Re: [PATCH RFC] random: fix syzkaller fuzzer test int overflow
From: Theodore Ts'o
Date: Sun Oct 29 2017 - 17:57:38 EST
On Sat, Oct 28, 2017 at 11:22:00AM +0800, Chen Feng wrote:
>
> I checked the ioctl. What's the purpose of RNDADDTOENTCNT ioctl to
> userspace?
It's a legacy ioctl which is probably not used anywhere; it's been
replaced by RNDADDENTROPY. It previously allows root to bump the
entropy estimate, but the right way to do this by rngd is to
atomically add entropy to the pool land and bump the entropy estimate
at the same time.
The UBSAN is harmless. The ioctl requires root, and the entropy_total
field, which is involved in the UBSAN, is only used in the first few
seconds of boot, to determine when the entropy pool has been
initialized. In general on desktop and servers this happens before
userspace has a chance to run.
In any case, here's a fix for this.
- Ted
commit 6f7034d0c52e21f30002b95126b6b98e4618dc57
Author: Theodore Ts'o <tytso@xxxxxxx>
Date: Sun Oct 29 14:17:26 2017 -0400
random: use a tighter cap in credit_entropy_bits_safe()
This fixes a harmless UBSAN where root could potentially end up
causing an overflow while bumping the entropy_total field (which is
ignored once the entropy pool has been initialized, and this generally
is completed during the boot sequence).
This is marginal for the stable kernel series, but it's a really
trivial patch, and it UBSAN warning that might cause security folks to
get overly excited for no reason.
Signed-off-by: Theodore Ts'o <tytso@xxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
diff --git a/drivers/char/random.c b/drivers/char/random.c
index 8ad92707e45f..ae8a2f829890 100644
--- a/drivers/char/random.c
+++ b/drivers/char/random.c
@@ -733,7 +733,7 @@ static void credit_entropy_bits(struct entropy_store *r, int nbits)
static int credit_entropy_bits_safe(struct entropy_store *r, int nbits)
{
- const int nbits_max = (int)(~0U >> (ENTROPY_SHIFT + 1));
+ const int nbits_max = r->poolinfo->poolwords * 32;
if (nbits < 0)
return -EINVAL;