 (18:43 +0100)] 
block: Handle permission errors in change_parent_backing_link()

Instead of just trying to change parents by parent over to reference @to
instead of @from, and abort()ing whenever the permissions don't allow
this, do proper permission checking beforehand and pass any error to the
callers.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Ignore multiple children in bdrv_check_update_perm()
Kevin Wolf [Mon, 6 Mar 2017 14:00:13 +0000 (15:00 +0100)] 
block: Ignore multiple children in bdrv_check_update_perm()

change_parent_backing_link() will need to update multiple BdrvChild
objects at once. Checking permissions reference by reference doesn't
work because permissions need to be consistent only with all parents
moved to the new child.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Factor out bdrv_replace_child_noperm()
Kevin Wolf [Mon, 6 Mar 2017 12:45:28 +0000 (13:45 +0100)] 
block: Factor out bdrv_replace_child_noperm()

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agoblock: Factor out should_update_child()
Kevin Wolf [Wed, 1 Mar 2017 16:30:41 +0000 (17:30 +0100)] 
block: Factor out should_update_child()

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
5 years agoblock: Fix blockdev-snapshot error handling
Kevin Wolf [Thu, 2 Mar 2017 14:26:18 +0000 (15:26 +0100)] 
block: Fix blockdev-snapshot error handling

For blockdev-snapshot, external_snapshot_prepare() accepts an arbitrary
node reference at first and only checks later whether it already has a
backing file. Between those places, other errors can occur.

Therefore checking in external_snapshot_abort() whether state->new_bs
has a backing file is not sufficient to tell whether bdrv_append() was
already completed or not. Trying to undo the bdrv_append() when it
wasn't even executed is wrong.

Introduce a new boolean flag in the state to fix this.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agomirror: Fix error path for dirty bitmap creation
Kevin Wolf [Mon, 6 Mar 2017 15:12:44 +0000 (16:12 +0100)] 
mirror: Fix error path for dirty bitmap creation

mirror_top_bs must be removed from the graph again when creating the
dirty bitmap fails.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agomirror: Fix permissions for removing mirror_top_bs
Kevin Wolf [Mon, 6 Mar 2017 15:03:00 +0000 (16:03 +0100)] 
mirror: Fix permissions for removing mirror_top_bs

mirror_top_bs takes write permissions on its backing file, which can
make it impossible to attach that backing file node to another parent.
However, this is exactly what needs to be done in order to remove
mirror_top_bs from the backing chain. So give up the write permission
first.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agomirror: Fix permission problem with 'replaces'
Kevin Wolf [Thu, 2 Mar 2017 16:48:14 +0000 (17:48 +0100)] 
mirror: Fix permission problem with 'replaces'

The 'replaces' option of drive-mirror can be used to mirror a Quorum
node to a new image and then let the target image replace one of the
Quorum children. In order for this graph modification to succeed, the
mirror job needs to lift its restrictions on the target node first
before actually replacing the child.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
5 years agocommit: Fix error handling
Kevin Wolf [Fri, 3 Mar 2017 15:54:21 +0000 (16:54 +0100)] 
commit: Fix error handling

Apparently some kind of mismerge happened in commit 8dfba279, which
broke the error handling without any real reason by removing the
assignment of the return value to ret in a blk_insert_bs() call.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>