Re: [PATCH 09/35] e2fsck: abort on read error beyond end of FS

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Apr 03, 2015 at 03:11:39PM -0600, Andreas Dilger wrote:
> On Apr 2, 2015, at 12:00 AM, Darrick J. Wong <darrick.wong@xxxxxxxxxx> wrote:
> > On Wed, Apr 01, 2015 at 11:10:48PM -0500, Andreas Dilger wrote:
> >> It isn't clear what the benefit of this patch is?  There are times (e.g.
> >> if a partition table is broken or if a single-disk filesystem is changed
> >> to an MD RAID device) that it is useful to run e2fsck on such a
> >> filesystem.
> >> 
> >> With this patch, it's essentially turning a small error into a fatal one,
> >> but what is the benefit?
> > 
> > This came from a debugging patch I was using to see if I could trick fsck
> > into using garbage values when it tries to "fix" things.
> > 
> > I'm not really sure what the usage scenario is for running fsck on a FS
> > that's too big for the device containing it -- wouldn't you want fsck to
> > stop immediately?
> 
> As mentioned above, this can happen in some cases, and e2fsck already has
> a check for it and it will prompt to abort:
> 
>         { PR_0_FS_SIZE_WRONG,
>           N_("The @f size (according to the @S) is %b @bs\n"
>           "The physical size of the @v is %c @bs\n"
>           "Either the @S or the partition table is likely to be corrupt!\n"),
>           PROMPT_ABORT, 0 },
> 
> However, it doesn't _force_ e2fsck to abort, because someone might want to
> be able to recover from such a situation (e.g. lose a few files that use
> blocks beyond the end of the device) rather than have an unusable filesystem.

Yeah, you're probably right, this is at best a debugging patch.  Let's drop it.

--D

> 
> Cheers, Andreas
> 
> > --D
> > 
> >> 
> >> Cheers, Andreas
> >> 
> >>> On Apr 1, 2015, at 21:35, Darrick J. Wong <darrick.wong@xxxxxxxxxx> wrote:
> >>> 
> >>> Abort if we fail to read a block that's past the end of the FS.
> >>> Includes a flag to disable the abort behavior for selected parts of
> >>> the fsck run, so that we don't fail on a busted object prior to fixing
> >>> it.
> >>> 
> >>> Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> >>> ---
> >>> e2fsck/e2fsck.h   |    1 +
> >>> e2fsck/ehandler.c |    7 +++++--
> >>> e2fsck/extents.c  |    2 ++
> >>> e2fsck/journal.c  |    3 +++
> >>> e2fsck/message.c  |   12 +++++++++++-
> >>> e2fsck/pass1.c    |   28 ++++++++++++++++------------
> >>> e2fsck/pass1b.c   |    4 ++++
> >>> 7 files changed, 42 insertions(+), 15 deletions(-)
> >>> 
> >>> 
> >>> diff --git a/e2fsck/e2fsck.h b/e2fsck/e2fsck.h
> >>> index 5fda863..453b552 100644
> >>> --- a/e2fsck/e2fsck.h
> >>> +++ b/e2fsck/e2fsck.h
> >>> @@ -193,6 +193,7 @@ struct resource_track {
> >>> #define E2F_FLAG_TIME_INSANE    0x2000 /* Time is insane */
> >>> #define E2F_FLAG_PROBLEMS_FIXED    0x4000 /* At least one problem was fixed */
> >>> #define E2F_FLAG_ALLOC_OK    0x8000 /* Can we allocate blocks? */
> >>> +#define E2F_FLAG_IGNORE_READ_ERROR 0x10000 /* Don't rewrite read error blocks */
> >>> 
> >>> #define E2F_RESET_FLAGS (E2F_FLAG_TIME_INSANE | E2F_FLAG_PROBLEMS_FIXED)
> >>> 
> >>> diff --git a/e2fsck/ehandler.c b/e2fsck/ehandler.c
> >>> index 71ca301..847f8e5 100644
> >>> --- a/e2fsck/ehandler.c
