638206 – sys-kernel/gentoo-sources-4.14.0 - bcache destroys filesystem...

Bug 638206 - sys-kernel/gentoo-sources-4.14.0 - bcache destroys filesystems

Using bcache can destroy the filesystem. Mine was gone after 3rd non sucesfull try to mount the rootfs. It was not possible to recover any file. (I had backups...)

Another report on the bcache mailing list:
https://www.spinics.net/lists/linux-bcache/msg05290.html

I would sugest to mask it with a warning that there is an bcache issue.

thanks for the report

masked the package and searching for more informations

maybe  no need to mask , 4.14.1 released !

Sent to stable list an hour ago:

https://www.spinics.net/lists/stable/msg198707.html:
Hi---

Previously sent to stable was this patch, and it's made it into
Linus's tree for 4.16.  It's a rather urgent data corruption issue
that affects at minimum bcache but possibly other block subsystems.
Several users have lost data.  The issue was introduced in the 4.15
branch by 74d46992e0d9 ("block: replace bi_bdev with a gendisk pointer
and partitions index").

I'd really appreciate anything that can be done to get this into a
stable 4.15 release soon.

Thanks--

Mike

+ update (not yet indexed)

I don't know what I was thinking here, but I meant 4.14/4.15 and not
4.15/4.16.  :P

Mike

Can you please try this patch? Probably a good candidate along with the 4.14.1 bump.

commit 62530ed8b1d07a45dec94d46e521c0c6c2d476e6
Author: Michael Lyle <[email protected]>
Date:   Thu Nov 16 23:47:25 2017 -0800

    bio: ensure __bio_clone_fast copies bi_partno
    
    A new field was introduced in 74d46992e0d9, bi_partno, instead of using
    bdev->bd_contains and encoding the partition information in the bi_bdev
    field.  __bio_clone_fast was changed to copy the disk information, but
    not the partition information.  At minimum, this regressed bcache and
    caused data corruption.
    
    Signed-off-by: Michael Lyle <[email protected]>
    Fixes: 74d46992e0d9 ("block: replace bi_bdev with a gendisk pointer and partitions index")
    Reported-by: Pavel Goran <[email protected]>
    Reported-by: Campbell Steven <[email protected]>
    Reviewed-by: Coly Li <[email protected]>
    Reviewed-by: Ming Lei <[email protected]>
    Cc: <[email protected]> # 4.14
    Signed-off-by: Jens Axboe <[email protected]>

diff --git a/block/bio.c b/block/bio.c
index b94a802f8ba3..459cc857f3d9 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -597,6 +597,7 @@ void __bio_clone_fast(struct bio *bio, struct bio *bio_src)
         * so we don't set nor calculate new physical/hw segment counts here
         */
        bio->bi_disk = bio_src->bi_disk;
+       bio->bi_partno = bio_src->bi_partno;
        bio_set_flag(bio, BIO_CLONED);
        bio->bi_opf = bio_src->bi_opf;
        bio->bi_write_hint = bio_src->bi_write_hint;

(In reply to Tomáš Mózes from comment #6)
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/
> ?id=62530ed8b1d07a45dec94d46e521c0c6c2d476e6

Thanks for the info. I will add this patch and release with 4.14.1.

added in gentoo-sources-4.14.1

This bug affects sys-kernel/vanilla-sources-4.14.0 and sys-kernel/vanilla-sources-4.14.1 too.

you are right

masked as well

Masked:

commit 3f3676066690d9a176a613f9779360b2030bb4a0
Author: Alice Ferrazzi <[email protected]>                                                                                                                                            
Date:   Tue Nov 21 12:33:45 2017 +0000                                                                                                                                                
                                                                                                                                                                                      
    profiles: mask sys-kernel/vanilla-sources-4.14.1.ebuild (bug 638206)                                                                                                              
                                                                                                                                                                                      
commit 1c37d6db7962c1a7db9c1ca63960ab2c8254894b
Author: Alice Ferrazzi <[email protected]>                                                                                                                                            
Date:   Tue Nov 21 17:26:28 2017 +0900                                                                                                                                                
                                                                                                                                                                                      
    profiles: mask sys-kernel/gentoo-sources-4.14.0 (bug 638206)


Released new version with patch included:

commit 0a8c30502306c781db8b2c973506e3c62da400e5 (HEAD -> master, origin/master, origin/HEAD)
Author: Mike Pagano <[email protected]>                                                                                                                                              
Date:   Tue Nov 21 07:47:51 2017 -0500                                                                                                                                                
                                                                                                                                                                                      
    sys-kernel/gentoo-sources: Linux patch 4.14.1 including a fix for bcache data corruption bug. See bug #638206.                                                                    

    Package-Manager: Portage-2.3.13, Repoman-2.3.3   


I think we're done, right?

Thanks, it looks good indeed, should be fine :)

(In reply to bug2017 from comment #0)
> Using bcache can destroy the filesystem. Mine was gone after 3rd non
> sucesfull try to mount the rootfs. It was not possible to recover any file.
> (I had backups...)
> 
> Another report on the bcache mailing list:
> https://www.spinics.net/lists/linux-bcache/msg05290.html
> 
> I would sugest to mask it with a warning that there is an bcache issue.

Can you please report back if it helped you? I don't use bcache and 4.14.0 works fine for me (except for one machine).

masked and removed affected kernel
sys-kernel/vanilla-sources-4.14.0
sys-kernel/vanilla-sources-4.14.1
sys-kernel/gentoo-sources-4.14.0

masked ck-sources-4.14.0
waiting mainteiner for removal

Acknowledged. Thanks for the ping.

System is restored, about 170GB written, everything looks fine. (Btrfs scrub, reboot ...)

@bug2017 thanks for the follow up