Re: -mm: xfs lockdep warning

From: Torsten Kaiser
Date: Thu Oct 07 2010 - 09:21:46 EST

Next message: Kumar Gala: "Re: [PATCH] edac: mpc85xx: Add support for new MPCxxx/Pxxxx EDAC controllers (fix)"
Previous message: David Howells: "Re: linux-next: manual merge of the irqflags tree with Linus' tree"
In reply to: Tejun Heo: "Re: -mm: xfs lockdep warning"
Next in thread: Tejun Heo: "Re: -mm: xfs lockdep warning"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Thu, Oct 7, 2010 at 1:14 PM, Tejun Heo <tj@xxxxxxxxxx> wrote:
> (restoring cc to lkml)
> Hello,
>
> On 10/06/2010 08:32 PM, Tejun Heo wrote:
>> On 10/06/2010 07:20 PM, Torsten Kaiser wrote:
>>>> It seems the system isn't completely stuck yet. The above process is
>>>> still trying to free memory. Maybe memory reclaim is just extremely
>>>> slow for some reason?
>>>
>>> How can I check this? With the hund -rc3 I repeatedly hit SysRq+M
>>> every time waiting 10..60 seconds.
>>> Can you see from these outputs if reclaim ins making any progress?
>>
>> I think I have an idea of what's going on. I'll prep a patch and send
>> it to you tomorrow.
>
> Can you test whether the following patch fix the issue?

Yes. After adding this patch to 2.6.36-rc6 it survived my testcase. It
first seems to get stuck again, but after less then a minute the
system recovered and all compiles finished successfully.

I suspect the combination of CFLAGs, the gcc version and maybe even
changes in KOffice from 2.2.1 to 2.2.2 are import for my testcase. It
looks like at the start of the complies something is really eating an
enormous amount of memory and I did not hit such a case with earlier
kernels.

output from vmstat 60 during the start of the testcase:
1 0 0 2442196 1060 698180 0 0 242 19 6847 15039 26 7 66 1
12 1 0 1786212 1060 1040988 0 0 1248 15 6401 21721 56 20 19 5
18 0 0 787880 1060 1097136 0 0 270 9 742 1614 72 28 0 0
18 0 0 322680 1060 1110036 0 0 47 4 611 1234 71 29 0 0
18 1 161208 975064 96 412960 38 2709 757 2722 1616 2958 79 21 0 0
27 28 1000036 14728 0 135520 12 13991 503 14023 1808 3781 25 8 0 66
18 4 1296652 815832 0 131852 1824 8245 3268 8266 2037 4365 42 16 0 42
21 0 857012 903888 0 185292 1351 0 1810 65 846 1678 87 13 0 0
16 0 817320 1890752 0 217404 102 0 275 18 530 1142 89 11 0 0
11 0 652620 2441520 0 270528 1132 0 1424 38 934 2783 79 18 1 2
11 1 636148 1629492 0 318020 43 0 413 144 613 1417 84 16 0 0

As the system recovered from that, I think your patch is the correct
solution for my problem.
I will run my testcase later again, but if you do not hear from me, I
was not able to get it stuck with your patch.

Thanks for looking into this and providing this patch!

