libburnia/libisofs
2
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Bug fix ticket 144: File size damage by non-unique inode numbers of

Browse Source 
pre-RRIP 1.12 images

Replaced the extent-LBA based inode number by a static 32 bit counter.
I deem it safe because if the numbers repeat within a single image
reading, then we are doomed anyway. I banned inode number 0 just in
case it has a meaning.

TODO: what if ER says IEEE_1282 but the PX field has only 36 bytes ?
This commit is contained in:
Thomas Schmitt 2008-11-25 14:34:47 +01:00
parent 0026c93cd4
commit 28e8936b4f
1 changed files with 32 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
libisofs/fs_image.c
View File

@ -1206,7 +1206,10 @@ int iso_file_source_new_ifs(IsoImageFilesystem *fs, IsoFileSource *parent,
"Sparse files not supported."); "Sparse files not supported.");
break; break;
} else if (SUSP_SIG(sue, 'R', 'R')) { } else if (SUSP_SIG(sue, 'R', 'R')) {
/* TODO I've seen this RR on mkisofs images. what's this? */ /* This was an optional flag byte in RRIP 1.09 which told the
reader what other RRIP fields to expect.
mkisofs emits it. We don't.
*/
continue; continue;
} else if (SUSP_SIG(sue, 'S', 'P')) { } else if (SUSP_SIG(sue, 'S', 'P')) {
/* /*
@ -1385,13 +1388,41 @@ int iso_file_source_new_ifs(IsoImageFilesystem *fs, IsoFileSource *parent,
return ISO_SUCCESS; return ISO_SUCCESS;
} }
/* ts Nov 25 2008: TODO
This seems not fully consistent with read_rr_PX() which decides
by (px->len_sue[0] == 44) whether an inode number is present or not.
What if read_rr_PX finds a PX of length 36 in a IEEE_1282 image ?
It is illegal but could confuse the image by duplicate inode numbers.
Regrettably it is not enough to just use single default numbers.
If only one number misses in the image, then all would need to be
defaulted by the following iso_global_inode code.
Why do duplicate inode numbers confuse the file lengths, anyway ?
(See ticket 144)
*/
if (fsdata->rr != RR_EXT_112) { if (fsdata->rr != RR_EXT_112) {
/* /*
* Only RRIP 1.12 provides valid inode numbers. If not, it is not easy * Only RRIP 1.12 provides valid inode numbers. If not, it is not easy
* to generate those serial numbers, and we use extend block instead. * to generate those serial numbers, and we use extend block instead.
* It BREAKS POSIX SEMANTICS, but its suitable for our needs * It BREAKS POSIX SEMANTICS, but its suitable for our needs
*/ */
#define Libisofs_patch_ticket_144 yes
#ifdef Libisofs_patch_ticket_144
/* Trying to ensure unique inode numbers */
{
static ino_t iso_global_inode= 0;
if(iso_global_inode <= 0)
iso_global_inode= 1;
atts.st_ino = iso_global_inode++;
}
#else
/* Ticket 144: This produces duplicate numbers with empty files.
*/
atts.st_ino = (ino_t) iso_read_bb(record->block, 4, NULL); atts.st_ino = (ino_t) iso_read_bb(record->block, 4, NULL);
#endif
if (fsdata->rr == 0) { if (fsdata->rr == 0) {
atts.st_nlink = 1; atts.st_nlink = 1;
} }