[mm,03/11] kasan: improve kasan_non_canonical_hook
Commit Message
From: Andrey Konovalov <andreyknvl@google.com>
Make kasan_non_canonical_hook to be more sure in its report (i.e. say
"probably" instead of "maybe") if the address belongs to the shadow memory
region for kernel addresses.
Also use the kasan_shadow_to_mem helper to calculate the original address.
Also improve the comments in kasan_non_canonical_hook.
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
mm/kasan/kasan.h | 6 ++++++
mm/kasan/report.c | 34 ++++++++++++++++++++--------------
2 files changed, 26 insertions(+), 14 deletions(-)
@@ -307,6 +307,12 @@ struct kasan_stack_ring {
#if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
+static __always_inline bool addr_in_shadow(const void *addr)
+{
+ return addr >= (void *)KASAN_SHADOW_START &&
+ addr < (void *)KASAN_SHADOW_END;
+}
+
#ifndef kasan_shadow_to_mem
static inline const void *kasan_shadow_to_mem(const void *shadow_addr)
{
@@ -635,37 +635,43 @@ void kasan_report_async(void)
#if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
/*
- * With CONFIG_KASAN_INLINE, accesses to bogus pointers (outside the high
- * canonical half of the address space) cause out-of-bounds shadow memory reads
- * before the actual access. For addresses in the low canonical half of the
- * address space, as well as most non-canonical addresses, that out-of-bounds
- * shadow memory access lands in the non-canonical part of the address space.
- * Help the user figure out what the original bogus pointer was.
+ * With compiler-based KASAN modes, accesses to bogus pointers (outside of the
+ * mapped kernel address space regions) cause faults when KASAN tries to check
+ * the shadow memory before the actual memory access. This results in cryptic
+ * GPF reports, which are hard for users to interpret. This hook helps users to
+ * figure out what the original bogus pointer was.
*/
void kasan_non_canonical_hook(unsigned long addr)
{
unsigned long orig_addr;
const char *bug_type;
+ /*
+ * All addresses that came as a result of the memory-to-shadow mapping
+ * (even for bogus pointers) must be >= KASAN_SHADOW_OFFSET.
+ */
if (addr < KASAN_SHADOW_OFFSET)
return;
- orig_addr = (addr - KASAN_SHADOW_OFFSET) << KASAN_SHADOW_SCALE_SHIFT;
+ orig_addr = (unsigned long)kasan_shadow_to_mem((void *)addr);
+
/*
* For faults near the shadow address for NULL, we can be fairly certain
* that this is a KASAN shadow memory access.
- * For faults that correspond to shadow for low canonical addresses, we
- * can still be pretty sure - that shadow region is a fairly narrow
- * chunk of the non-canonical address space.
- * But faults that look like shadow for non-canonical addresses are a
- * really large chunk of the address space. In that case, we still
- * print the decoded address, but make it clear that this is not
- * necessarily what's actually going on.
+ * For faults that correspond to the shadow for low or high canonical
+ * addresses, we can still be pretty sure: these shadow regions are a
+ * fairly narrow chunk of the address space.
+ * But the shadow for non-canonical addresses is a really large chunk
+ * of the address space. For this case, we still print the decoded
+ * address, but make it clear that this is not necessarily what's
+ * actually going on.
*/
if (orig_addr < PAGE_SIZE)
bug_type = "null-ptr-deref";
else if (orig_addr < TASK_SIZE)
bug_type = "probably user-memory-access";
+ else if (addr_in_shadow((void *)addr))
+ bug_type = "probably wild-memory-access";
else
bug_type = "maybe wild-memory-access";
pr_alert("KASAN: %s in range [0x%016lx-0x%016lx]\n", bug_type,