On Thu, Jan 30, 2014 at 1:18 PM, Filipe David Manana <fdmanana@xxxxxxxxx> wrote:
> On Thu, Jan 30, 2014 at 3:11 AM, Darrick J. Wong
> <darrick.wong@xxxxxxxxxx> wrote:
>> On Sun, Jan 12, 2014 at 02:22:46AM +0000, Filipe David Borba Manana wrote:
>>> After the change titled "Btrfs: add support for inode properties", if
>>> btrfs was built-in the kernel (i.e. not as a module), it would cause a
>>> kernel panic, as reported recently by Fengguang:
>>>
>>> [ 2.024722] BUG: unable to handle kernel NULL pointer dereference at (null)
>>> [ 2.027814] IP: [<ffffffff81501594>] crc32c+0xc/0x6b
>>> [ 2.028684] PGD 0
>>> [ 2.028684] Oops: 0000 [#1] SMP
>>> [ 2.028684] Modules linked in:
>>> [ 2.028684] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.13.0-rc7-04795-ga7b57c2 #1
>>> [ 2.028684] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
>>> [ 2.028684] task: ffff88000edba100 ti: ffff88000edd6000 task.ti: ffff88000edd6000
>>> [ 2.028684] RIP: 0010:[<ffffffff81501594>] [<ffffffff81501594>] crc32c+0xc/0x6b
>>> [ 2.028684] RSP: 0000:ffff88000edd7e58 EFLAGS: 00010246
>>> [ 2.028684] RAX: 0000000000000000 RBX: ffffffff82295550 RCX: 0000000000000000
>>> [ 2.028684] RDX: 0000000000000011 RSI: ffffffff81efe393 RDI: 00000000fffffffe
>>> [ 2.028684] RBP: ffff88000edd7e60 R08: 0000000000000003 R09: 0000000000015d20
>>> [ 2.028684] R10: ffffffff81ef225e R11: ffffffff811b0222 R12: ffffffffffffffff
>>> [ 2.028684] R13: 0000000000000239 R14: 0000000000000000 R15: 0000000000000000
>>> [ 2.028684] FS: 0000000000000000(0000) GS:ffff88000fa00000(0000) knlGS:0000000000000000
>>> [ 2.028684] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
>>> [ 2.028684] CR2: 0000000000000000 CR3: 000000000220c000 CR4: 00000000000006f0
>>> [ 2.028684] Stack:
>>> [ 2.028684] ffffffff82295550 ffff88000edd7e80 ffffffff8238af62 ffffffff8238ac05
>>> [ 2.028684] 0000000000000000 ffff88000edd7e98 ffffffff8238ac0f ffffffff8238ac05
>>> [ 2.028684] ffff88000edd7f08 ffffffff810002ba ffff88000edd7f00 ffffffff810e2404
>>> [ 2.028684] Call Trace:
>>> [ 2.028684] [<ffffffff8238af62>] btrfs_props_init+0x4f/0x96
>>> [ 2.028684] [<ffffffff8238ac05>] ? ftrace_define_fields_btrfs_space_reservation+0x145/0x145
>>> [ 2.028684] [<ffffffff8238ac0f>] init_btrfs_fs+0xa/0xf0
>>> [ 2.028684] [<ffffffff8238ac05>] ? ftrace_define_fields_btrfs_space_reservation+0x145/0x145
>>> [ 2.028684] [<ffffffff810002ba>] do_one_initcall+0xa4/0x13a
>>> [ 2.028684] [<ffffffff810e2404>] ? parse_args+0x25f/0x33d
>>> [ 2.028684] [<ffffffff8234cf75>] kernel_init_freeable+0x1aa/0x230
>>> [ 2.028684] [<ffffffff8234c785>] ? do_early_param+0x88/0x88
>>> [ 2.028684] [<ffffffff819f61b5>] ? rest_init+0x89/0x89
>>> [ 2.028684] [<ffffffff819f61c3>] kernel_init+0xe/0x109
>>>
>>> The issue here is that the initialization function of btrfs (super.c:init_btrfs_fs)
>>> started using crc32c (from lib/libcrc32c.c). But when it needs to call crc32c (as
>>> part of the properties initialization routine), the libcrc32c is not yet initialized,
>>> so crc32c derreferenced a NULL pointer (lib/libcrc32c.c:tfm), causing the kernel
>>> panic on boot.
>>>
>>> The approach to fix this is to use crypto component directly to use its crc32c (which
>>> is basically what lib/libcrc32c.c is, a wrapper around crypto). This is what ext4 is
>>> doing as well, it uses crypto directly to get crc32c functionality.
>>
>> Sorry, I didn't even see this until your other patch today...
