Re: [PATCH v4 4/4] scsi: ufs: inject errors to verify error handling

[Gilad Broner]

>> +static bool inject_cmd_hang_tr(struct ufs_hba *hba)
>> +{
>> +       int tag;
>> +
>> +       tag = find_first_bit(&hba->outstanding_reqs, hba->nutrs);
>> +       if (tag == hba->nutrs)
>> +               return 0;
>> +
>> +       __clear_bit(tag, &hba->outstanding_reqs);
>> +       hba->lrb[tag].cmd = NULL;
>> +       __clear_bit(tag, &hba->lrb_in_use);
>
> hba->lrb_in_use is a bitmap set by test_and_set_bit_lock().  So
> this should be cleared by clear_bit_unlock().

You are correct. Thanks.

>
> And as soon as the bit corresponds to this slot in hba->lrb_in_use is
> cleared, this slot could be reused.  But if the request corresponds
> to the slot is not completed yet, it could sacrifice the new request,
> too.  So should we only inject into the commands which have been
> completed like this?

Please note that we only clear the bit in hba->lrb_in_use. scsi_done is
not called for this request. Therefore, the tag is not yet free in the
block layer and next calls for queuecommand will not pass down this tag to
be used in the UFS driver. So there is no danger of a new request being
sacrificed.

On a different note, we are debating internally on a few other changes so
until we consolidate those I will drop this patch with error injection.

Gilad.

-- 
Qualcomm Israel, on behalf of Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/