> >>> +++ b/e2fsck/ehandler.c
> >>> @@ -60,8 +60,11 @@ static errcode_t e2fsck_handle_read_error(io_channel channel,
> >>>   preenhalt(ctx);
> >>> 
> >>>   /* Don't rewrite a block past the end of the FS. */
> >>> -    if (block >= ext2fs_blocks_count(fs->super))
> >>> -        return 0;
> >>> +    if (block >= ext2fs_blocks_count(fs->super)) {
> >>> +        if (ctx->flags & E2F_FLAG_IGNORE_READ_ERROR)
> >>> +            return 0;
> >>> +        abort();
> >>> +    }
> >>> 
> >>>   if (ask(ctx, _("Ignore error"), 1)) {
> >>>       if (ask(ctx, _("Force rewrite"), 1))
> >>> diff --git a/e2fsck/extents.c b/e2fsck/extents.c
> >>> index 8465299..cff265a 100644
> >>> --- a/e2fsck/extents.c
> >>> +++ b/e2fsck/extents.c
> >>> @@ -29,6 +29,7 @@ errcode_t e2fsck_rebuild_extents_later(e2fsck_t ctx, ext2_ino_t ino)
> >>> {
> >>>   if (!EXT2_HAS_INCOMPAT_FEATURE(ctx->fs->super,
> >>>                      EXT3_FEATURE_INCOMPAT_EXTENTS) ||
> >>> +        (ctx->flags & (E2F_FLAG_RESTART_LATER | E2F_FLAG_RESTART)) ||
> >>>       (ctx->options & E2F_OPT_NO) ||
> >>>       (ino != EXT2_ROOT_INO && ino < ctx->fs->super->s_first_ino))
> >>>       return 0;
> >>> @@ -339,6 +340,7 @@ static void rebuild_extents(e2fsck_t ctx, const char *pass_name, int pr_header)
> >>> 
> >>>   if (!EXT2_HAS_INCOMPAT_FEATURE(ctx->fs->super,
> >>>                      EXT3_FEATURE_INCOMPAT_EXTENTS) ||
> >>> +        (ctx->flags & (E2F_FLAG_RESTART_LATER | E2F_FLAG_RESTART)) ||
> >>>       !ext2fs_test_valid(ctx->fs) ||
> >>>       ctx->invalid_bitmaps) {
> >>>       if (ctx->inodes_to_rebuild)
> >>> diff --git a/e2fsck/journal.c b/e2fsck/journal.c
> >>> index 9f32095..c195797 100644
> >>> --- a/e2fsck/journal.c
> >>> +++ b/e2fsck/journal.c
> >>> @@ -315,6 +315,7 @@ static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
> >>>   journal->j_inode = NULL;
> >>>   journal->j_blocksize = ctx->fs->blocksize;
> >>> 
> >>> +    ctx->flags |= E2F_FLAG_IGNORE_READ_ERROR;
> >>>   if (uuid_is_null(sb->s_journal_uuid)) {
> >>>       if (!sb->s_journal_inum) {
> >>>           retval = EXT2_ET_BAD_INODE_NUM;
> >>> @@ -518,9 +519,11 @@ static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
> >>> 
> >>>   *ret_journal = journal;
> >>>   e2fsck_use_inode_shortcuts(ctx, 0);
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>>   return 0;
> >>> 
> >>> errout:
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>>   e2fsck_use_inode_shortcuts(ctx, 0);
> >>>   if (dev_fs)
> >>>       ext2fs_free_mem(&dev_fs);
> >>> diff --git a/e2fsck/message.c b/e2fsck/message.c
> >>> index 9c1433f..510f291 100644
> >>> --- a/e2fsck/message.c
> >>> +++ b/e2fsck/message.c
> >>> @@ -199,14 +199,24 @@ static void print_pathname(FILE *f, ext2_filsys fs, ext2_ino_t dir,
> >>> {
> >>>   errcode_t    retval = 0;
> >>>   char        *path;
> >>> +    e2fsck_t    ctx = fs ? (e2fsck_t) fs->priv_data : NULL;
> >>> +    int        flags;
> >>> 
> >>>   if (!dir && (ino < num_special_inodes)) {
> >>>       fputs(_(special_inode_name[ino]), f);
> >>>       return;
> >>>   }
> >>> 
> >>> -    if (fs)
> >>> +    if (fs) {
> >>> +        if (ctx) {
> >>> +            flags = ctx->flags;
> >>> +            ctx->flags |= E2F_FLAG_IGNORE_READ_ERROR;
> >>> +        }
