On 1/23/24 10:55 AM, Keith Busch wrote:
> On Tue, Jan 23, 2024 at 10:30:36AM -0700, Jens Axboe wrote:
>> static void sched_update_worker(struct task_struct *tsk)
>> {
>> - if (tsk->flags & (PF_WQ_WORKER | PF_IO_WORKER)) {
>> + if (tsk->flags & (PF_WQ_WORKER | PF_IO_WORKER | PF_BLOCK_TS)) {
>> + if (tsk->flags & PF_BLOCK_TS)
>> + blk_plug_invalidate_ts(tsk);
>> if (tsk->flags & PF_WQ_WORKER)
>> wq_worker_running(tsk);
>> else
>
> I don't even know what is in the 'else' case, but it doesn't look right
> now that you added another flag to enter into this block. Before this
> patch, it looks like we could assume PF_IO_WORKER was set in the 'else'
> condition, but that's not necessarily the case with this change.
Will add an else if for PF_IO_WORKER for that case, to be on the safe
side.
--
Jens Axboe
