On Sun, Oct 30 2005, Tejun Heo wrote:
> Tejun Heo wrote:
> >Hello, Paul.
> >
> >Paul Collins wrote:
> >
> >>I boot with elevator=cfq (wanted to try the ionice stuff, never got
> >>around to it). Having decided to go back to the anticipatory
> >>scheduler, I did the following:
> >>
> >># echo anticipatory > /sys/block/hda/queue/scheduler
> >># echo anticipatory > /sys/block/hdc/queue/scheduler
> >>
> >>A while later I did 'sudo snooze', which produced the Oops below.
> >>
> >>Booting with elevator=as and then changing to cfq, sleep works fine.
> >>But if I resume and change back to anticipatory I get a similar Oops
> >>on the next 'sudo snooze'.
> >>
> >>
> >> Oops: kernel access of bad area, sig: 11 [#1]
> >> NIP: C01E1948 LR: C01D6A60 SP: EFBC5C20 REGS: efbc5b70 TRAP: 0300
> >>Not tainted
> >> MSR: 00001032 EE: 0 PR: 0 FP: 0 ME: 1 IR/DR: 11
> >> DAR: 00000020, DSISR: 40000000
> >> TASK = efb012c0[1213] 'pmud' THREAD: efbc4000
> >> Last syscall: 54 GPR00: 00080000 EFBC5C20 EFB012C0 EFE9E044
> >>EFBC5CE8 00000002 00000000 C03B0000 GPR08: C046E5D8 00000000
> >>C03B47C8 E6A58360 22042422 1001E4DC 10010000 10000000 GPR16:
> >>10000000 10000000 10000000 7FE4EB40 10000000 10000000 10010000
> >>C0400000 GPR24: C0380000 00000002 00000002 C046E0C0 00000000
> >>00000002 00000000 EFBC5CE8 NIP [c01e1948] as_insert_request+0xa8/0x6b0
> >> LR [c01d6a60] __elv_add_request+0xa0/0x100
> >> Call trace:
> >> [c01d6a60] __elv_add_request+0xa0/0x100
> >> [c01ffb84] ide_do_drive_cmd+0xb4/0x190
> >> [c01fc1c0] generic_ide_suspend+0x80/0xa0
> >> [c01d4574] suspend_device+0x104/0x160
> >> [c01d47c0] device_suspend+0x120/0x330
> >> [c03f3b50] pmac_suspend_devices+0x50/0x1b0
> >> [c03f4294] pmu_ioctl+0x344/0x9b0
> >> [c0082aa4] do_ioctl+0x84/0x90
> >> [c0082b3c] vfs_ioctl+0x8c/0x460
> >> [c0082f50] sys_ioctl+0x40/0x80
> >> [c0004850] ret_from_syscall+0x0/0x4c
> >>
> >
> >Can you please post assembly of as_insert_request? You can get this by
> >doing 'objdump -d drivers/block/as-iosched.o | less' and copy & pasting
> >as_insert_request part.
> >
> >I'm also trying to reproduce the oops but haven't succeeded yet. Does
> >the oops occur only if the disk is loaded while switching scheduler /
> >snoozing or does it happen regardless of disk load?
> >
> >And one more thing. Can you please try the following program and see if
> >it causes the oops? The program simply writes 3, sleeps one second and
> >then writes 0. When redirected to the disk's power/state sysfs node, it
> >will make the disk sleep for 1 second and then wake it up.
> >
> >#include <stdio.h>
> >#include <stdlib.h>
> >
> >int main(int argc, char **argv)
> >{
> > int level = 3;
> > if (argc > 1)
> > level = atoi(argv[1]);
> > setvbuf(stdout, NULL, _IONBF, 0);
> > printf("%d", level);
> > sleep(1);
> > printf("0");
> > return 0;
> >}
> >
> >After compiling, do the following.
> >
> >./a.out > /sys/block/hd?/device/power/state
> >
> >Thanks. :-)
> >
>
> Oh.. Never mind. Jens seems to know how to fix it. :-)
This should fix it.
-------
Don't clear ->elevator_data on exit, if we are switching queues we are
overwriting the data of the new io scheduler.
Signed-off-by: Jens Axboe <[email protected]>
diff --git a/drivers/block/cfq-iosched.c b/drivers/block/cfq-iosched.c
index cd056e7..23df2d2 100644
--- a/drivers/block/cfq-iosched.c
+++ b/drivers/block/cfq-iosched.c
@@ -2260,10 +2260,8 @@ static void cfq_put_cfqd(struct cfq_data
if (!atomic_dec_and_test(&cfqd->ref))
return;
- blk_put_queue(q);
-
cfq_shutdown_timer_wq(cfqd);
- q->elevator->elevator_data = NULL;
+ blk_put_queue(q);
mempool_destroy(cfqd->crq_pool);
kfree(cfqd->crq_hash);
--
Jens Axboe
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]