Torsten

> diff --git a/block/blk-core.c b/block/blk-core.c
> index 32a1c12..622602b 100644
> --- a/block/blk-core.c
> +++ b/block/blk-core.c
> @@ -2584,7 +2584,7 @@ int __init blk_dev_init(void)
> BUILD_BUG_ON(__REQ_NR_BITS > 8 *
> sizeof(((struct request *)0)->cmd_flags));
>
> - kblockd_workqueue = create_workqueue("kblockd");
> + kblockd_workqueue = alloc_workqueue("kblockd", WQ_MEM_RECLAIM, 1);
> if (!kblockd_workqueue)
> panic("Failed to create kblockd\n");
>
> diff --git a/crypto/crypto_wq.c b/crypto/crypto_wq.c
> index fdcf624..4c893a1 100644
> --- a/crypto/crypto_wq.c
> +++ b/crypto/crypto_wq.c
> @@ -20,7 +20,7 @@ EXPORT_SYMBOL_GPL(kcrypto_wq);
>
> static int __init crypto_wq_init(void)
> {
> - kcrypto_wq = create_workqueue("crypto");
> + kcrypto_wq = alloc_workqueue("crypto", WQ_MEM_RECLAIM, 1);
> if (unlikely(!kcrypto_wq))
> return -ENOMEM;
> return 0;
> diff --git a/crypto/pcrypt.c b/crypto/pcrypt.c
> index de30782..2ee6ba2 100644
> --- a/crypto/pcrypt.c
> +++ b/crypto/pcrypt.c
> @@ -455,7 +455,7 @@ static int pcrypt_init_padata(struct padata_pcrypt *pcrypt,
>
> get_online_cpus();
>
> - pcrypt->wq = create_workqueue(name);
> + pcrypt->wq = alloc_workqueue(name, WQ_MEM_RECLAIM, 1);
> if (!pcrypt->wq)
> goto err;
>
> diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
> index 64e0903..2631930 100644
> --- a/drivers/infiniband/core/cm.c
> +++ b/drivers/infiniband/core/cm.c
> @@ -3804,7 +3804,7 @@ static int __init ib_cm_init(void)
> if (ret)
> return -ENOMEM;
>
> - cm.wq = create_workqueue("ib_cm");
> + cm.wq = alloc_workqueue("ib_cm", WQ_MEM_RECLAIM, 1);
> if (!cm.wq) {
> ret = -ENOMEM;
> goto error1;
> diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c
> index f1d16d3..0f559f2 100644
> --- a/drivers/infiniband/hw/qib/qib_init.c
> +++ b/drivers/infiniband/hw/qib/qib_init.c
> @@ -1053,7 +1053,7 @@ static int __init qlogic_ib_init(void)
> * so flush_scheduled_work() can deadlock during device
> * removal.
> */
> - qib_wq = create_workqueue("qib");
> + qib_wq = alloc_workqueue("qib", WQ_MEM_RECLAIM, 1);
> if (!qib_wq) {
> ret = -ENOMEM;
> goto bail_dev;
> diff --git a/drivers/md/dm-delay.c b/drivers/md/dm-delay.c
> index baa1191..1f8d1e4 100644
> --- a/drivers/md/dm-delay.c
> +++ b/drivers/md/dm-delay.c
> @@ -352,7 +352,7 @@ static int __init dm_delay_init(void)
> {
> int r = -ENOMEM;
>
> - kdelayd_wq = create_workqueue("kdelayd");
> + kdelayd_wq = alloc_workqueue("kdelayd", WQ_MEM_RECLAIM, 1);
> if (!kdelayd_wq) {
> DMERR("Couldn't start kdelayd");
> goto bad_queue;
> diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
> index 487ecda..e60f22d 100644
> --- a/drivers/md/dm-mpath.c
> +++ b/drivers/md/dm-mpath.c
> @@ -1687,7 +1687,7 @@ static int __init dm_multipath_init(void)
> return -EINVAL;
> }
>
> - kmultipathd = create_workqueue("kmpathd");
> + kmultipathd = alloc_workqueue("kmpathd", WQ_MEM_RECLAIM, 1);
> if (!kmultipathd) {
> DMERR("failed to create workqueue kmpathd");
> dm_unregister_target(&multipath_target);
> diff --git a/drivers/message/i2o/driver.c b/drivers/message/i2o/driver.c
> index a0421ef..8a5b2d8 100644
> --- a/drivers/message/i2o/driver.c
> +++ b/drivers/message/i2o/driver.c
> @@ -84,7 +84,8 @@ int i2o_driver_register(struct i2o_driver *drv)
> osm_debug("Register driver %s\n", drv->name);
>
> if (drv->event) {
> - drv->event_queue = create_workqueue(drv->name);
> + drv->event_queue = alloc_workqueue(drv->name,
> + WQ_MEM_RECLAIM, 1);
> if (!drv->event_queue) {
> osm_err("Could not initialize event queue for driver "
> "%s\n", drv->name);
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 1e4bff6..a6dd94d 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -351,7 +351,7 @@ static int qla25xx_setup_mode(struct scsi_qla_host *vha)
> "Can't create request queue\n");
> goto fail;
> }
> - ha->wq = create_workqueue("qla2xxx_wq");
> + ha->wq = alloc_workqueue("qla2xxx_wq", WQ_MEM_RECLAIM, 1);
> vha->req = ha->req_q_map[req];
> options |= BIT_1;
> for (ques = 1; ques < ha->max_rsp_queues; ques++) {
> diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
> index 4d0ff5e..b37653f 100644
> --- a/fs/bio-integrity.c
> +++ b/fs/bio-integrity.c
> @@ -782,7 +782,7 @@ void __init bio_integrity_init(void)
> {
> unsigned int i;
>
> - kintegrityd_wq = create_workqueue("kintegrityd");
> + kintegrityd_wq = alloc_workqueue("kintegrityd", WQ_MEM_RECLAIM, 1);
> if (!kintegrityd_wq)
> panic("Failed to create kintegrityd\n");
>
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index 2614774..5efccc6 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -3038,7 +3038,8 @@ no_journal:
> goto failed_mount_wq;
> }
>
> - EXT4_SB(sb)->dio_unwritten_wq = create_workqueue("ext4-dio-unwritten");
> + EXT4_SB(sb)->dio_unwritten_wq = alloc_workqueue("ext4-dio-unwritten",
> + WQ_MEM_RECLAIM, 1);
> if (!EXT4_SB(sb)->dio_unwritten_wq) {
> printk(KERN_ERR "EXT4-fs: failed to create DIO workqueue\n");
> goto failed_mount_wq;
> diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c
> index 812e2c0..03275f1 100644
> --- a/fs/reiserfs/journal.c
> +++ b/fs/reiserfs/journal.c
> @@ -2950,7 +2950,7 @@ int journal_init(struct super_block *sb, const char *j_dev_name,
> reiserfs_mounted_fs_count++;
> if (reiserfs_mounted_fs_count <= 1) {
> reiserfs_write_unlock(sb);
> - commit_wq = create_workqueue("reiserfs");
> + commit_wq = alloc_workqueue("reiserfs", WQ_MEM_RECLAIM, 1);
> reiserfs_write_lock(sb);
> }
>
> diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h
> index 25e02c9..77cb9d9 100644
> --- a/include/linux/workqueue.h
> +++ b/include/linux/workqueue.h
> @@ -247,6 +247,8 @@ enum {
> WQ_HIGHPRI = 1 << 4, /* high priority */
> WQ_CPU_INTENSIVE = 1 << 5, /* cpu instensive workqueue */
>
> + WQ_MEM_RECLAIM = WQ_RESCUER | WQ_HIGHPRI,
> +
> WQ_DYING = 1 << 6, /* internal: workqueue is dying */
>
> WQ_MAX_ACTIVE = 512, /* I like 512, better ideas? */
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Kumar Gala: "Re: [PATCH] edac: mpc85xx: Add support for new MPCxxx/Pxxxx EDAC controllers (fix)"
Previous message: David Howells: "Re: linux-next: manual merge of the irqflags tree with Linus' tree"
In reply to: Tejun Heo: "Re: -mm: xfs lockdep warning"
Next in thread: Tejun Heo: "Re: -mm: xfs lockdep warning"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]