Hi,
2012-01-10 1:41 AM, Tejun Heo wrote:
Hello,
On Mon, Jan 09, 2012 at 11:57:43AM +0900, Namhyung Kim wrote:
I understand your concerns. However, the blktrace cannot get
bio->bi_flags info in its current form AFAIK. Doing it will require
extending struct blk_io_trace and it'll cause a compatibility issue,
I guess.
Umm? Why can't blk_add_trace_bio_complete() look at the flags (or
whatever other states) to decide to fire off BLK_TA_COMPLETE or not?
What's the difference? No userland visible change is necessary at
all. Just make blktrace.c do the right thing. Am I missing
something?
Thanks.
Oh I misunderstood what you said. I was thinking about filtering in pure
userspace, but you meant in-kernel probe side.
Right, we can change the probe to filter BIO_BOUNCED case out. But IMHO
BIO_CLONED is different, since it usually routed to another device as a
separate IO request. I'll cook a patch for the former soon.
Thanks for the comment.
Namhyung Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
