Re: [Xen-devel] Xen blktap driver for Ceph RBD : Anybody wants to test ? :p

[Sylvain Munaut <s.munaut@whatever-company.com>]

Hi,

>> > > tapdisk[9180]: segfault at 7f7e3a5c8c10 ip 00007f7e387532d4 sp
>> > 00007f7e3a5c8c10 error 4 in libpthread-2.13.so[7f7e38748000+17000]
>> > > tapdisk:9180 blocked for more than 120 seconds.
>> > > tapdisk         D ffff88043fc13540     0  9180      1 0x00000000

You can try generating a core file by changing the ulimit on the running process

http://superuser.com/questions/404239/setting-ulimit-on-a-running-process

A backtrace would be useful :)


> Actually maybe not. What I was reading only applies for large number of bytes written to the pipe, and even then I got confused by the double negatives. Sorry for the noise.

Yes, as you discovered but size < PIPE_BUF, they should be atomic even
in non-blocking mode. But I could still add assert() there to make
sure it is.


I did find a bug where it could "leak" requests which may lead to
hang. But it shouldn't crash ...

Here's an (untested yet) patch in the rbd error path:


diff --git a/drivers/block-rbd.c b/drivers/block-rbd.c
index 68fbed7..ab2d2c5 100644
--- a/drivers/block-rbd.c
+++ b/drivers/block-rbd.c
@@ -560,6 +560,9 @@ err:
        if (c)
                rbd_aio_release(c);

+       list_move(&req->queue, &prv->reqs_free);
+       prv->reqs_free_count++;
+
        return rv;
 }


Cheers,

     Sylvain