| Message ID | 1446911439-9235-1-git-send-email-penguin-kernel@I-love.SAKURA.ne.jp |
|---|---|
| State | Changes Requested, archived |
| Delegated to: | David Miller |
| Headers | show |
Tetsuo Handa wrote: > Commit 095dc8e0c3686d58 ("tcp: fix/cleanup inet_ehash_locks_alloc()") > silently changed from kmalloc() to kmalloc_array(). The latter has > overflow check whereas the former doesn't have. > > If nblocks * locksz might overflow, we need to do like > > - if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) > + if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) Oops, I meant - if (!hashinfo->ehash_locks) + if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) here. > hashinfo->ehash_locks = vmalloc(nblocks * locksz); > > because kmalloc_array() detects overflow and returns NULL. > But if nblocks * locksz is guaranteed not to overflow, there is > no need to use kmalloc_array(). > > Since I assume it won't overflow, use kmalloc() than kmalloc_array(). I don't know about possible value range. Please confirm that it can't overflow. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Date: Sun, 8 Nov 2015 00:58:50 +0900 > Tetsuo Handa wrote: >> Commit 095dc8e0c3686d58 ("tcp: fix/cleanup inet_ehash_locks_alloc()") >> silently changed from kmalloc() to kmalloc_array(). The latter has >> overflow check whereas the former doesn't have. >> >> If nblocks * locksz might overflow, we need to do like >> >> - if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) >> + if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) > > Oops, I meant > > - if (!hashinfo->ehash_locks) > + if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) > > here. > >> hashinfo->ehash_locks = vmalloc(nblocks * locksz); >> >> because kmalloc_array() detects overflow and returns NULL. >> But if nblocks * locksz is guaranteed not to overflow, there is >> no need to use kmalloc_array(). >> >> Since I assume it won't overflow, use kmalloc() than kmalloc_array(). > > I don't know about possible value range. > Please confirm that it can't overflow. Whether it can overflow or not, I don't like your change at all. If kmalloc_array() provides overflow protection, we want to keep using it, rather than reverting back to not checking for overflow. kmalloc_array() can fail for us at this stage for two reasons, either: 1) overflow 2) allocation size exceeds kmalloc() maximum alloc size so if you want to be completely perfect about all of this you need to add code to distinguish between these two cases as you suggest above. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index ccc5980..8f4ab27 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -648,8 +648,8 @@ int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo) /* no more locks than number of hash buckets */ nblocks = min(nblocks, hashinfo->ehash_mask + 1); - hashinfo->ehash_locks = kmalloc_array(nblocks, locksz, - GFP_KERNEL | __GFP_NOWARN); + hashinfo->ehash_locks = kmalloc(nblocks * locksz, + GFP_KERNEL | __GFP_NOWARN); if (!hashinfo->ehash_locks) hashinfo->ehash_locks = vmalloc(nblocks * locksz);
Commit 095dc8e0c3686d58 ("tcp: fix/cleanup inet_ehash_locks_alloc()") silently changed from kmalloc() to kmalloc_array(). The latter has overflow check whereas the former doesn't have. If nblocks * locksz might overflow, we need to do like - if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) + if (!hashinfo->ehash_locks && nblocks > SIZE_MAX / locksz) hashinfo->ehash_locks = vmalloc(nblocks * locksz); because kmalloc_array() detects overflow and returns NULL. But if nblocks * locksz is guaranteed not to overflow, there is no need to use kmalloc_array(). Since I assume it won't overflow, use kmalloc() than kmalloc_array(). Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> --- net/ipv4/inet_hashtables.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)