Re: Global hotspare functionality

[Anand Jain <anand.jain@oracle.com>]

On 03/31/2016 06:17 AM, Yauhen Kharuzhy wrote:
> On Tue, Mar 29, 2016 at 10:40:40PM +0300, Yauhen Kharuzhy wrote:
>> Hi.
>>
>> I am testing hotspare v2 on kernel v4.4.5 (I will try latest Chris' tree later)
>> now with lockdep debugging enabled. At starting of replacement, lockdep warning is displayed,
>> because kstrdup() is called with GFP_NOFS inside of rcu_read_lock/unlock()
>> block (GFP_NOFS can sleep).
>
> Similar thing in the btrfs_auto_replace_start(): rcu_str_deref() without
> rcu_read_lock():
>
> int btrfs_auto_replace_start(struct btrfs_root *root,
>                                  struct btrfs_device *src_device)
> {
>          int ret;
>          char *tgt_path;
>
>          if (btrfs_get_spare_device(&tgt_path)) {
>                  btrfs_err(root->fs_info,
>                          "No spare device found/configured in the kernel");
>                  return -EINVAL;
>          }
>
>          ret = btrfs_dev_replace_start(root, tgt_path,
>                                          src_device->devid,
>                                          rcu_str_deref(src_device->name),

This is fixed in V3.

Thanks, Anand


>                  BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID);
>          if (ret)
>                  btrfs_put_spare_device(tgt_path);
>
>          kfree(tgt_path);
>
>          return 0;
> }
>
> [  156.168133] ===============================
> [  156.168963] [ INFO: suspicious RCU usage. ]
> [  156.169822] 4.4.5-scst31x+ #20 Not tainted
> [  156.170656] -------------------------------
> [  156.171488] fs/btrfs/dev-replace.c:990 suspicious rcu_dereference_check() usage!
> [  156.172920]
> [  156.172920] other info that might help us debug this:
> [  156.172920]
> [  156.174825]
> [  156.174825] rcu_scheduler_active = 1, debug_locks = 0
> [  156.176152] 1 lock held by btrfs-casualty/4807:
> [  156.181917]  #0:  (&fs_info->casualty_mutex){+.+...}, at: [<ffffffffa0165364>] casualty_kthread+0x64/0x390 [btrfs]
> [  156.193511]
> [  156.193511] stack backtrace:
> [  156.194680] CPU: 0 PID: 4807 Comm: btrfs-casualty Not tainted 4.4.5-scst31x+ #20
> [  156.201650] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
> [  156.219100]  0000000000000000 ffff88005d79fda0 ffffffff813529e3 ffff88005e19c600
> [  156.221216]  0000000000000001 ffff88005d79fdd0 ffffffff810d6407 0000000000000000
> [  156.224287]  0000000000000000 ffff88005f4a0c00 ffff88005da36000 ffff88005d79fe08
> [  156.226375] Call Trace:
> [  156.227078]  [<ffffffff813529e3>] dump_stack+0x85/0xc2
> [  156.228152]  [<ffffffff810d6407>] lockdep_rcu_suspicious+0xd7/0x110
> [  156.229418]  [<ffffffffa01e6236>] btrfs_auto_replace_start+0xa6/0xd0 [btrfs]
> [  156.230714]  [<ffffffffa01655c4>] casualty_kthread+0x2c4/0x390 [btrfs]
> [  156.231915]  [<ffffffffa016549c>] ? casualty_kthread+0x19c/0x390 [btrfs]
> [  156.233105]  [<ffffffffa0165300>] ? btrfs_check_devices+0x200/0x200 [btrfs]
> [  156.234339]  [<ffffffff810a70df>] kthread+0xef/0x110
> [  156.235309]  [<ffffffff810dc081>] ? __raw_callee_save___pv_queued_spin_unlock+0x11/0x20
> [  156.236940]  [<ffffffff810a6ff0>] ? kthread_create_on_node+0x200/0x200
> [  156.239489]  [<ffffffff81637c2f>] ret_from_fork+0x3f/0x70
> [  156.240533]  [<ffffffff810a6ff0>] ? kthread_create_on_node+0x200/0x200
>
>