public inbox for pve-devel@lists.proxmox.com
 help / color / mirror / Atom feed
From: Wolfgang Bumiller <w.bumiller@proxmox.com>
To: Fabian Ebner <f.ebner@proxmox.com>
Cc: pve-devel@lists.proxmox.com
Subject: Re: [pve-devel] [PATCH qemu] fix #4101: acquire job's aio context before calling job_unref
Date: Thu, 9 Jun 2022 14:11:29 +0200	[thread overview]
Message-ID: <20220609121129.osaveyngxn2zk2xy@wobu-vie.proxmox.com> (raw)
In-Reply-To: <20220609115538.135041-1-f.ebner@proxmox.com>

minor nit

but otherwise LGTM

On Thu, Jun 09, 2022 at 01:55:38PM +0200, Fabian Ebner wrote:
> Otherwise, we might not run into an abort via bdrv_co_yield_to_drain()
> (can at least happen when a disk with iothread is used):
> > #0  0x00007fef4f5dece1 __GI_raise (libc.so.6 + 0x3bce1)
> > #1  0x00007fef4f5c8537 __GI_abort (libc.so.6 + 0x25537)
> > #2  0x00005641bce3c71f error_exit (qemu-system-x86_64 + 0x80371f)
> > #3  0x00005641bce3d02b qemu_mutex_unlock_impl (qemu-system-x86_64 + 0x80402b)
> > #4  0x00005641bcd51655 bdrv_co_yield_to_drain (qemu-system-x86_64 + 0x718655)
> > #5  0x00005641bcd52de8 bdrv_do_drained_begin (qemu-system-x86_64 + 0x719de8)
> > #6  0x00005641bcd47e07 blk_drain (qemu-system-x86_64 + 0x70ee07)
> > #7  0x00005641bcd498cd blk_unref (qemu-system-x86_64 + 0x7108cd)
> > #8  0x00005641bcd31e6f block_job_free (qemu-system-x86_64 + 0x6f8e6f)
> > #9  0x00005641bcd32d65 job_unref (qemu-system-x86_64 + 0x6f9d65)
> > #10 0x00005641bcd93b3d pvebackup_co_complete_stream (qemu-system-x86_64 + 0x75ab3d)
> > #11 0x00005641bce4e353 coroutine_trampoline (qemu-system-x86_64 + 0x815353)
> 
> Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
> ---
>  ...ensure-jobs-in-di_list-are-referenced.patch | 18 ++++++++++++------
>  ...id-segfault-issues-upon-backup-cancel.patch |  6 +++---
>  2 files changed, 15 insertions(+), 9 deletions(-)
> 
> diff --git a/debian/patches/pve/0055-PVE-Backup-ensure-jobs-in-di_list-are-referenced.patch b/debian/patches/pve/0055-PVE-Backup-ensure-jobs-in-di_list-are-referenced.patch
> index db86632..ebb7919 100644
> --- a/debian/patches/pve/0055-PVE-Backup-ensure-jobs-in-di_list-are-referenced.patch
> +++ b/debian/patches/pve/0055-PVE-Backup-ensure-jobs-in-di_list-are-referenced.patch
> @@ -17,26 +17,29 @@ freed. With unlucky timings it seems possible that:
>  Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
>  Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
>  ---
> - pve-backup.c | 13 +++++++++++++
> - 1 file changed, 13 insertions(+)
> + pve-backup.c | 19 +++++++++++++++++++
> + 1 file changed, 19 insertions(+)
>  
>  diff --git a/pve-backup.c b/pve-backup.c
> -index 5bed6f4014..cd45e66a61 100644
> +index 5bed6f4014..7b094e5018 100644
>  --- a/pve-backup.c
>  +++ b/pve-backup.c
> -@@ -316,6 +316,11 @@ static void coroutine_fn pvebackup_co_complete_stream(void *opaque)
> +@@ -316,6 +316,14 @@ static void coroutine_fn pvebackup_co_complete_stream(void *opaque)
>           }
>       }
>   
>  +    if (di->job) {
> ++        AioContext *ctx = di->job->job.aio_context;
> ++        aio_context_acquire(ctx);
>  +        job_unref(&di->job->job);
> ++        aio_context_release(ctx);
>  +        di->job = NULL;

(I think it might be nicer to have this assignment before the release
call)

>  +    }
>  +
>       // remove self from job list
>       backup_state.di_list = g_list_remove(backup_state.di_list, di);
>   
> -@@ -494,6 +499,9 @@ static void create_backup_jobs_bh(void *opaque) {
> +@@ -494,6 +502,9 @@ static void create_backup_jobs_bh(void *opaque) {
>           aio_context_release(aio_context);
>   
>           di->job = job;
> @@ -46,13 +49,16 @@ index 5bed6f4014..cd45e66a61 100644
>   
>           if (!job || local_err) {
>               error_setg(errp, "backup_job_create failed: %s",
> -@@ -528,6 +536,11 @@ static void create_backup_jobs_bh(void *opaque) {
> +@@ -528,6 +539,14 @@ static void create_backup_jobs_bh(void *opaque) {
>                   aio_context_release(ctx);
>                   canceled = true;
>               }
>  +
>  +            if (di->job) {
> ++                AioContext *ctx = di->job->job.aio_context;
> ++                aio_context_acquire(ctx);

Since now both the above concelation and this unref acquire the aio
context, we could just move the cancellation down into this `if(job)` as

    if (!canceled) {
        job_cancel_sync(...);
        canceled = true;
    }

to be a bit more concise

>  +                job_unref(&di->job->job);
> ++                aio_context_release(ctx);
>  +                di->job = NULL;

^ (and also assign before release)




      parent reply	other threads:[~2022-06-09 12:11 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-06-09 11:55 Fabian Ebner
2022-06-09 11:57 ` Fabian Ebner
2022-06-09 12:11 ` Wolfgang Bumiller [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220609121129.osaveyngxn2zk2xy@wobu-vie.proxmox.com \
    --to=w.bumiller@proxmox.com \
    --cc=f.ebner@proxmox.com \
    --cc=pve-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal