On Tue, 24 Oct 2023 at 23:45, Juan Quintela wrote:
>
> The following changes since commit a95260486aa7e78d7c7194eba65cf03311ad94ad:
>
> Merge tag 'pull-tcg-20231023' of https://gitlab.com/rth7680/qemu into
> staging (2023-10-23 14:45:46 -0700)
>
> are available in the Git repository at:
>
>
On Wed, Oct 25, 2023 at 09:41:40AM +0200, Michal Prívozník wrote:
> On 10/3/23 16:56, Andrea Bolognani wrote:
> > Andrea Bolognani (6):
> > meson: Do less when not building from git
> > meson: Move all handling of test options together
> > meson: Handle -Dtests=enabled with Clang
> >
unsubscribe
--
Alvaro
On Wed, Oct 25, 2023 at 05:27:07PM +0100, Daniel P. Berrangé wrote:
> On Wed, Oct 25, 2023 at 09:12:32AM -0700, Andrea Bolognani wrote:
> > On Fri, Oct 20, 2023 at 12:48:26PM +0200, Michal Prívozník wrote:
> > > On 10/4/23 07:58, Narayana Murty N wrote:
> > > > This patch fixes this issue by
There are some cases in which we automatically disable tests when
using Clang as the compiler. If the user has explicitly asked for
tests to be enabled, however, we should error out instead of
silently disabling things.
Signed-off-by: Andrea Bolognani
Reviewed-by: Michal Privoznik
---
Currently, passing -Dtests=disabled only disables a subset of
tests: those that are written in C and thus require compilation.
Other tests, such as the syntax-check ones and those that are
implemented as scripts, are always enabled.
There's a potentially dangerous consequence of this behavior:
Test cases that depend on duplicating fds are using fairly big
values as targets.
This works fine on Linux, where RLIMIT_NOFILE is 1024 by
default, but fails on macOS which uses 256 as the default.
Decrease the values so that they're valid across all platforms.
Signed-off-by: Andrea Bolognani
Given that this variable now controls not just whether C tests
are built, but also whether any test at all is executed, the new
name is more appropriate.
Update the description for the corresponding meson option
accordingly.
Signed-off-by: Andrea Bolognani
Reviewed-by: Michal Privoznik
---
It only makes sense to enable expensive tests when tests are
enabled. Disallow invalid configurations.
Signed-off-by: Andrea Bolognani
Reviewed-by: Michal Privoznik
---
meson.build | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/meson.build b/meson.build
index
As explained in the comment, the syntax-check machinery uses git
to figure out the list of files it should operate on, so we can
only enable it when building from git.
Despite only registering the various tests with meson in that
case, however, we unconditionally perform a bunch of preparation
This will make future patches nicer.
Note that we need to handle these somewhat late because of the
dependency on information about the compiler and the flags it
supports.
Signed-off-by: Andrea Bolognani
Reviewed-by: Michal Privoznik
---
meson.build | 57
Clang can be too aggressive at optimizations, which can end up
breaking our test suite. See f9f5ab57189b for details.
As a result of this, since 7944700b4037 we are automatically
disabling tests when Clang is used unless it supports the
-fsemantic-interposition compiler flag.
Since the version
Test pipeline: https://gitlab.com/abologna/libvirt/-/pipelines/1049326523
Changes from [v1]
* fix test suite on macOS and ensure it is run as part of the
pipeline (with the previous version it would just always fail);
* disable a couple of tests (check-html, check-html-references)
On Wed, Oct 25, 2023 at 09:12:32AM -0700, Andrea Bolognani wrote:
> On Fri, Oct 20, 2023 at 12:48:26PM +0200, Michal Prívozník wrote:
> > On 10/4/23 07:58, Narayana Murty N wrote:
> > > This patch fixes this issue by returning the size=0 for architectures
> > > other than x86 and SuperH.
> >
> >
On Fri, Oct 20, 2023 at 12:48:26PM +0200, Michal Prívozník wrote:
> On 10/4/23 07:58, Narayana Murty N wrote:
> > This patch fixes this issue by returning the size=0 for architectures
> > other than x86 and SuperH.
>
> Whoa, I had no idea that SH is still alive (an well?).
As of a few months ago,
On a Wednesday in 2023, Peter Krempa wrote:
Both tests pass a disk source definition which didn't go through the
preparation steps and thus contains only the target information that
were originally present, thus we should be using the
QEMU_BLOCK_STORAGE_SOURCE_BACKEND_PROPS_TARGET_ONLY flag.
