Re: [PATCH] mm/sl[aou]b: make kfree() aware of error pointers

[Jiri Kosina]

On Wed, 10 Sep 2014, Theodore Ts'o wrote:

> So I wouldn't be so sure that we don't have these sorts of bugs hiding
> somewhere; and it's extremely easy for them to sneak in.  That being
> said, I'm not in favor of making changes to kfree; I'd much rather
> depending on better testing and static checkers to fix them, since
> kfree *is* a hot path.

I of course have no objections to this check being added to whatever 
static checker, that would be very welcome improvement.

Still, I believe that kernel shouldn't be just ignoring kfree(ERR_PTR) 
happening. Would something like the below be more acceptable?



From: Jiri Kosina <jkosina@xxxxxxx>
Subject: [PATCH] mm/sl[aou]b: make kfree() aware of error pointers

Freeing if ERR_PTR is not covered by ZERO_OR_NULL_PTR() check already
present in kfree(), but it happens in the wild and has disastrous effects.

Issue a warning and don't proceed trying to free the memory if
CONFIG_DEBUG_SLAB is set.

Inspired by a9cfcd63e8d ("ext4: avoid trying to kfree an ERR_PTR pointer").

Signed-off-by: Jiri Kosina <jkosina@xxxxxxx>
---
 mm/slab.c | 6 ++++++
 mm/slob.c | 7 ++++++-
 mm/slub.c | 7 ++++++-
 3 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/mm/slab.c b/mm/slab.c
index a467b30..6f49d6b 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3612,6 +3612,12 @@ void kfree(const void *objp)
 
 	trace_kfree(_RET_IP_, objp);
 
+#ifdef CONFIG_DEBUG_SLAB
+	if (unlikely(IS_ERR(objp))) {
+			WARN(1, "trying to free ERR_PTR\n");
+			return;
+	}
+#endif
 	if (unlikely(ZERO_OR_NULL_PTR(objp)))
 		return;
 	local_irq_save(flags);
diff --git a/mm/slob.c b/mm/slob.c
index 21980e0..66422a0 100644
--- a/mm/slob.c
+++ b/mm/slob.c
@@ -488,7 +488,12 @@ void kfree(const void *block)
 	struct page *sp;
 
 	trace_kfree(_RET_IP_, block);
-
+#ifdef CONFIG_DEBUG_SLAB
+	if (unlikely(IS_ERR(block))) {
+		WARN(1, "trying to free ERR_PTR\n");
+		return;
+	}
+#endif
 	if (unlikely(ZERO_OR_NULL_PTR(block)))
 		return;
 	kmemleak_free(block);
diff --git a/mm/slub.c b/mm/slub.c
index 3e8afcc..21155ae 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -3337,7 +3337,12 @@ void kfree(const void *x)
 	void *object = (void *)x;
 
 	trace_kfree(_RET_IP_, x);
-
+#ifdef CONFIG_DEBUG_SLAB
+	if (unlikely(IS_ERR(x))) {
+		WARN(1, "trying to free ERR_PTR\n");
+		return;
+	}
+#endif
 	if (unlikely(ZERO_OR_NULL_PTR(x)))
 		return;
 

-- 
Jiri Kosina
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/