>>
>> Yes, ext4 bypasses libcrc32c, but notice how ext4/jbd2 only call
>> crypto_alloc_shash() when you mount the filesystem. This silly trick enables
>> ext4 to take advantage of faster crc32c implementations that don't get loaded
>> until after the initial insmod, just in case you do this:
>>
>> # modprobe btrfs
>> # mount /dev/X / <-- / uses software crc32c implementation
>> # modprobe crc32c-intel
>> # mount /dev/Y /home <-- /home still uses software crc32c, because
>> *tfm still points to the sw implementation!
>>
>> Of course, one could argue that this is a distro problem, and that the
>> initramfs or kernel config should take care to load the appropriate hardware
>> accelerated drivers /before/ btrfs.
>>
>> On the other hand, btrfs might as well always enjoy the fastest implementation
>> that it can get its hands on.
>
> Thanks Darrick, that's very useful information.
>
> So, if I understood it correctly, before this change btrfs was using
> libcrc32c, which is initialized only once and therefore it's possible
> it wouldn't use crc32c-intel - if crc32c-intel was built as a module
> and not loaded before libcrc32c is initialized.
>
> Also that ext4 approach, of using crypto_alloc_shash() on each mount
> (and make the result accessible via super block structure) doesn't
> solve the case where the partition is mounted before crc32c-intel is
> loaded but then it's never unmounted and mounted again (like a root or
> home partition). By solve I mean not profiting from intel's hardware
> based crc32c.
>
> I suppose that if crc32c-intel is built-in (i.e. not as a module) that
> either with the current approach (after this patch) or with the old
> approach, we'll be using crc32c-intel. Is there any way to verify
> this?
Just answered my own last question by printing
tfm->base.__crt_alg->cra_driver_name in
fs/btrfs/hash.c:btrfs_hash_init(), which printed "crc32c-intel".
>
>>
>> (Also I suppose metadata_csum is optional in ext4/jbd2...)
>>
>> --D
>>
>>>
>>> Verified this works both when btrfs is built-in and when it's loadable kernel module.
>>>
>>> Signed-off-by: Filipe David Borba Manana <fdmanana@xxxxxxxxx>
>>> ---
>>>
>>> V2: Removed unnecessary __exit qualifier.
>>>
>>> fs/btrfs/Kconfig | 3 ++-
>>> fs/btrfs/Makefile | 2 +-
>>> fs/btrfs/extent-tree.c | 6 +++---
>>> fs/btrfs/hash.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++++
>>> fs/btrfs/hash.h | 11 ++++++++---
>>> fs/btrfs/super.c | 10 +++++++++-
>>> 6 files changed, 72 insertions(+), 9 deletions(-)
>>> create mode 100644 fs/btrfs/hash.c
>>>
>>> diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig
>>> index aa976ec..a66768e 100644
>>> --- a/fs/btrfs/Kconfig
>>> +++ b/fs/btrfs/Kconfig
>>> @@ -1,6 +1,7 @@
>>> config BTRFS_FS
>>> tristate "Btrfs filesystem support"
>>> - select LIBCRC32C
>>> + select CRYPTO
>>> + select CRYPTO_CRC32C
>>> select ZLIB_INFLATE
>>> select ZLIB_DEFLATE
>>> select LZO_COMPRESS
>>> diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
>>> index af7f000..f341a98 100644
>>> --- a/fs/btrfs/Makefile
>>> +++ b/fs/btrfs/Makefile
>>> @@ -9,7 +9,7 @@ btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
>>> export.o tree-log.o free-space-cache.o zlib.o lzo.o \
>>> compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o \
>>> reada.o backref.o ulist.o qgroup.o send.o dev-replace.o raid56.o \
>>> - uuid-tree.o props.o
>>> + uuid-tree.o props.o hash.o
>>>
>>> btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o
>>> btrfs-$(CONFIG_BTRFS_FS_CHECK_INTEGRITY) += check-integrity.o
>>> diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
>>> index 77acc08..db19ed7 100644
>>> --- a/fs/btrfs/extent-tree.c
>>> +++ b/fs/btrfs/extent-tree.c
>>> @@ -1072,11 +1072,11 @@ static u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
>>> __le64 lenum;
>>>
>>> lenum = cpu_to_le64(root_objectid);
>>> - high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
>>> + high_crc = btrfs_crc32c(high_crc, &lenum, sizeof(lenum));
>>> lenum = cpu_to_le64(owner);
>>> - low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
>>> + low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
>>> lenum = cpu_to_le64(offset);
>>> - low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
>>> + low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
>>>
>>> return ((u64)high_crc << 31) ^ (u64)low_crc;
>>> }
>>> diff --git a/fs/btrfs/hash.c b/fs/btrfs/hash.c
>>> new file mode 100644
>>> index 0000000..ab3a938
>>> --- /dev/null
>>> +++ b/fs/btrfs/hash.c
>>> @@ -0,0 +1,49 @@
>>> +/*
>>> + * Copyright (C) 2014 Filipe David Borba Manana <fdmanana@xxxxxxxxx>
>>> + *
>>> + * This program is free software; you can redistribute it and/or
>>> + * modify it under the terms of the GNU General Public
>>> + * License v2 as published by the Free Software Foundation.