> Sent: Tuesday, October 24, 2023 at 5:28 PM
> From: "Martin Kletzander"
> To: "daggs"
> Cc: libvir-list@redhat.com
> Subject: Re: hdd kills vm
>
> On Mon, Oct 23, 2023 at 04:59:08PM +0200, daggs wrote:
> >Greetings Martin,
> >
> >> Sent: Sunday, October 22, 2023 at 12:37 PM
> >> From: "Martin
On a Wednesday in 2023, Peter Krempa wrote:
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 29 -
src/qemu/qemu_domain.c | 2 +-
2 files changed, 17 insertions(+), 14 deletions(-)
diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
index
On a Wednesday in 2023, Peter Krempa wrote:
Certain disk config fields are mirrored between the disk and storage
source definitions. We nee
Do we?
Signed-off-by: Peter Krempa
---
src/conf/domain_conf.h | 37 -
src/conf/virconftypes.h | 39
Use the same ordering of the relevant fields as we do for the format
layer -blockdev so that later they can be refactored without test
fallout.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 2 +-
.../xml2json/dir-fat-cache.json | 6 ++---
QEMU deprecated and removed support for those protocols, but due to a
logic bug in the tests it was not caught. Remove the test cases first.
Signed-off-by: Peter Krempa
---
tests/qemublocktest.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/tests/qemublocktest.c
QEMU_BLOCK_STORAGE_SOURCE_BACKEND_PROPS_SKIP_UNMAP is no longer
referenced inside the code.
QEMU_BLOCK_STORAGE_SOURCE_BACKEND_PROPS_AUTO_READONLY is passed from
various code paths to the qemuBlockStorageSourceGetBackendProps helper,
but it's no longer used.
Both thus can be removed.
Use the qemuBlockStorageSourceAddBlockdevCommonProps helper when
formatting protocol layer both when it's used as backing for a format
node and when it's used as the effective node.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 29 +++--
1 file changed, 3
The resulting properties are identical, as the hostdev backend code
doesn't set any of the extra properties.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_command.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
index
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 29 -
src/qemu/qemu_domain.c | 2 +-
2 files changed, 17 insertions(+), 14 deletions(-)
diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
index 0f47b5b37f..41038fb994 100644
---
At this point only a single code path (for formatting -drive for legacy
SD cards) uses the 'legacy' output and that code path doesn't populate
the node name. Thus we can unify the code block and simplify the JSON
formatters.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 40
Change the return value type to 'virDomainDiskGetDetectZeroes'.
Signed-off-by: Peter Krempa
---
src/conf/domain_conf.c | 2 +-
src/conf/domain_conf.h | 2 +-
src/qemu/qemu_block.c | 4 ++--
src/qemu/qemu_command.c | 4 ++--
4 files changed, 6 insertions(+), 6 deletions(-)
diff --git
The new helper replaces qemuBlockStorageSourceGetBlockdevFormatCommonProps
and the two inline instances generating the common properties for a
blockdev layer.
The new helper is to be used for both the format layer and the storage
backing layer, thus a new parameter 'effective' switches between
Return whether a relevant cachemode was presented rather than returning
an error, so that callers can be simplified. Use the proper enum type as
argument rather than typecasting in the switch statement.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 5 +
src/qemu/qemu_command.c
Introduce a mode where the protocol layer -blockdev will be formatted
so that it can be used as the effective node (used to access data from
the device). For this new mode we'll use
qemuBlockStorageSourceAddBlockdevCommonProps.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 40
Condition handling failure of the first virTestRun was lacking the 'ret
= -1' line thus the subsequent line was taken as it's body rendering the
first invocation useless.
Signed-off-by: Peter Krempa
---
tests/qemublocktest.c | 1 +
1 file changed, 1 insertion(+)
diff --git
Use the new helper in qemuBlockStorageSourceGetBlockdevStorageSliceProps
to format the common bits.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 17 ++---
.../disk-slices.x86_64-latest.args | 6 +++---
2 files changed, 9
Second part cleans up tests and various bits and refactors the protocol
blockdev JSON configuration generator to be reusable.
This still doesn't change the configuration of the blockdevs, just
reorders some arguments.
Peter Krempa (22):
qemuxml2(argv|xml)test: Add network backed disk type='sd'
Legacy mode used to be needed for use with -drive, which was almost
completely deleted. We now have qemuxml2argvtest test cases checking a
few cases and the rest uses the modern mode only. Thus we don't need to
test the legacy mode any more.
Signed-off-by: Peter Krempa
---
tests/qemublocktest.c
The aforementioned fields in virStorageSource struct are copies of the
disk properties, but were not converted to the proper type yet.
