The virtio device/driver (e.g., vhost-scsi or vhost-net) may hang due to
the loss of doorbell kick, e.g.,
https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg01711.html
... or due to the loss of IRQ, e.g., as fixed by linux kernel commit
fe200ae48ef5 ("genirq: Mark polled irqs and defer the
This is to implement the device event interface for virtio-blk-pci.
Signed-off-by: Dongli Zhang
---
hw/block/virtio-blk.c | 9 +
hw/virtio/virtio-blk-pci.c | 10 ++
include/hw/virtio/virtio-blk.h | 2 ++
3 files changed, 21 insertions(+)
diff --git
This is to implement the device event interface for vhost-scsi-pci.
Signed-off-by: Dongli Zhang
---
hw/scsi/vhost-scsi.c | 6 ++
hw/virtio/vhost-scsi-pci.c | 10 ++
include/hw/virtio/vhost-scsi.h | 3 +++
3 files changed, 19 insertions(+)
diff --git
This is to implement the device event interface for virtio-scsi-pci.
Signed-off-by: Dongli Zhang
---
hw/scsi/virtio-scsi.c | 9 +
hw/virtio/virtio-scsi-pci.c | 10 ++
include/hw/virtio/virtio-scsi.h | 3 +++
3 files changed, 22 insertions(+)
diff --git
This is to introduce the helper function for virtio device to kick or
call.
Signed-off-by: Dongli Zhang
---
hw/virtio/virtio.c | 64 ++
include/hw/virtio/virtio.h | 3 ++
2 files changed, 67 insertions(+)
diff --git a/hw/virtio/virtio.c
This is to implement the device event interface for virtio-net-pci.
Signed-off-by: Dongli Zhang
---
hw/net/virtio-net.c| 9 +
hw/virtio/virtio-net-pci.c | 10 ++
include/hw/virtio/virtio-net.h | 3 +++
3 files changed, 22 insertions(+)
diff --git
The virtio device/driver (e.g., vhost-scsi or vhost-net) may hang due to
the loss of doorbell kick, e.g.,
https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg01711.html
... or due to the loss of IRQ, e.g., as fixed by linux kernel commit
fe200ae48ef5 ("genirq: Mark polled irqs and defer the
All boards calling pflash_cfi02_register() use nb_mappings=1,
which does not do any mapping:
$ git grep -wl pflash_cfi02_register hw/
hw/arm/xilinx_zynq.c
hw/block/pflash_cfi02.c
hw/lm32/lm32_boards.c
hw/ppc/ppc405_boards.c
hw/sh4/r2d.c
We can remove this now unneeded code.
The previous commit removed the mapping code from TYPE_PFLASH_CFI02.
pflash_cfi02_register() doesn't use the 'nb_mappings' argument
anymore. Simply remove it to simplify.
Signed-off-by: Philippe Mathieu-Daudé
---
include/hw/block/flash.h | 1 -
hw/arm/digic_boards.c| 1 -
hw/arm/musicpal.c
>On Thu, Mar 25, 2021 at 8:07 AM ChangLimin wrote:
>>On Wed, Mar 24, 2021 at 4:52 PM Max Reitz wrote:
>>On 22.03.21 10:25, ChangLimin wrote:
>>> For Linux 5.10/5.11, qemu write zeros to a multipath device using
>>> ioctl(fd, BLKZEROOUT, range) with cache none or directsync return -EBUSY
>>>
On Thu, 25 Mar 2021 17:43:10 +
Stefan Hajnoczi wrote:
> On Thu, Mar 25, 2021 at 01:05:16PM -0400, Michael S. Tsirkin wrote:
> > On Thu, Mar 25, 2021 at 04:07:27PM +0100, Greg Kurz wrote:
> > > Now that virtio-scsi-pci and virtio-blk-pci map 1 virtqueue per vCPU,
> > > a serious slow down may
ping. Do we want it for 6.0?
19.03.2021 13:08, Vladimir Sementsov-Ogievskiy wrote:
Look at 03 for the problem and fix. 01 is preparation and 02 is the
test.
Actually previous version of this thing is
[PATCH v2(RFC) 0/3] qcow2: fix parallel rewrite and discard
Still
[PATCH v3 0/6]
On 23.03.21 19:19, Paolo Bonzini wrote:
This series adds a few usability improvements to qemu-iotests, in
particular:
- arguments can be passed to Python unittests scripts, for example
to run only a subset of the test cases (patches 1-2)
- it is possible to do "./check --
On Thu, 25 Mar 2021 13:05:16 -0400
"Michael S. Tsirkin" wrote:
> On Thu, Mar 25, 2021 at 04:07:27PM +0100, Greg Kurz wrote:
> > Now that virtio-scsi-pci and virtio-blk-pci map 1 virtqueue per vCPU,
> > a serious slow down may be observed on setups with a big enough number
> > of vCPUs.