> >>>       retval = ext2fs_get_pathname(fs, dir, ino, &path);
> >>> +        if (ctx)
> >>> +            ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR |
> >>> +                    (flags & E2F_FLAG_IGNORE_READ_ERROR);
> >>> +    }
> >>>   if (!fs || retval)
> >>>       fputs("???", f);
> >>>   else {
> >>> diff --git a/e2fsck/pass1.c b/e2fsck/pass1.c
> >>> index 308a95a..760fbde 100644
> >>> --- a/e2fsck/pass1.c
> >>> +++ b/e2fsck/pass1.c
> >>> @@ -510,6 +510,7 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>   int            extent_fs;
> >>>   int            inlinedata_fs;
> >>> 
> >>> +    ctx->flags |= E2F_FLAG_IGNORE_READ_ERROR;
> >>>   /*
> >>>    * If the mode looks OK, we believe it.  If the first block in
> >>>    * the i_block array is 0, this cannot be a directory. If the
> >>> @@ -519,7 +520,7 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>    */
> >>>   if (LINUX_S_ISDIR(inode->i_mode) || LINUX_S_ISREG(inode->i_mode) ||
> >>>       LINUX_S_ISLNK(inode->i_mode) || inode->i_block[0] == 0)
> >>> -        return;
> >>> +        goto out;
> >>> 
> >>>   /* 
> >>>    * Check the block numbers in the i_block array for validity:
> >>> @@ -552,13 +553,13 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>       struct ext2_dir_entry de;
> >>> 
> >>>       if (ext2fs_inline_data_size(ctx->fs, pctx->ino, &size))
> >>> -            return;
> >>> +            goto out;
> >>>       /*
> >>>        * If the size isn't a multiple of 4, it's probably not a
> >>>        * directory??
> >>>        */
> >>>       if (size & 3)
> >>> -            return;
> >>> +            goto out;
> >>>       /*
> >>>        * If the first 10 bytes don't look like a directory entry,
> >>>        * it's probably not a directory.
> >>> @@ -578,14 +579,14 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>            de.inode != 0) ||
> >>>           rec_len > EXT4_MIN_INLINE_DATA_SIZE -
> >>>                 EXT4_INLINE_DATA_DOTDOT_SIZE)
> >>> -            return;
> >>> +            goto out;
> >>>       /* device files never have a "system.data" entry */
> >>>       goto isdir;
> >>>   } else if (extent_fs && (inode->i_flags & EXT4_EXTENTS_FL)) {
> >>>       /* extent mapped */
> >>>       if  (ext2fs_bmap2(ctx->fs, pctx->ino, inode, 0, 0, 0, 0,
> >>>                &blk))
> >>> -            return;
> >>> +            goto out;
> >>>       /* device files are never extent mapped */
> >>>       not_device++;
> >>>   } else {
> >>> @@ -600,7 +601,7 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>               blk >= ext2fs_blocks_count(ctx->fs->super) ||
> >>>               ext2fs_fast_test_block_bitmap2(ctx->block_found_map,
> >>>                              blk))
> >>> -                return;    /* Invalid block, can't be dir */
> >>> +                goto out;    /* Invalid block, can't be dir */
> >>>       }
> >>>       blk = inode->i_block[0];
> >>>   }
> >>> @@ -612,45 +613,48 @@ static void check_is_really_dir(e2fsck_t ctx, struct problem_context *pctx,
> >>>    */
> >>>   if ((LINUX_S_ISCHR(inode->i_mode) || LINUX_S_ISBLK(inode->i_mode)) &&
> >>>       (inode->i_links_count == 1) && !not_device)
> >>> -        return;
> >>> +        goto out;
> >>> 
> >>>   /* read the first block */
