Re: [Qemu-devel] [PATCH 07/11] blockjob: introduce block_job_cancel_async, check iostatus invariants

[Paolo Bonzini <pbonzini@redhat.com>]

On 26/04/2017 22:25, John Snow wrote:
>>  
>>  void block_job_iostatus_reset(BlockJob *job)
>>  {
>> +    if (job->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
>> +        return;
>> +    }
>> +    assert(job->user_paused && job->pause_count > 0);
> Why assert that it's user-paused? Will that be true from:
> 
> (A) All users of block_job_cancel_async, including:
> 
> - block_job_cancel
> - block_job_completed
>     block_job_completed_txn_abort
> 
> (B) all users of blk_iostatus_reset:
> - blk_do_attach_dev
> - qmp_cont
> 
> It's ... not really clear to me that this is true, can you help me out?

Setting iostatus to not-ok only happens in block_job_iostatus_set_err, 
which is called from block_job_error_action:

    if (action == BLOCK_ERROR_ACTION_STOP) {
        /* make the pause user visible, which will be resumed from QMP. */
        block_job_user_pause(job);
        block_job_iostatus_set_err(job, error);
    }

Paolo
Paolo