diff mbox

[v2] Btrfs: fix RCU correctness warning when running sanity tests

Message ID 1402424236-13459-1-git-send-email-fdmanana@gmail.com (mailing list archive)
State Accepted
Headers show

Commit Message

Filipe Manana June 10, 2014, 6:17 p.m. UTC 
When CONFIG_PROVE_RCU=y and CONFIG_PROVE_RCU_REPEATEDLY=y, the
following was dumped in dmesg:

[ 3197.218064] ===============================
[ 3197.218064] [ INFO: suspicious RCU usage. ]
[ 3197.218066] 3.15.0-rc8-fdm-btrfs-next-33+ #4 Not tainted
[ 3197.218067] -------------------------------
[ 3197.218068] include/linux/radix-tree.h:196 suspicious rcu_dereference_check() usage!
[ 3197.218068]
[ 3197.218068] other info that might help us debug this:
[ 3197.218068]
[ 3197.218070]
[ 3197.218070] rcu_scheduler_active = 1, debug_locks = 1
[ 3197.218071] 1 lock held by modprobe/12024:
[ 3197.218072]  #0:  (&(&fs_info->buffer_lock)->rlock){+.+...}, at: [<ffffffffa025c5fa>] btrfs_free_dummy_root+0x5a/0x1d0 [btrfs]
[ 3197.218093]
[ 3197.218093] stack backtrace:
[ 3197.218095] CPU: 3 PID: 12024 Comm: modprobe Not tainted 3.15.0-rc8-fdm-btrfs-next-33+ #4
[ 3197.218096] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[ 3197.218097]  0000000000000001 ffff8800af18fc18 ffffffff81685c5a 000000000000feb0
[ 3197.218099]  ffff8800cf6ccb40 ffff8800af18fc48 ffffffff810a6316 ffff8801d955f640
[ 3197.218101]  ffff8800d719e328 ffff8800d719e370 ffff8800d719c000 ffff8800af18fcb8
[ 3197.218102] Call Trace:
[ 3197.218105]  [<ffffffff81685c5a>] dump_stack+0x4e/0x68
[ 3197.218108]  [<ffffffff810a6316>] lockdep_rcu_suspicious+0xe6/0x130
[ 3197.218119]  [<ffffffffa025c728>] btrfs_free_dummy_root+0x188/0x1d0 [btrfs]
[ 3197.218129]  [<ffffffffa025f56a>] btrfs_test_qgroups+0xea/0x1bb [btrfs]
[ 3197.218137]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
[ 3197.218144]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
[ 3197.218151]  [<ffffffffa03a1ab7>] init_btrfs_fs+0xe5/0x184 [btrfs]
[ 3197.218154]  [<ffffffff81000352>] do_one_initcall+0x102/0x150
[ 3197.218157]  [<ffffffff8103d223>] ? set_memory_nx+0x43/0x50
[ 3197.218160]  [<ffffffff81682668>] ? set_section_ro_nx+0x6d/0x74
[ 3197.218162]  [<ffffffff810d91cc>] load_module+0x1cdc/0x2630
[ 3197.218164]  [<ffffffff810d4e90>] ? show_initstate+0x60/0x60
[ 3197.218166]  [<ffffffff810d9c9e>] SyS_finit_module+0x8e/0x90
[ 3197.218168]  [<ffffffff81698212>] system_call_fastpath+0x16/0x1b

Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
---

V2: Added missing rcu read unlock if a retry is needed.