> >>>   ehandler_operation(_("reading directory block"));
> >>>   retval = ext2fs_read_dir_block4(ctx->fs, blk, buf, 0, pctx->ino);
> >>>   ehandler_operation(0);
> >>>   if (retval)
> >>> -        return;
> >>> +        goto out;
> >>> 
> >>>   dirent = (struct ext2_dir_entry *) buf;
> >>>   retval = ext2fs_get_rec_len(ctx->fs, dirent, &rec_len);
> >>>   if (retval)
> >>> -        return;
> >>> +        goto out;
> >>>   if ((ext2fs_dirent_name_len(dirent) != 1) ||
> >>>       (dirent->name[0] != '.') ||
> >>>       (dirent->inode != pctx->ino) ||
> >>>       (rec_len < 12) ||
> >>>       (rec_len % 4) ||
> >>>       (rec_len >= ctx->fs->blocksize - 12))
> >>> -        return;
> >>> +        goto out;
> >>> 
> >>>   dirent = (struct ext2_dir_entry *) (buf + rec_len);
> >>>   retval = ext2fs_get_rec_len(ctx->fs, dirent, &rec_len);
> >>>   if (retval)
> >>> -        return;
> >>> +        goto out;
> >>>   if ((ext2fs_dirent_name_len(dirent) != 2) ||
> >>>       (dirent->name[0] != '.') ||
> >>>       (dirent->name[1] != '.') ||
> >>>       (rec_len < 12) ||
> >>>       (rec_len % 4))
> >>> -        return;
> >>> +        goto out;
> >>> 
> >>> isdir:
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>>   if (fix_problem(ctx, PR_1_TREAT_AS_DIRECTORY, pctx)) {
> >>>       inode->i_mode = (inode->i_mode & 07777) | LINUX_S_IFDIR;
> >>>       e2fsck_write_inode_full(ctx, pctx->ino, inode,
> >>>                   EXT2_INODE_SIZE(ctx->fs->super),
> >>>                   "check_is_really_dir");
> >>>   }
> >>> +out:
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>> }
> >>> 
> >>> void e2fsck_setup_tdb_icount(e2fsck_t ctx, int flags,
> >>> diff --git a/e2fsck/pass1b.c b/e2fsck/pass1b.c
> >>> index cd967f4..10136a6 100644
> >>> --- a/e2fsck/pass1b.c
> >>> +++ b/e2fsck/pass1b.c
> >>> @@ -234,7 +234,9 @@ void e2fsck_pass1_dupblocks(e2fsck_t ctx, char *block_buf)
> >>>   dict_set_allocator(&clstr_dict, NULL, cluster_dnode_free, NULL);
> >>> 
> >>>   init_resource_track(&rtrack, ctx->fs->io);
> >>> +    ctx->flags |= E2F_FLAG_IGNORE_READ_ERROR;
> >>>   pass1b(ctx, block_buf);
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>>   print_resource_track(ctx, "Pass 1b", &rtrack, ctx->fs->io);
> >>> 
> >>>   init_resource_track(&rtrack, ctx->fs->io);
> >>> @@ -242,7 +244,9 @@ void e2fsck_pass1_dupblocks(e2fsck_t ctx, char *block_buf)
> >>>   print_resource_track(ctx, "Pass 1c", &rtrack, ctx->fs->io);
> >>> 
> >>>   init_resource_track(&rtrack, ctx->fs->io);
> >>> +    ctx->flags |= E2F_FLAG_IGNORE_READ_ERROR;
> >>>   pass1d(ctx, block_buf);
> >>> +    ctx->flags &= ~E2F_FLAG_IGNORE_READ_ERROR;
> >>>   print_resource_track(ctx, "Pass 1d", &rtrack, ctx->fs->io);
> >>> 
> >>>   /*
> >>> 
> >>> --
> >>> 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-ext4" in
> >> the body of a message to majordomo@xxxxxxxxxxxxxxx
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 
> 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




[Index of Archives]     [Reiser Filesystem Development]     [Ceph FS]     [Kernel Newbies]     [Security]     [Netfilter]     [Bugtraq]     [Linux FS]     [Yosemite National Park]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Samba]     [Device Mapper]     [Linux Media]

  Powered by Linux