Skip to contentSkip to conversation

Question regarding get_block() functions of filesystems.

DISCUSSION Manish Katiyar 1 participant 1 message Feb 17, 2008 VFS

lore
Share & tools
lore
#0 lore raw

Hi,

I can see at lot of places in the kernel code, sector_t and long have been used interchangeably. Since sector_t is defined as u64 or unsigned long
it will be issue for large file block numbers.

Searching google I found that there have been discussions regarding this and some patch submitted too for some modules (links below), but the current latest stable kernel
still seems to have this. As there appears to be printing issues with using sector_t can anyone tell whether we need the cleanup in existing Filesystem code or not ?

Previous patches :

1) http://lwn.net/Articles/8437/
2) http://osdir.com/ml/file-systems.ext2.devel/2006-05/msg00037.html

Below are some of the codes where they are used.

547 int ext2_get_block(struct inode *inode, sector_t iblock, struct
buffer_head *bh_result, int create)
548 {

 549  int err = -EIO;
 550  int offsets[4];
 551  Indirect chain[4];
 552  Indirect *partial;
 553  unsigned long goal;
 554  int left;
 555  int boundary = 0;
 556  int depth = ext2_block_to_path(inode, iblock, offsets, &boundary);

..........
...........

196 static int ext2_block_to_path(struct inode *inode,
197 long i_block, int offsets[4], int *boundary)
198 {
.........
.........

52 int V1_minix_get_block(struct inode * inode, long block,
53 struct buffer_head *bh_result, int create)
54 {
.............
.............

Signature · 10 lines
Thanks a lot in advance.

Thanks & Regards,
********************************************
Manish Katiyar  ( http://mkatiyar.googlepages.com )
3rd Floor, Fair Winds Block
EGL Software Park
Off Intermediate Ring Road
Bangalore 560071, India
***********************************************