> >
> >
On Thu, Mar 25, 2021 at 01:05:16PM -0400, Michael S. Tsirkin wrote:
> On Thu, Mar 25, 2021 at 04:07:27PM +0100, Greg Kurz wrote:
> > Now that virtio-scsi-pci and virtio-blk-pci map 1 virtqueue per vCPU,
> > a serious slow down may be observed on setups with a big enough number
> > of vCPUs.
> >
>
On Thu, Mar 25, 2021 at 04:36:43PM +, Peter Maydell wrote:
> On Thu, 25 Mar 2021 at 16:28, Stefan Hajnoczi wrote:
> >
> > On Thu, Mar 25, 2021 at 01:17:50PM +0300, Vladimir Sementsov-Ogievskiy
> > wrote:
> > > Thanks! My path modifies migration/block-dirty-bitmap.c. qsd-jobs runs
> > >
On Thu, Mar 25, 2021 at 04:07:27PM +0100, Greg Kurz wrote:
> Now that virtio-scsi-pci and virtio-blk-pci map 1 virtqueue per vCPU,
> a serious slow down may be observed on setups with a big enough number
> of vCPUs.
>
> Exemple with a pseries guest on a bi-POWER9 socket system (128 HW threads):
>
On 25/03/21 12:12, Olaf Hering wrote:
Am Mon, 22 Mar 2021 18:09:17 -0400
schrieb John Snow :
My understanding is that XEN has some extra disks that it unplugs when
it later figures out it doesn't need them. How exactly this works is
something I've not looked into too closely.
It has no extra
On Thu, 25 Mar 2021 at 16:28, Stefan Hajnoczi wrote:
>
> On Thu, Mar 25, 2021 at 01:17:50PM +0300, Vladimir Sementsov-Ogievskiy wrote:
> > Thanks! My path modifies migration/block-dirty-bitmap.c. qsd-jobs runs
> > block-commit and block-stream jobs and don't start any kind of migration or
> >
On 3/24/21 1:54 PM, Philippe Mathieu-Daudé wrote:
+for (i = 0; i < ARRAY_SIZE(d->bus); i++) {
+ide_bus_new(>bus[i], sizeof(d->bus[i]), ds, i, MAX_IDE_DEVS);
I bet nothing good happens if this value is ever not 2, but I bet that's
no worse than the current reality. :)
ACK
On Thu, Mar 25, 2021 at 01:17:50PM +0300, Vladimir Sementsov-Ogievskiy wrote:
> 25.03.2021 12:56, Stefan Hajnoczi wrote:
> > On Wed, Mar 24, 2021 at 08:42:27PM +, Peter Maydell wrote:
> > > On Wed, 24 Mar 2021 at 20:18, Vladimir Sementsov-Ogievskiy
> > > wrote:
> > > >
> > > > 24.03.2021
On 3/24/21 1:54 PM, Philippe Mathieu-Daudé wrote:
To avoid abusing isa_get_irq(NULL) using a hidden ISA bridge
under the hood, let the IDE function expose 2 output IRQs,
and connect them to the ISA function inputs when creating
the south bridge chipset model in vt82c686b_southbridge_init.
On Thu, Mar 25, 2021 at 8:07 AM ChangLimin wrote:
> >On Wed, Mar 24, 2021 at 4:52 PM Max Reitz wrote:
> >On 22.03.21 10:25, ChangLimin wrote:
> >> For Linux 5.10/5.11, qemu write zeros to a multipath device using
> >> ioctl(fd, BLKZEROOUT, range) with cache none or directsync return -EBUSY
> >>
On 18.03.21 21:09, Connor Kuehl wrote:
Oops, sorry about the churn. I can see why this would have caused a
failure but I'm surprised I can't reproduce this when I run the test
locally.
Christian, would you be willing to test this patch out as a quick sanity
check too?
Connor Kuehl (1):
It is useful to use different connect/disconnect event handlers
on device initialization and operation as seen from the further
commit fixing a bug on device initialization.
This patch refactors the code to make use of them: we don't rely any
more on the VM state for choosing how to cleanup the
This allows the virtio-blk-pci device to batch additions and deletions
of host notifiers. This significantly improves boot time of VMs with a
high number of vCPUs, e.g. from 3m26.408s down to 0m59.923s for a pseries
machine with 384 vCPUs.
