Re: [PATCH] jbd2: gate checksum calculations on crc driver presence, not sb flags

From: Darrick J. Wong
Date: Thu Oct 01 2015 - 14:18:46 EST


On Thu, Oct 01, 2015 at 12:35:12AM -0600, Andreas Dilger wrote:
> On Sep 30, 2015, at 11:47 AM, Darrick J. Wong <darrick.wong@xxxxxxxxxx> wrote:
> >
> > Change the journal's checksum functions to gate on whether or not the
> > crc32c driver is loaded, and gate the loading on the superblock bits.
> > This prevents a journal crash if someone loads a journal in no-csum
> > mode and then randomizes the superblock, thus flipping on the feature
> > bits.
> >
> > Reported-by: Nikolay Borisov <kernel@xxxxxxxx>
> > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> > ---
> > fs/jbd2/journal.c | 12 +++++++++---
> > include/linux/jbd2.h | 10 ++++++----
> > 2 files changed, 15 insertions(+), 7 deletions(-)
> >
> > diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
> > index 8270fe9..16e3a46 100644
> > --- a/fs/jbd2/journal.c
> > +++ b/fs/jbd2/journal.c
> > @@ -122,9 +122,15 @@ EXPORT_SYMBOL(__jbd2_debug);
> > #endif
> >
> > /* Checksumming functions */
> > +static bool journal_has_csum_v2or3_feature(journal_t *j)
> > +{
> > + return JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2) ||
> > + JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V3);
> > +}
> > +
> > static int jbd2_verify_csum_type(journal_t *j, journal_superblock_t *sb)
> > {
> > - if (!jbd2_journal_has_csum_v2or3(j))
> > + if (!journal_has_csum_v2or3_feature(j))
> > return 1;
> >
> > return sb->s_checksum_type == JBD2_CRC32C_CHKSUM;
> > @@ -1531,7 +1537,7 @@ static int journal_get_superblock(journal_t *journal)
> > goto out;
> > }
> >
> > - if (jbd2_journal_has_csum_v2or3(journal) &&
> > + if (journal_has_csum_v2or3_feature(journal) &&
> > JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM)) {
> > /* Can't have checksum v1 and v2 on at the same time! */
> > printk(KERN_ERR "JBD2: Can't enable checksumming v1 and v2/3 "
> > @@ -1545,7 +1551,7 @@ static int journal_get_superblock(journal_t *journal)
> > }
> >
> > /* Load the checksum driver */
> > - if (jbd2_journal_has_csum_v2or3(journal)) {
> > + if (journal_has_csum_v2or3_feature(journal)) {
> > journal->j_chksum_driver = crypto_alloc_shash("crc32c", 0, 0);
> > if (IS_ERR(journal->j_chksum_driver)) {
> > printk(KERN_ERR "JBD2: Cannot load crc32c driver.\n");
> > diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h
> > index df07e78..c74c786 100644
> > --- a/include/linux/jbd2.h
> > +++ b/include/linux/jbd2.h
> > @@ -1340,11 +1340,13 @@ extern size_t journal_tag_bytes(journal_t *journal);
> >
> > static inline int jbd2_journal_has_csum_v2or3(journal_t *journal)
> > {
> > - if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2) ||
> > - JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3))
> > - return 1;
> > + WARN_ON_ONCE((JBD2_HAS_INCOMPAT_FEATURE(journal,
> > + JBD2_FEATURE_INCOMPAT_CSUM_V2) ||
> > + JBD2_HAS_INCOMPAT_FEATURE(journal,
> > + JBD2_FEATURE_INCOMPAT_CSUM_V3)) &&
> > + journal->j_chksum_driver == NULL);
> >
> > - return 0;
> > + return journal->j_chksum_driver != NULL;
> > }
>
> Why not use:
>
> WARN_ON_ONCE(journal_has_csum_v2orv3_feature() &&
> journal->j_chksum_driver == NULL);
>
> rather than open-coding it? Yes, you would have to move that function
> to the header and give it a better name.

Sounds like a good idea, thanks.

>
> As a side note, I've long thought about changing the macros to be shorter:
>
> #define JBD2_HAS_INCOMPAT_FEATURE(j, name) \
> ((j)->j_format_version >= 2 && \
> ((j)->j_superblock->s_feature_incompat & \
> cpu_to_be32((JBD2_HAS_INCOMPAT_FEATURE_ ## name))))
>
> so they can be used like:
>
> static bool jbd2_journal_has_csum_v2or3_feature(journal_t *journal)
> {
> return JBD2_HAS_INCOMPAT_FEATURE(journal, CSUM_V2) ||
> JBD2_HAS_INCOMPAT_FEATURE(journal, CSUM_V3);
> }
>
> This not only makes the code much shorter and more readable, it also
> avoids potentially hard-to-spot bugs like the following:
>
> JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_COMPAT_CHECKSUM)
>
> The same would be useful for the equivalent ext4 macros as well.

Yes it will... as a separate patch. Shorter lines and fewer opportunities
to screw things up. :)

--D

>
> Cheers, Andreas
>
>
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
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/