debian/patches/pve/0056-PVE-fix-aborting-multiple-CREATED-jobs-in-sequential.patch

   1 From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
   2 From: Stefan Reiter <s.reiter@proxmox.com>
   3 Date: Mon, 4 Jan 2021 14:49:14 +0100
   4 Subject: [PATCH] PVE: fix aborting multiple 'CREATED' jobs in sequential
   5  transaction
   6
   7 Deadlocks could occur in the AIO_WAIT_WHILE loop in job_finish_sync,
   8 which would wait for CREATED but not running jobs to complete, even
   9 though job_enter is a no-op in that scenario. Mark offending jobs as
  10 ABORTING immediately via job_update_rc if required.
  11
  12 Manifested itself in cancelling or failing backups with more than 2
  13 drives.
  14
  15 Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
  16 Tested-by: Mira Limbeck <m.limbeck@proxmox.com>
  17 Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
  18 ---
  19  job.c | 7 +++++++
  20  1 file changed, 7 insertions(+)
  21
  22 diff --git a/job.c b/job.c
  23 index 8f06e05fbf..05b7797e82 100644
  24 --- a/job.c
  25 +++ b/job.c
  26 @@ -1035,6 +1035,13 @@ int job_finish_sync(Job *job, void (*finish)(Job *, Error **errp), Error **errp)
  27          return -EBUSY;
  28      }
  29
  30 +    /* in a sequential transaction jobs with status CREATED can appear at time
  31 +     * of cancelling, these have not begun work so job_enter won't do anything,
  32 +     * let's ensure they are marked as ABORTING if required */
  33 +    if (job->status == JOB_STATUS_CREATED && job->txn->sequential) {
  34 +        job_update_rc(job);
  35 +    }
  36 +
  37      AIO_WAIT_WHILE(job->aio_context,
  38                     (job_enter(job), !job_is_completed(job)));
  39