pend_bio uses that for
its own purposes.
- Once the bios from osd_request have been spent (added to a
blk_request), they can be NULL-ified.
I'd like hear if these are OK.
3) PSCSI needs to free the bios in pscsi_map_sg, but also needs to clear
the bios from the request in pscsi_exe
On 1/31/18 6:24 AM, Ming Lei wrote:
On Tue, Jan 30, 2018 at 04:24:14PM +0100, Jiri Palecek wrote:
On 1/30/18 1:53 PM, Ming Lei wrote:
On Thu, Jan 25, 2018 at 9:58 PM, Jiří Paleček <jpale...@web.de> wrote:
Avoids page leak from bounced requests
---
block/blk-map.c | 3 ++-
io at all under any circumstances. I believe it should stay that
way and incrementing the remaining counter, which effectively nullifies
the extra bio_endio call, does that pretty straightforwardly.
Regards
Jiri Palecek
allocate
those pages. Prior to caa4b02476e3, it would be done by bio_endio, which
is IMHO the correct way to do it.
2) What if the bio bounces and goes into the splitting path of
__blk_queue_bounce. That would submit the request, which would be a
problem should this test or some other later fail, cancelling the whole
request.
For 1, I will send a patch. As for 2, it would be a problem for a long
time so I have a hunch it doesn't actually happen, however, I'd like to
know if it is somehow guaranteed.
Regards
Jiri Palecek