On Sat, 1 Apr 2000, Andrew Morton wrote:
>A week of two back I was discussing a couple of elevator hiccups with
>Andrea. I have now found a way to reproduce them. /bin/sync.
>
>mnm:/usr/src/linux> alias make="make -j3"
>mnm:/usr/src/linux> while true ; do sync ; sleep 15 ; done &
>mnm:/usr/src/linux> while true ; do make clean && make dep && make
>bzImage ; done
Good.
>After a few minutes this will hit the first BUG() in
For other developers, this is a full trace from earlier email:
21:05: elevator req->cmd 99 req->nr_segments 1074483200 req->q 00000000
21:05: elevator req->q NULL req->nr_segments 1074483200
21:05: elevator req->cmd 99 req->nr_segments 1074483200 req->q 00000000
21:05: elevator req->q NULL req->nr_segments 1074483200
21:05: elevator req->cmd 99 req->nr_segments 1074483200 req->q 00000000
21:05: elevator req->q NULL req->nr_segments 1074483200
According to the above trace the device where you are getting these is
/dev/sg5.
Also in previous email you said you was getting this other trace:
> 16:06: elevator req->cmd 99 req->nr_segments 3276455432 req->q 00000000
> 16:06: elevator req->q NULL req->nr_segments 3276455432
And the previous one (as said in earlier email) made no sense since 16:06
should be the sixth gscd and you said you don't have such device even
compiled into the kernel.
For SCSI devices the only request in the request_queue that can have
req->q NULL gets inserted by scsi_insert_special_cmd and
scsi_insert_special_req (they are special commands) and before inserting
it we _always_ set nr_segment to zero too. It's kind of strange problem.
IMHO you are getting someway some memory corruption (so maybe you are not
even using /dev/sg5?). Do you have/ever had fs corruption problems or
everything remains rock solid?
Which blockdevice are you using to generate such messages?
>I first observed this problem on a uniprocessor machine running an SMP
>2.3.99-pre1 kernel (with make -j0). I am reproducing it on a dual
>Celeron 2.3.99-pre4-1.
Before 2.3.99-pre1 there wasn't such debugging check, so it's normal you
couldn't see that message before.
Also which compiler are you using? The recommended are egcs-1.1.2 or gcc
2.95.2.
Andrea