Signed-off-by: Greg Kurz
---
Commit 4bcad76f4c39 ("vhost-user-blk: delay vhost_user_blk_disconnect")
introduced postponing vhost_dev cleanup aiming to eliminate qemu aborts
because of connection problems with vhost-blk daemon.
However, it introdues a new problem. Now, any communication errors
during execution of
This allows the virtio-scsi-pci device to batch additions and deletions
of host notifiers. This significantly improves boot time of VMs with a
high number of vCPUs, e.g. from 6m13.969s down to 1m4.268s for a pseries
machine with 384 vCPUs.
Signed-off-by: Greg Kurz
---
v3:
* 0003: a new patch added fixing the problem on vm shutdown
I stumbled on this bug after v2 sending.
* 0001: gramma fixing (Raphael)
* 0002: commit message fixing (Raphael)
v2:
* split the initial patch into two (Raphael)
* rename init to realized (Raphael)
* remove unrelated
Introduce VirtioBusClass methods to begin and commit a transaction
of setting/unsetting host notifiers. These handlers will be implemented
by virtio-pci to batch addition and deletion of ioeventfds for multiqueue
devices like virtio-scsi-pci or virtio-blk-pci.
Convert
Qemu crashes on shutdown if the chardev used by vhost-user-blk has been
finalized before the vhost-user-blk.
This happens with char-socket chardev operating in the listening mode (server).
The char-socket chardev emits "close" event at the end of finalizing when
its internal data is destroyed.
Host notifiers are guaranteed to be idle until the callbacks are
hooked up with virtio_queue_aio_set_host_notifier_handler(). They
thus don't need to be set or unset with the AioContext lock held.
Do this outside the critical section, like virtio-blk already
does : basically splitting
When dataplane multiqueue support was added in QEMU 2.7, the path
that would rollback guest notifiers assignment in case of error
simply got dropped.
Later on, when Error was added to blk_set_aio_context() in QEMU 4.1,
another error path was introduced, but it ommits to rollback both
host and
Implement the ioeventfd_assign_begin() and ioeventfd_assign_commit()
handlers of VirtioBusClass. Basically track that a transaction was
already requested by the device and use this information to prevent
the memory code to generate a transaction for each individual eventfd.
Devices that want to
Multiqueue devices such as virtio-scsi or virtio-blk, all open-code the
same pattern to setup/tear down host notifiers of the request virtqueues.
Consolidate the pattern in a new virtio_bus_set_host_notifiers() API.
Since virtio-scsi and virtio-blk both fallback to userspace if host
notifiers
Now that virtio-scsi-pci and virtio-blk-pci map 1 virtqueue per vCPU,
a serious slow down may be observed on setups with a big enough number
of vCPUs.
Exemple with a pseries guest on a bi-POWER9 socket system (128 HW threads):
1 0m20.922s 0m21.346s
2 0m21.230s
Each addition or deletion of an eventfd happens in its own MR
transaction. This doesn't scale well with multiqueue devices
that do 1:1 queue:vCPU mapping (e.g. virtio-scsi-pci or
virtio-blk-pci) : these devices typically create at least one
eventfd per queue and memory_region_transaction_commit(),
On 24.03.21 09:43, Tao Xu wrote:
There is an typo in iotest 051, correct it.
Signed-off-by: Tao Xu
---
tests/qemu-iotests/051| 2 +-
tests/qemu-iotests/051.pc.out | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
Thanks, applied to my block branch:
On 3/25/21 1:29 PM, Richard Henderson wrote:
> On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
>> To avoid abusing isa_get_irq(NULL) using a hidden ISA bridge
>> under the hood, let the IDE function expose 2 output IRQs,
>> and connect them to the ISA function inputs when creating
>> the south
On 3/25/21 6:09 AM, Philippe Mathieu-Daudé wrote:
When no mapping is requested, it is pointless to create
alias regions.
Only create them when multiple mappings are requested to
simplify the memory layout. The flatview is not changed.
For example using 'qemu-system-sh4 -M r2d -S -monitor
On 3/25/21 6:09 AM, Philippe Mathieu-Daudé wrote:
The ROMD mode isn't related to mapping setup.
Ideally we'd set this mode when the state machine resets,
but for now simply move it to pflash_cfi02_realize() to
not introduce logical change.
