All error handling paths, except these ones, branch to the 'bad' label in
the error handling path.
If not done, there is a memory leak and some sensitive data may be kept
around.
So, fix these error handling paths and also do the needed clean-up.
Fixes: e7f57d2c47e2 ("dm-inlinecrypt: add target for inline block device
encryption")
Signed-off-by: Christophe JAILLET <[email protected]>
Reviewed-by: Eric Biggers <[email protected]>
---
Compile tested only
Changes in v2:
- Fix another missing goto [Eric Biggers]
- Add R-b tag (received privately)
v1:
https://lore.kernel.org/all/2b6e6b71e66fbe90c75791669a9e16a36b48f7f6.1781983971.git.christophe.jail...@wanadoo.fr/
---
drivers/md/dm-inlinecrypt.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/md/dm-inlinecrypt.c b/drivers/md/dm-inlinecrypt.c
index be1b4aa8f28b..41293c18d10f 100644
--- a/drivers/md/dm-inlinecrypt.c
+++ b/drivers/md/dm-inlinecrypt.c
@@ -347,7 +347,8 @@ static int inlinecrypt_ctr(struct dm_target *ti, unsigned
int argc, char **argv)
err = get_key_size(&argv[1]);
if (err < 0) {
ti->error = "Cannot parse key size";
- return -EINVAL;
+ err = -EINVAL;
+ goto bad;
}
ctx->key_size = err;
@@ -398,6 +399,7 @@ static int inlinecrypt_ctr(struct dm_target *ti, unsigned
int argc, char **argv)
if (ctx->iv_offset & ((ctx->sector_size >> SECTOR_SHIFT) - 1)) {
ti->error = "Wrong alignment of iv_offset sector";
err = -EINVAL;
+ goto bad;
}
ctx->max_dun = (ctx->iv_offset + ti->len - 1) >>
--
2.54.0
