On 32-bit systems, cheking for IS_ERR_VALUE(pos) is not
correct. Expanding that code we get (loff_t cast is added for clarity):
(loff_t)pos >= (unsigned long)-MAX_ERRNO
given that loff_t is a 64-bit signed value, any perfectly valid seek
offset that is greater than 0xffffc000 will result in false
positive. Change the logic to check if position returned by
fsdrv->lseek() is what's been requested. If it is, we can assume that
operation was succesfull. If not, that's likely means failure and
return value is a negative error code.
This should accomodate both 32-bit systems, where we /dev/mem doesn't
present any range problems, as well as 64-bit systems where both file
offset and size of /dev/mem couldn't really be correctly captured by
loff_t and we have to rely on 2's complement and overflow.
Signed-off-by: Andrey Smirnov <andrew.smirnov@xxxxxxxxx>
---
fs/fs.c | 13 +++++--------
1 file changed, 5 insertions(+), 8 deletions(-)
diff --git a/fs/fs.c b/fs/fs.c
index a304bf186..6a62fb98b 100644
--- a/fs/fs.c
+++ b/fs/fs.c
@@ -405,8 +405,7 @@ loff_t lseek(int fildes, loff_t offset, int whence)
{
struct fs_driver_d *fsdrv;
FILE *f;
- loff_t pos;
- int ret;
+ loff_t pos, ret;
if (check_fd(fildes))
return -1;
@@ -442,13 +441,11 @@ loff_t lseek(int fildes, loff_t offset, int whence)
goto out;
}
- pos = fsdrv->lseek(&f->fsdev->dev, f, pos);
- if (IS_ERR_VALUE(pos)) {
- errno = -pos;
- return -1;
- }
+ ret = fsdrv->lseek(&f->fsdev->dev, f, pos);
+ if (ret != pos)
+ goto out;
- return pos;
+ return ret;
out:
if (ret)
--
2.20.1
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
