linux-kernel - [PATCH v7 6/9] kfence, kasan: make KFENCE compatible with KASAN

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Tue,  3 Nov 2020 18:58:38 +0100
From:   Marco Elver <elver@...gle.com>
To:     elver@...gle.com, akpm@...ux-foundation.org, glider@...gle.com
Cc:     hpa@...or.com, paulmck@...nel.org, andreyknvl@...gle.com,
        aryabinin@...tuozzo.com, luto@...nel.org, bp@...en8.de,
        catalin.marinas@....com, cl@...ux.com, dave.hansen@...ux.intel.com,
        rientjes@...gle.com, dvyukov@...gle.com, edumazet@...gle.com,
        gregkh@...uxfoundation.org, hdanton@...a.com, mingo@...hat.com,
        jannh@...gle.com, Jonathan.Cameron@...wei.com, corbet@....net,
        iamjoonsoo.kim@....com, joern@...estorage.com,
        keescook@...omium.org, mark.rutland@....com, penberg@...nel.org,
        peterz@...radead.org, sjpark@...zon.com, tglx@...utronix.de,
        vbabka@...e.cz, will@...nel.org, x86@...nel.org,
        linux-doc@...r.kernel.org, linux-kernel@...r.kernel.org,
        kasan-dev@...glegroups.com, linux-arm-kernel@...ts.infradead.org,
        linux-mm@...ck.org
Subject: [PATCH v7 6/9] kfence, kasan: make KFENCE compatible with KASAN

From: Alexander Potapenko <glider@...gle.com>

Make KFENCE compatible with KASAN. Currently this helps test KFENCE
itself, where KASAN can catch potential corruptions to KFENCE state, or
other corruptions that may be a result of freepointer corruptions in the
main allocators.

Reviewed-by: Dmitry Vyukov <dvyukov@...gle.com>
Reviewed-by: Jann Horn <jannh@...gle.com>
Co-developed-by: Marco Elver <elver@...gle.com>
Signed-off-by: Marco Elver <elver@...gle.com>
Signed-off-by: Alexander Potapenko <glider@...gle.com>
---
v7:
* Remove EXPERT restriction for enabling KASAN+KFENCE. In future, MTE-based KASAN
  without stack traces will benefit from having KFENCE (which has stack
  traces). Removing EXPERT restriction allows this for production
  builds. The Kconfig help-text should still make it clear that in most
  cases KFENCE+KASAN does not make sense.
* Also skip kasan_poison_shadow() if KFENCE object. It turns out that
  kernel/scs.c is a user of kasan_{poison,unpoison}_object_data().
* Add Jann's Reviewed-by.

v5:
* Also guard kasan_unpoison_shadow with is_kfence_address(), as it may
  be called from SL*B internals, currently ksize().
* Make kasan_record_aux_stack() compatible with KFENCE, which may be
  called from outside KASAN runtime.
---
 lib/Kconfig.kfence |  2 +-
 mm/kasan/common.c  | 19 +++++++++++++++++++
 mm/kasan/generic.c |  3 ++-
 3 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/lib/Kconfig.kfence b/lib/Kconfig.kfence
index b209cd02042b..d2e3c6724226 100644
--- a/lib/Kconfig.kfence
+++ b/lib/Kconfig.kfence
@@ -5,7 +5,7 @@ config HAVE_ARCH_KFENCE
 
 menuconfig KFENCE
 	bool "KFENCE: low-overhead sampling-based memory safety error detector"
-	depends on HAVE_ARCH_KFENCE && !KASAN && (SLAB || SLUB)
+	depends on HAVE_ARCH_KFENCE && (SLAB || SLUB)
 	depends on JUMP_LABEL # To ensure performance, require jump labels
 	select STACKTRACE
 	help
diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index 950fd372a07e..de92da1b637a 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -18,6 +18,7 @@
 #include <linux/init.h>
 #include <linux/kasan.h>
 #include <linux/kernel.h>
+#include <linux/kfence.h>
 #include <linux/kmemleak.h>
 #include <linux/linkage.h>
 #include <linux/memblock.h>
@@ -124,6 +125,10 @@ void kasan_poison_shadow(const void *address, size_t size, u8 value)
 	 */
 	address = reset_tag(address);
 
+	/* Skip KFENCE memory if called explicitly outside of sl*b. */
+	if (is_kfence_address(address))
+		return;
+
 	shadow_start = kasan_mem_to_shadow(address);
 	shadow_end = kasan_mem_to_shadow(address + size);
 
@@ -141,6 +146,14 @@ void kasan_unpoison_shadow(const void *address, size_t size)
 	 */
 	address = reset_tag(address);
 
+	/*
+	 * Skip KFENCE memory if called explicitly outside of sl*b. Also note
+	 * that calls to ksize(), where size is not a multiple of machine-word
+	 * size, would otherwise poison the invalid portion of the word.
+	 */
+	if (is_kfence_address(address))
+		return;
+
 	kasan_poison_shadow(address, size, tag);
 
 	if (size & KASAN_SHADOW_MASK) {
@@ -396,6 +409,9 @@ static bool __kasan_slab_free(struct kmem_cache *cache, void *object,
 	tagged_object = object;
 	object = reset_tag(object);
 
+	if (is_kfence_address(object))
+		return false;
+
 	if (unlikely(nearest_obj(cache, virt_to_head_page(object), object) !=
 	    object)) {
 		kasan_report_invalid_free(tagged_object, ip);
@@ -444,6 +460,9 @@ static void *__kasan_kmalloc(struct kmem_cache *cache, const void *object,
 	if (unlikely(object == NULL))
 		return NULL;
 
+	if (is_kfence_address(object))
+		return (void *)object;
+
 	redzone_start = round_up((unsigned long)(object + size),
 				KASAN_SHADOW_SCALE_SIZE);
 	redzone_end = round_up((unsigned long)object + cache->object_size,
diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index 248264b9cb76..1069ecd1cd55 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c
@@ -21,6 +21,7 @@
 #include <linux/init.h>
 #include <linux/kasan.h>
 #include <linux/kernel.h>
+#include <linux/kfence.h>
 #include <linux/kmemleak.h>
 #include <linux/linkage.h>
 #include <linux/memblock.h>
@@ -332,7 +333,7 @@ void kasan_record_aux_stack(void *addr)
 	struct kasan_alloc_meta *alloc_info;
 	void *object;
 
-	if (!(page && PageSlab(page)))
+	if (is_kfence_address(addr) || !(page && PageSlab(page)))
 		return;
 
 	cache = page->slab_cache;
-- 
2.29.1.341.ge80a0c044ae-goog