Signed-off-by: Philippe Mathieu-Daudé
---
Hi,
I'm using KVM from command line to run a VM and I have to create a ivshmem
between host and guest. The options that I pass are:
-device ivshmem-plain, memdev=id -object
memory-backend-file,size=1M,share,mem-path=/dev/shm/ivshmem,id=id
Now, from host side I can read and write the shmem. From
On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
To avoid abusing isa_get_irq(NULL) using a hidden ISA bridge
under the hood, let the IDE function expose 2 output IRQs,
and connect them to the ISA function inputs when creating
the south bridge chipset model in vt82c686b_southbridge_init.
On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
The 2 cascaded 8259 PIC are managed by the PCI function #0
(ISA bridge). Expose the 16 IRQs on this function, so other
functions from the same chipset can access them.
Signed-off-by: Philippe Mathieu-Daudé
---
hw/isa/vt82c686.c | 13
On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
Signed-off-by: Philippe Mathieu-Daudé
---
hw/ide/via.c | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
Reviewed-by: Richard Henderson
r~
On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
Instead of creating an input IRQ with qemu_allocate_irqs()
to pass it as output IRQ of the PIC, with its handler simply
dispatching into the "intr" output IRQ, simplify by directly
connecting the PIC to the "intr" named output.
Fixes:
When no mapping is requested, it is pointless to create
alias regions.
Only create them when multiple mappings are requested to
simplify the memory layout. The flatview is not changed.
For example using 'qemu-system-sh4 -M r2d -S -monitor stdio',
* before:
(qemu) info mtree
address-space:
The ROMD mode isn't related to mapping setup.
Ideally we'd set this mode when the state machine resets,
but for now simply move it to pflash_cfi02_realize() to
not introduce logical change.
Signed-off-by: Philippe Mathieu-Daudé
---
hw/block/pflash_cfi02.c | 2 +-
1 file changed, 1 insertion(+),
Simplify memory layout when no pflash_cfi02 mapping requested.
For example using the r2d machine:
(qemu) info mtree
address-space: memory
- (prio 0, i/o): system
-00ff (prio 0, i/o): pflash
On 3/24/21 11:54 AM, Philippe Mathieu-Daudé wrote:
Named IRQs are easier to understand in the monitor.
Name the single output interrupt as 'intr'.
Signed-off-by: Philippe Mathieu-Daudé
---
hw/isa/vt82c686.c | 2 +-
hw/mips/fuloong2e.c | 2 +-
2 files changed, 2 insertions(+), 2
Am Mon, 22 Mar 2021 18:09:17 -0400
schrieb John Snow :
> My understanding is that XEN has some extra disks that it unplugs when
> it later figures out it doesn't need them. How exactly this works is
> something I've not looked into too closely.
It has no extra disks, why would it?
I assume
25.03.2021 12:56, Stefan Hajnoczi wrote:
On Wed, Mar 24, 2021 at 08:42:27PM +, Peter Maydell wrote:
On Wed, 24 Mar 2021 at 20:18, Vladimir Sementsov-Ogievskiy
wrote:
24.03.2021 21:05, Peter Maydell wrote:
On Wed, 24 Mar 2021 at 14:52, Stefan Hajnoczi wrote:
Vladimir
On Wed, Mar 24, 2021 at 08:42:27PM +, Peter Maydell wrote:
> On Wed, 24 Mar 2021 at 20:18, Vladimir Sementsov-Ogievskiy
> wrote:
> >
> > 24.03.2021 21:05, Peter Maydell wrote:
> > > On Wed, 24 Mar 2021 at 14:52, Stefan Hajnoczi wrote:
> > >>
> > >> Vladimir Sementsov-Ogievskiy (2):
> > >>
25.03.2021 10:42, Vladimir Sementsov-Ogievskiy wrote:
25.03.2021 00:13, no-re...@patchew.org wrote:
Patchew URL:
https://patchew.org/QEMU/20210324205132.464899-1-vsement...@virtuozzo.com/
Hi,
This series seems to have some coding style problems. See output below for
more information:
25.03.2021 00:13, no-re...@patchew.org wrote:
Patchew URL:
https://patchew.org/QEMU/20210324205132.464899-1-vsement...@virtuozzo.com/
Hi,
This series seems to have some coding style problems. See output below for
more information:
Type: series
Message-id:
>On Wed, Mar 24, 2021 at 4:52 PM Max Reitz wrote:
>On 22.03.21 10:25, ChangLimin wrote:
>> For Linux 5.10/5.11, qemu write zeros to a multipath device using
>> ioctl(fd, BLKZEROOUT, range) with cache none or directsync return -EBUSY
>> permanently.
>
>So as far as I can track back the discussion,
56 matches
Mail list logo
