Skip to contentSkip to conversation

isofs: fix minor filesystem corruption, take 2

PATCH In review Adam Greenblatt 2 participants 2 messages Jul 16, 2008 Isofs filesystem

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 isofs: fix minor filesystem corruption, take 2

Some iso9660 images contain files with rockridge data that is either incorrect or incompletely parsed. Prior to commit f2966632a134e865db3c819346a1dc7d96e05309, (included with kernel 2.6.13) the kernel ignored the rockridge data for these files, while still allowing the files to be accessed under their non-rockridge names. That commit inadvertently changed things so that files with invalid rockridge data could not be accessed at all. (I ran across the problem when comparing some old CDs with hard disk copies I had made long ago under kernel 2.4: a few of the files on the hard disk copies were no longer visible on the CDs.)

This change reverts to the pre-2.6.13 behavior.

---
Pekka Enberg correctly pointed out that there was a similar change made to parse_rock_ridge_inode_internal() that should also be reverted. Please ignore my previous patch, and use this patch instead. Thanks Pekka!

fs/isofs/rock.c 2008-07-15 21:52:43.000000000 -1000 renamed from fs/isofs/rock.c.orig 2008-07-13 11:51:29.000000000 -1000 +0 -0
@@ -209,6 +209,9 @@ repeat:
 
       while (rs.len > 2) { /* There may be one byte for padding 

somewhere */

               rr = (struct rock_ridge *)rs.chr;
+               /* Ignore rock ridge info if rr->len is out of range, but
+                  don't return -EIO because that would make the file
+                  invisible. */
               if (rr->len < 3)
                       goto out;       /* Something got screwed up here */
               sig = isonum_721(rs.chr);
@@ -216,8 +219,9 @@ repeat:
                       goto eio;
               rs.chr += rr->len;
               rs.len -= rr->len;
+               /* As above, ignore rock ridge info if rr->len is bogus. */
               if (rs.len < 0)
-                       goto eio;       /* corrupted isofs */
+                       goto out;       /* Something got screwed up here */
               switch (sig) {
               case SIG('R', 'R'):

@@ -307,6 +311,9 @@ parse_rock_ridge_inode_internal(struct i repeat:
while (rs.len > 2) { /* There may be one byte for padding
somewhere */

               rr = (struct rock_ridge *)rs.chr;
+               /* Ignore rock ridge info if rr->len is out of range, but
+                  don't return -EIO because that would make the file
+                  invisible. */
               if (rr->len < 3)
                       goto out;       /* Something got screwed up here */
               sig = isonum_721(rs.chr);
@@ -314,8 +321,9 @@ repeat:
                       goto eio;
               rs.chr += rr->len;
               rs.len -= rr->len;
+               /* As above, ignore rock ridge info if rr->len is bogus. */
               if (rs.len < 0)
-                       goto eio;       /* corrupted isofs */
+                       goto out;       /* Something got screwed up here */
               switch (sig) {

#ifndef CONFIG_ZISOFS /* No flag for SF or ZF */

Reviewed-by on v1 1/1
#1 lore raw

Hi Adam,

Quoted 16 lines · Adam Greenblatt

Some iso9660 images contain files with rockridge data that is either incorrect or
incompletely parsed. Prior to commit f2966632a134e865db3c819346a1dc7d96e05309,
(included with kernel 2.6.13) the kernel ignored the rockridge data for these files,
while still allowing the files to be accessed under their non-rockridge names. That
commit inadvertently changed things so that files with invalid rockridge data could
not be accessed at all. (I ran across the problem when comparing some old CDs with
hard disk copies I had made long ago under kernel 2.4: a few of the files on the hard
disk copies were no longer visible on the CDs.)

Quoted 8 lines

This change reverts to the pre-2.6.13 behavior.

---
Pekka Enberg correctly pointed out that there was a similar change made to parse_rock_ridge_inode_internal() that should also be reverted. Please ignore
my previous patch, and use this patch instead. Thanks Pekka!

Looks good to me.