 fs/btrfs/tests/btrfs-tests.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

Comments

Filipe Manana June 13, 2014, 9:09 a.m. UTC | #1 
On Tue, Jun 10, 2014 at 7:17 PM, Filipe David Borba Manana
<fdmanana@gmail.com> wrote:
> When CONFIG_PROVE_RCU=y and CONFIG_PROVE_RCU_REPEATEDLY=y, the
> following was dumped in dmesg:
>
> [ 3197.218064] ===============================
> [ 3197.218064] [ INFO: suspicious RCU usage. ]
> [ 3197.218066] 3.15.0-rc8-fdm-btrfs-next-33+ #4 Not tainted
> [ 3197.218067] -------------------------------
> [ 3197.218068] include/linux/radix-tree.h:196 suspicious rcu_dereference_check() usage!
> [ 3197.218068]
> [ 3197.218068] other info that might help us debug this:
> [ 3197.218068]
> [ 3197.218070]
> [ 3197.218070] rcu_scheduler_active = 1, debug_locks = 1
> [ 3197.218071] 1 lock held by modprobe/12024:
> [ 3197.218072]  #0:  (&(&fs_info->buffer_lock)->rlock){+.+...}, at: [<ffffffffa025c5fa>] btrfs_free_dummy_root+0x5a/0x1d0 [btrfs]
> [ 3197.218093]
> [ 3197.218093] stack backtrace:
> [ 3197.218095] CPU: 3 PID: 12024 Comm: modprobe Not tainted 3.15.0-rc8-fdm-btrfs-next-33+ #4
> [ 3197.218096] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
> [ 3197.218097]  0000000000000001 ffff8800af18fc18 ffffffff81685c5a 000000000000feb0
> [ 3197.218099]  ffff8800cf6ccb40 ffff8800af18fc48 ffffffff810a6316 ffff8801d955f640
> [ 3197.218101]  ffff8800d719e328 ffff8800d719e370 ffff8800d719c000 ffff8800af18fcb8
> [ 3197.218102] Call Trace:
> [ 3197.218105]  [<ffffffff81685c5a>] dump_stack+0x4e/0x68
> [ 3197.218108]  [<ffffffff810a6316>] lockdep_rcu_suspicious+0xe6/0x130
> [ 3197.218119]  [<ffffffffa025c728>] btrfs_free_dummy_root+0x188/0x1d0 [btrfs]
> [ 3197.218129]  [<ffffffffa025f56a>] btrfs_test_qgroups+0xea/0x1bb [btrfs]
> [ 3197.218137]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
> [ 3197.218144]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
> [ 3197.218151]  [<ffffffffa03a1ab7>] init_btrfs_fs+0xe5/0x184 [btrfs]
> [ 3197.218154]  [<ffffffff81000352>] do_one_initcall+0x102/0x150
> [ 3197.218157]  [<ffffffff8103d223>] ? set_memory_nx+0x43/0x50
> [ 3197.218160]  [<ffffffff81682668>] ? set_section_ro_nx+0x6d/0x74
> [ 3197.218162]  [<ffffffff810d91cc>] load_module+0x1cdc/0x2630
> [ 3197.218164]  [<ffffffff810d4e90>] ? show_initstate+0x60/0x60
> [ 3197.218166]  [<ffffffff810d9c9e>] SyS_finit_module+0x8e/0x90
> [ 3197.218168]  [<ffffffff81698212>] system_call_fastpath+0x16/0x1b
>
> Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>

Chris,

Please drop this patch from your integration branch.
Sasha fixed this too but in a simpler way:
https://patchwork.kernel.org/patch/4337091/
(In fact both patches applied probably trigger another warning the rcu
usage correctness checker)

