On 13/09/21 7:33 pm, Bart Van Assche wrote:

On 9/13/21 1:53 AM, Adrian Hunter wrote:

quoted

scsi_dec_host_busy() is called for any non-zero return value like
SCSI_MLQUEUE_HOST_BUSY:

i.e.
    reason = scsi_dispatch_cmd(cmd);
    if (reason) {
        scsi_set_blocked(cmd, reason);
        ret = BLK_STS_RESOURCE;
        goto out_dec_host_busy;
    }

    return BLK_STS_OK;

out_dec_host_busy:
    scsi_dec_host_busy(shost, cmd);

And that will wake the error handler:

static void scsi_dec_host_busy(struct Scsi_Host *shost, struct scsi_cmnd *cmd)
{
    unsigned long flags;

    rcu_read_lock();
    __clear_bit(SCMD_STATE_INFLIGHT, &cmd->state);
    if (unlikely(scsi_host_in_recovery(shost))) {
        spin_lock_irqsave(shost->host_lock, flags);
        if (shost->host_failed || shost->host_eh_scheduled)
            scsi_eh_wakeup(shost);
        spin_unlock_irqrestore(shost->host_lock, flags);
    }
    rcu_read_unlock();
}

Returning SCSI_MLQUEUE_HOST_BUSY is not sufficient to wake up the SCSI
error handler because of the following test in scsi_error_handler():

    shost->host_failed != scsi_host_busy(shost)

SCSI_MLQUEUE_HOST_BUSY causes scsi_host_busy() to decrement by calling
scsi_dec_host_busy() as described above, so the request is not being
counted in that condition anymore.

As I mentioned in a previous email, all pending commands must have failed
or timed out before the error handler is woken up. Returning
SCSI_MLQUEUE_HOST_BUSY from ufshcd_queuecommand() does not fail a command
and prevents it from timing out. Hence my suggestion to change
"return SCSI_MLQUEUE_HOST_BUSY" into set_host_byte(cmd, DID_IMM_RETRY)
followed by cmd->scsi_done(cmd). A possible alternative is to move the
blk_mq_start_request() call in the SCSI core such that the block layer
request timer is not reset if a SCSI LLD returns SCSI_MLQUEUE_HOST_BUSY.

Bart.