>>> + *
>>> + * This program is distributed in the hope that it will be useful,
>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
>>> + * General Public License for more details.
>>> + */
>>> +
>>> +#include <crypto/hash.h>
>>> +#include "hash.h"
>>> +
>>> +static struct crypto_shash *tfm;
>>> +
>>> +int __init btrfs_hash_init(void)
>>> +{
>>> + tfm = crypto_alloc_shash("crc32c", 0, 0);
>>> + if (IS_ERR(tfm))
>>> + return PTR_ERR(tfm);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +void btrfs_hash_exit(void)
>>> +{
>>> + crypto_free_shash(tfm);
>>> +}
>>> +
>>> +u32 btrfs_crc32c(u32 crc, const void *address, unsigned int length)
>>> +{
>>> + struct {
>>> + struct shash_desc shash;
>>> + char ctx[crypto_shash_descsize(tfm)];
>>> + } desc;
>>> + int err;
>>> +
>>> + desc.shash.tfm = tfm;
>>> + desc.shash.flags = 0;
>>> + *(u32 *)desc.ctx = crc;
>>> +
>>> + err = crypto_shash_update(&desc.shash, address, length);
>>> + BUG_ON(err);
>>> +
>>> + return *(u32 *)desc.ctx;
>>> +}
>>> diff --git a/fs/btrfs/hash.h b/fs/btrfs/hash.h
>>> index 1d98281..118a231 100644
>>> --- a/fs/btrfs/hash.h
>>> +++ b/fs/btrfs/hash.h
>>> @@ -19,10 +19,15 @@
>>> #ifndef __HASH__
>>> #define __HASH__
>>>
>>> -#include <linux/crc32c.h>
>>> +int __init btrfs_hash_init(void);
>>> +
>>> +void btrfs_hash_exit(void);
>>> +
>>> +u32 btrfs_crc32c(u32 crc, const void *address, unsigned int length);
>>> +
>>> static inline u64 btrfs_name_hash(const char *name, int len)
>>> {
>>> - return crc32c((u32)~1, name, len);
>>> + return btrfs_crc32c((u32)~1, name, len);
>>> }
>>>
>>> /*
>>> @@ -31,7 +36,7 @@ static inline u64 btrfs_name_hash(const char *name, int len)
>>> static inline u64 btrfs_extref_hash(u64 parent_objectid, const char *name,
>>> int len)
>>> {
>>> - return (u64) crc32c(parent_objectid, name, len);
>>> + return (u64) btrfs_crc32c(parent_objectid, name, len);
>>> }
>>>
>>> #endif
>>> diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
>>> index 461e41c..f44cc6a 100644
>>> --- a/fs/btrfs/super.c
>>> +++ b/fs/btrfs/super.c
>>> @@ -48,6 +48,7 @@
>>> #include "transaction.h"
>>> #include "btrfs_inode.h"
>>> #include "print-tree.h"
>>> +#include "hash.h"
>>> #include "props.h"
>>> #include "xattr.h"
>>> #include "volumes.h"
>>> @@ -1866,11 +1867,15 @@ static int __init init_btrfs_fs(void)
>>> {
>>> int err;
>>>
>>> + err = btrfs_hash_init();
>>> + if (err)
>>> + return err;
>>> +
>>> btrfs_props_init();
>>>
>>> err = btrfs_init_sysfs();
>>> if (err)
>>> - return err;
>>> + goto free_hash;
>>>
>>> btrfs_init_compress();
>>>
>>> @@ -1945,6 +1950,8 @@ free_cachep:
>>> free_compress:
>>> btrfs_exit_compress();
>>> btrfs_exit_sysfs();
>>> +free_hash:
>>> + btrfs_hash_exit();
>>> return err;
>>> }
>>>
>>> @@ -1963,6 +1970,7 @@ static void __exit exit_btrfs_fs(void)
>>> btrfs_exit_sysfs();
>>> btrfs_cleanup_fs_uuids();
>>> btrfs_exit_compress();
>>> + btrfs_hash_exit();
>>> }
>>>
>>> module_init(init_btrfs_fs)
>>> --
>>> 1.7.9.5
>>>
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
>>> the body of a message to majordomo@xxxxxxxxxxxxxxx
>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
>
> --
> Filipe David Manana,
>
> "Reasonable men adapt themselves to the world.
> Unreasonable men adapt the world to themselves.
> That's why all progress depends on unreasonable men."
--
Filipe David Manana,
"Reasonable men adapt themselves to the world.
Unreasonable men adapt the world to themselves.
That's why all progress depends on unreasonable men."
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
