Hello Nab,
> The first is specific to FILEIO, where the size of the submitted I/O
> (512) for COMPARE_AND_WRITE is different from the fabric received
> cmd->data_length (1024).
I see.
> So the (ret != cmd->data_length) checks should either be aware of the
> difference for the special COMPARE_AND_WRITE case, or cmd->data_length
> should be reset to the different value before the initial READ I/O
> submission.
Got it.
> The second appears to be some manner of double completion in the failure
> path for the WRITE payload submission of COMPARE_AND_WRITE.
I see. Wow. I was really off with my wild guesses. I'm eager to see
your patches and try them out.
> In any event, I'm tracking these down now, and should be posting a
> bugfix series later this afternoon.
Perfect. Please give me a ping, I'll pull, rebuild the kernel and report
back ASAP.
Cheers,
Thomas
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