Thanks

> ---
>
> V2: Added missing rcu read unlock if a retry is needed.
>
>  fs/btrfs/tests/btrfs-tests.c | 12 +++++++-----
>  1 file changed, 7 insertions(+), 5 deletions(-)
>
> diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c
> index a5dcacb..bdb1f05 100644
> --- a/fs/btrfs/tests/btrfs-tests.c
> +++ b/fs/btrfs/tests/btrfs-tests.c
> @@ -130,8 +130,8 @@ static void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info)
>         struct radix_tree_iter iter;
>         void **slot;
>
> -       spin_lock(&fs_info->buffer_lock);
>  restart:
> +       rcu_read_lock();
>         radix_tree_for_each_slot(slot, &fs_info->buffer_radix, &iter, 0) {
>                 struct extent_buffer *eb;
>
> @@ -140,15 +140,17 @@ restart:
>                         continue;
>                 /* Shouldn't happen but that kind of thinking creates CVE's */
>                 if (radix_tree_exception(eb)) {
> -                       if (radix_tree_deref_retry(eb))
> +                       if (radix_tree_deref_retry(eb)) {
> +                               rcu_read_unlock();
>                                 goto restart;
> +                       }
>                         continue;
>                 }
> -               spin_unlock(&fs_info->buffer_lock);
> +               rcu_read_unlock();
>                 free_extent_buffer_stale(eb);
> -               spin_lock(&fs_info->buffer_lock);
> +               goto restart;
>         }
> -       spin_unlock(&fs_info->buffer_lock);
> +       rcu_read_unlock();
>
>         btrfs_free_qgroup_config(fs_info);
>         btrfs_free_fs_roots(fs_info);
> --
> 1.9.1
>
Chris Mason June 13, 2014, 2:12 p.m. UTC | #2 
On 06/13/2014 05:09 AM, Filipe David Manana wrote:
> On Tue, Jun 10, 2014 at 7:17 PM, Filipe David Borba Manana
> <fdmanana@gmail.com> wrote:
>> When CONFIG_PROVE_RCU=y and CONFIG_PROVE_RCU_REPEATEDLY=y, the
>> following was dumped in dmesg:
>>
>> [ 3197.218064] ===============================
>> [ 3197.218064] [ INFO: suspicious RCU usage. ]
>> [ 3197.218066] 3.15.0-rc8-fdm-btrfs-next-33+ #4 Not tainted
>> [ 3197.218067] -------------------------------
>> [ 3197.218068] include/linux/radix-tree.h:196 suspicious rcu_dereference_check() usage!
>> [ 3197.218068]
>> [ 3197.218068] other info that might help us debug this:
>> [ 3197.218068]
>> [ 3197.218070]
>> [ 3197.218070] rcu_scheduler_active = 1, debug_locks = 1
>> [ 3197.218071] 1 lock held by modprobe/12024:
>> [ 3197.218072]  #0:  (&(&fs_info->buffer_lock)->rlock){+.+...}, at: [<ffffffffa025c5fa>] btrfs_free_dummy_root+0x5a/0x1d0 [btrfs]
>> [ 3197.218093]
>> [ 3197.218093] stack backtrace:
>> [ 3197.218095] CPU: 3 PID: 12024 Comm: modprobe Not tainted 3.15.0-rc8-fdm-btrfs-next-33+ #4
>> [ 3197.218096] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
>> [ 3197.218097]  0000000000000001 ffff8800af18fc18 ffffffff81685c5a 000000000000feb0
>> [ 3197.218099]  ffff8800cf6ccb40 ffff8800af18fc48 ffffffff810a6316 ffff8801d955f640
>> [ 3197.218101]  ffff8800d719e328 ffff8800d719e370 ffff8800d719c000 ffff8800af18fcb8
>> [ 3197.218102] Call Trace:
>> [ 3197.218105]  [<ffffffff81685c5a>] dump_stack+0x4e/0x68
>> [ 3197.218108]  [<ffffffff810a6316>] lockdep_rcu_suspicious+0xe6/0x130
>> [ 3197.218119]  [<ffffffffa025c728>] btrfs_free_dummy_root+0x188/0x1d0 [btrfs]
>> [ 3197.218129]  [<ffffffffa025f56a>] btrfs_test_qgroups+0xea/0x1bb [btrfs]
>> [ 3197.218137]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
>> [ 3197.218144]  [<ffffffffa03a19d2>] ? ftrace_define_fields_btrfs_space_reservation+0xfd/0xfd [btrfs]
>> [ 3197.218151]  [<ffffffffa03a1ab7>] init_btrfs_fs+0xe5/0x184 [btrfs]
>> [ 3197.218154]  [<ffffffff81000352>] do_one_initcall+0x102/0x150
>> [ 3197.218157]  [<ffffffff8103d223>] ? set_memory_nx+0x43/0x50
>> [ 3197.218160]  [<ffffffff81682668>] ? set_section_ro_nx+0x6d/0x74
>> [ 3197.218162]  [<ffffffff810d91cc>] load_module+0x1cdc/0x2630
>> [ 3197.218164]  [<ffffffff810d4e90>] ? show_initstate+0x60/0x60
>> [ 3197.218166]  [<ffffffff810d9c9e>] SyS_finit_module+0x8e/0x90
>> [ 3197.218168]  [<ffffffff81698212>] system_call_fastpath+0x16/0x1b
>>
>> Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
> 
> Chris,
> 
> Please drop this patch from your integration branch.
> Sasha fixed this too but in a simpler way:
> https://urldefense.proofpoint.com/v1/url?u=https://patchwork.kernel.org/patch/4337091/&k=ZVNjlDMF0FElm4dQtryO4A%3D%3D%0A&r=6%2FL0lzzDhu0Y1hL9xm%2BQyA%3D%3D%0A&m=udAT%2BO62eFj7qxpld6pYEp%2FRKmeGPjqGP8i%2Fb8%2FcJ2s%3D%0A&s=e3bbde8e6b7bd327f6f32a8f959089bed8f4a25a27fa8b0ddf2589cf70b8a89f
> (In fact both patches applied probably trigger another warning the rcu

Thanks, it did look overlapping to me, was going to sort it out in the
morning.

-chris
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c
index a5dcacb..bdb1f05 100644
--- a/fs/btrfs/tests/btrfs-tests.c
+++ b/fs/btrfs/tests/btrfs-tests.c
@@ -130,8 +130,8 @@  static void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info)
 	struct radix_tree_iter iter;
 	void **slot;
 
-	spin_lock(&fs_info->buffer_lock);
 restart:
+	rcu_read_lock();
 	radix_tree_for_each_slot(slot, &fs_info->buffer_radix, &iter, 0) {
 		struct extent_buffer *eb;
 
@@ -140,15 +140,17 @@  restart:
 			continue;
 		/* Shouldn't happen but that kind of thinking creates CVE's */
 		if (radix_tree_exception(eb)) {
-			if (radix_tree_deref_retry(eb))
+			if (radix_tree_deref_retry(eb)) {
+				rcu_read_unlock();
 				goto restart;
+			}
 			continue;
 		}
-		spin_unlock(&fs_info->buffer_lock);
+		rcu_read_unlock();
 		free_extent_buffer_stale(eb);
-		spin_lock(&fs_info->buffer_lock);
+		goto restart;
 	}
-	spin_unlock(&fs_info->buffer_lock);
+	rcu_read_unlock();
 
 	btrfs_free_qgroup_config(fs_info);
 	btrfs_free_fs_roots(fs_info);