Although QEMU virtio-blk is quite fast, there is still some room for
improvements. Disk latency can be reduced if we handle virito-blk requests
in host kernel so we avoid a lot of syscalls and context switches.
The idea is quite simple - QEMU gives us block device and we translate
any incoming virt
Finally add ioctl to allow userspace to create additional workers
For now only allow to increase the number of workers
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
v2:
- Make code do what docs suggest. Previously ioctl-supplied new number
of workers were treated lik
We want to support several vhost workers. The first step is to
rework vhost to use array of workers rather than single pointer.
Update creation and cleanup routines.
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
v3:
set vq->worker to NULL in vhost_vq_reset()
this will
Allow vhost polls to be associated with vqs so we can queue them
on assigned workers.
If polls are not associated with specific vqs queue them on the first
worker.
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
drivers/vhost/vhost.c | 24
driv
Add worker pointer to every virtqueue. Add routine to assing
workers to virtqueues and call it after any worker creation
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
v4:
Set vq->worker = NULL in vhost_vq_reset. This will fix both
https://jira.sw.ru/browse/PSBM-14205
Add routines to queue works on virtqueue assigned workers
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
drivers/vhost/vhost.c | 22 ++
drivers/vhost/vhost.h | 5 +
2 files changed, 27 insertions(+)
diff --git a/drivers/vhost/vhost.c b/driver
Although QEMU virtio is quite fast, there is still some room for
improvements. Disk latency can be reduced if we handle virito-blk requests
in host kernel istead of passing them to QEMU. The patch adds vhost-blk
kernel module to do so.
Some test setups:
fio --direct=1 --rw=randread --bs=4k --ioe
Rework vhost_attach_cgroups to manipulate specified worker.
Implement vhost_worker_flush as we need to flush specific worker.
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
drivers/vhost/vhost.c | 18 ++
1 file changed, 14 insertions(+), 4 deletions(-)
Update vhost_blk to queue works on virtqueue workers. Together
with previous changes this allows us to split virtio blk requests
across several threads.
| randread, IOPS | randwrite, IOPS |
8vcpu, 1 kernel worker | 576k | 575k |
8vcpu, 2 kernel workers
Make vhost_work_dev_flush support several workers and flush
them simultaneously
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
v2:
- don't bother with checking dev->workers[0].worker since dev->nworkers
will always contain 0 in this case
drivers/vhost/vhost.c | 16 +
Add function to create a vhost worker and add it into the device.
Rework vhost_dev_set_owner
https://jira.sw.ru/browse/PSBM-139414
Signed-off-by: Andrey Zhadchenko
---
drivers/vhost/vhost.c | 68 +--
1 file changed, 40 insertions(+), 28 deletions(-)
diff
On 31.10.22 21:27, Denis V. Lunev wrote:
On 10/31/22 19:12, Alexander Atanasov wrote:
Initially in commit 80da9c2abac9 ("ploop: add a standby mode") a flag
on the block device's request queue was added to put the queue
into standby mode on EBUSY. Later on, the list with errors was extended
in co
ploop uses pr_info/pr_err for logging but it does not
print which device the message is about.
Add device name to log messages to make debugging easier.
Messages are in the following format:
ploop: dm-47274: log message here
Where 47274 is the device minor number which corresponds
to /dev/mappe
On 01.11.2022 08:36, Alexander Atanasov wrote:
On 31.10.22 21:27, Denis V. Lunev wrote:
On 10/31/22 19:12, Alexander Atanasov wrote:
[snip]
@@ -1163,6 +1166,26 @@ static void ploop_queue_resubmit(struct pio *pio)
queue_work(ploop->wq, &ploop->worker);
}
+static void ploop_check_sta
LGTM.
BTW, the dm-qcow2 driver also has some pr_err() calls,
do you plan to enhance them as well?
On 01.11.2022 12:33, Alexander Atanasov wrote:
ploop uses pr_info/pr_err for logging but it does not
print which device the message is about.
Add device name to log messages to make debugging eas
On 1.11.22 17:29, Konstantin Khorenko wrote:
LGTM.
BTW, the dm-qcow2 driver also has some pr_err() calls,
do you plan to enhance them as well?
I will create an issue for dm-qcow2 to do the same.
--
Regards,
Alexander Atanasov
___
Devel mailing lis
On 1.11.22 17:27, Konstantin Khorenko wrote:
On 01.11.2022 08:36, Alexander Atanasov wrote:
On 31.10.22 21:27, Denis V. Lunev wrote:
On 10/31/22 19:12, Alexander Atanasov wrote:
[snip]
@@ -1163,6 +1166,26 @@ static void ploop_queue_resubmit(struct pio
*pio)
queue_work(ploop->wq, &pl
-Original Message-
From: Alexander Atanasov
Date: Tuesday, 1 November 2022 at 11:57 PM
To: Konstantin Khorenko , "Denis V. Lunev"
, Devel
Cc: Kui Liu , Alexey Kuznetsov
Subject: Re: [PATCH vz9 v2] ploop: port and fix the standby mode feature.
On 1.11.22 17:27, Konstantin Khorenk
18 matches
Mail list logo
