diff options
authorH. Peter Anvin <hpa@linux.intel.com>2016-02-10 13:47:06 -0800
committerH. Peter Anvin <hpa@linux.intel.com>2016-02-17 15:25:11 -0800
commitc1cc1bcef1e23ecb9f650c75f9dafc7e61101218 (patch)
parent875001667b9cbb1af421ce3e84feb8ce548dc5e4 (diff)
getfssec: update comment explaining next_extent() semantics
Update comments which explains next_extent() semantics to be slightly clearer, and also correct the statement that next_extent can be NULL. Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
1 files changed, 8 insertions, 3 deletions
diff --git a/core/fs/getfssec.c b/core/fs/getfssec.c
index e099b64e..2ea122d9 100644
--- a/core/fs/getfssec.c
+++ b/core/fs/getfssec.c
@@ -39,18 +39,23 @@
* will store the initial sector number into inode->next_extent.lstart
* on return.)
+ * If inode->next_extent.pstart is EXTENT_ZERO, then no disk I/O is
+ * performed, and the data in the extent is all zero.
+ *
* If inode->next_extent.len != 0 on entry then the routine is allowed
* to assume inode->next_extent contains valid data from the previous
* usage, which can be used for optimization purposes.
* If the filesystem can map the entire file as a single extent
* (e.g. iso9660), then the filesystem can simply insert the extent
- * information into inode->next_extent at searchdir/iget time, and leave
- * next_extent() as NULL.
+ * information into inode->next_extent at searchdir/iget time, and point
+ * next_extent() to the generic function no_next_extent().
* Note: the filesystem driver is not required to do extent coalescing,
* if that is difficult to do; this routine will perform extent lookahead
- * and coalescing.
+ * and coalescing. However, if the filesystem can do extent coalescing
+ * very cheaply by using filesystem-specific knowledge, then that is
+ * preferred (e.g. FAT).
#include <dprintf.h>