Signed-off-by: Peter Krempa
---
src/conf/storage_source_conf.h | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git
The code was refactored to format the 'read-only' and 'auto-read-only'
flags via the common helper, so the logic determining their values can
be removed.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 36 +++-
1 file changed, 3 insertions(+), 33
Certain disk config fields are mirrored between the disk and storage
source definitions. We nee
Signed-off-by: Peter Krempa
---
src/conf/domain_conf.h | 37 -
src/conf/virconftypes.h | 39 ++-
2 files changed, 38
Both tests pass a disk source definition which didn't go through the
preparation steps and thus contains only the target information that
were originally present, thus we should be using the
QEMU_BLOCK_STORAGE_SOURCE_BACKEND_PROPS_TARGET_ONLY flag.
For the same reason
The 'auto-read-only' blockdev option is available in all supported qemu
versions so we can remove the migration hack which disabled it.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 8 ++--
src/qemu/qemu_block.h | 3 +--
src/qemu/qemu_command.c | 2 +-
Add a few examples of SD cards backed with network storage to capture
the current state as the formatter code is about to be refactored.
Signed-off-by: Peter Krempa
---
.../disk-arm-virtio-sd.aarch64-latest.args | 2 ++
tests/qemuxml2argvdata/disk-arm-virtio-sd.xml| 16
Formatting of the 'nbdkit' driven backend breaks out of the switch
statement so we don't need to have an unnecessary block and indentation
level for the case when nbdkit is not in use.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_block.c | 132 +-
1 file
'qemuBuildDriveSourceStr' used to build the legacy -drive commandline
for SD cards is the only user of qemuDiskSourceGetProps. Move the helper
directly inline.
Signed-off-by: Peter Krempa
---
src/qemu/qemu_command.c | 39 +++
1 file changed, 11 insertions(+),
On Wed, Oct 25, 2023 at 12:24 PM +0200, Martin Kletzander
wrote:
> On Wed, Oct 25, 2023 at 11:03:54AM +0200, Marc Hartmayer wrote:
>>Document the following changes:
>> + added `virsh console --resume` subcommand option
>> + improved `virsh start --console` behavior
>> + improved `virsh create
On Tue, Oct 24, 2023 at 04:28:58PM +0200, Martin Kletzander wrote:
On Mon, Oct 23, 2023 at 04:59:08PM +0200, daggs wrote:
Greetings Martin,
Sent: Sunday, October 22, 2023 at 12:37 PM
From: "Martin Kletzander"
To: "daggs"
Cc: libvir-list@redhat.com
Subject: Re: hdd kills vm
On Fri, Oct 20,
On Wed, Oct 25, 2023 at 11:03:54AM +0200, Marc Hartmayer wrote:
Document the following changes:
+ added `virsh console --resume` subcommand option
+ improved `virsh start --console` behavior
+ improved `virsh create --console` behavior
Signed-off-by: Marc Hartmayer
Reviewed-by: Martin
On Tue, Oct 24, 2023 at 09:29 AM -0700, Andrea Bolognani
wrote:
> On Tue, Oct 24, 2023 at 05:14:41PM +0200, Marc Hartmayer wrote:
>> On Tue, Oct 24, 2023 at 02:12 PM +0200, Michal Prívozník
>> wrote:
>> > On 9/28/23 17:37, Marc Hartmayer wrote:
>> >> Marc Hartmayer (3):
>> >> virsh: add
Document the following changes:
+ added `virsh console --resume` subcommand option
+ improved `virsh start --console` behavior
+ improved `virsh create --console` behavior
Signed-off-by: Marc Hartmayer
---
NEWS.rst | 15 +++
1 file changed, 15 insertions(+)
diff --git a/NEWS.rst
On 10/3/23 16:56, Andrea Bolognani wrote:
>
>
> Andrea Bolognani (6):
> meson: Do less when not building from git
> meson: Move all handling of test options together
> meson: Handle -Dtests=enabled with Clang
> meson: Make -Dexpensive_tests depend on -Dtests
> meson: Disable all tests
On 10/2/23 10:51, Andrea Bolognani wrote:
> This series addresses two pieces of feedback from my recent systemd
> changes: that some settings, such as LimitNOFILE, where still being
> repeated verbatim in multiple locations, and that only having the
> foo.{service,socket}.extra.in files for some
On 10/24/23 18:54, Andrea Bolognani wrote:
> The situation is the same as Linux: since glibc no
> longer includes the RPC functionality, libtirpc must
> be used to complement it.
>
> Signed-off-by: Andrea Bolognani
> ---
> meson.build | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
50 matches
Mail list logo