efs: fix direct extent lookup for indirect extents
PATCH 1/1 efs: fix direct extent lookup for indirect extents
The loop that finds the direct extent containing an indirect extent used cur < ibase as its condition. Since ibase starts at zero, the loop never advances for normal non-negative cur values, so only the first direct extent is considered.
Iterate over the direct extents and stop when cur falls within the range covered by the current direct extent.
--- fs/efs/inode.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)
fs/efs/inode.c +7 -2
@@ -254,9 +254,14 @@ efs_block_t efs_map_block(struct inode *inode, efs_block_t block) {* */ ibase =0 ; -for (dirext =0 ; cur < ibase && dirext < direxts; dirext++) { - ibase += in->extents[dirext].cooked.ex_length * +for (dirext =0 ; dirext < direxts; dirext++) { +int entries = in->extents[dirext].cooked.ex_length * (EFS_BLOCKSIZE /sizeof (efs_extent)); + +if (cur < ibase + entries) +break ; + + ibase += entries; }if (dirext == direxts) {
Signature · 1 line
2.39.5
Quoted 7 lines · Keshav Verma
The loop that finds the direct extent containing an indirect extent used cur < ibase as its condition. Since ibase starts at zero, the loop never advances for normal non-negative cur values, so only the first direct extent is considered.
Iterate over the direct extents and stop when cur falls within the range covered by the current direct extent.
Nah.
Quoted 5 lines · Matthew Wilcox
Nah.
Thanks, I missed your EFS removal patch.
I'll drop this one.
Thanks,
Keshav