@@ -124,11 +124,6 @@
#define BYTES_PER_POINTER sizeof(void *)
-/* GFP bitmask for kmemleak internal allocations */
-#define gfp_kmemleak_mask(gfp) (((gfp) & (GFP_KERNEL | GFP_ATOMIC)) | \
- __GFP_NORETRY | __GFP_NOMEMALLOC | \
- __GFP_NOWARN | __GFP_NOFAIL)
-
/* scanning area inside a memory block */
struct kmemleak_scan_area {
struct hlist_node node;
@@ -315,6 +310,30 @@ static void kmemleak_disable(void);
pr_warn(fmt, ##__VA_ARGS__); \
} while (0)
+/* GFP bitmask for kmemleak internal allocations */
+static inline gfp_t gfp_kmemleak_mask(gfp_t gfp)
+{
+ gfp = (gfp & (GFP_KERNEL | GFP_ATOMIC)) | __GFP_NORETRY |
+ __GFP_NOMEMALLOC | __GFP_NOWARN | __GFP_NOFAIL;
+
+/*
+ * PREEMPT_COUNT is set by either PREEMPT or DEBUG_ATOMIC_SLEEP which is
+ * normally found in a debug kernel just like kmemleak. Otherwise, it won't be
+ * able to catch irq unsafe spinlocks held.
+ */
+#ifdef CONFIG_PREEMPT_COUNT
+ /*
+ * The tracked memory was allocated successful, if the kmemleak object
+ * failed to allocate for some reasons, it ends up with the whole
+ * kmemleak disabled, so try it harder.
+ */
+ gfp |= ((in_atomic() || irqs_disabled()) ? GFP_ATOMIC :
+ __GFP_DIRECT_RECLAIM);
+#endif
+
+ return gfp;
+}
+
static void warn_or_seq_hex_dump(struct seq_file *seq, int prefix_type,
int rowsize, int groupsize, const void *buf,
size_t len, bool ascii)
Kmemleak could quickly fail to allocate an object structure and then disable itself below in a low-memory situation. For example, running a mmap() workload triggering swapping and OOM. This is especially problematic for running things like LTP testsuite where one OOM test case would disable the whole kmemleak and render the rest of test cases without kmemleak watching for leaking. Kmemleak allocation could fail even though the tracked memory is succeeded. Hence, it could still try to start a direct reclaim if it is not executed in an atomic context (spinlock, irq-handler etc), or a high-priority allocation in an atomic context as a last-ditch effort. Since kmemleak is a debug feature, it is unlikely to be used in production that memory resources is scarce where direct reclaim or high-priority atomic allocations should not be granted lightly. Unless there is a brave soul to reimplement the kmemleak to embed it's metadata into the tracked memory itself in a foreseeable future, this provides a good balance between enabling kmemleak in a low-memory situation and not introducing too much hackiness into the existing code for now. Another approach is to fail back the original allocation once kmemleak_alloc() failed, but there are too many call sites to deal with which makes it error-prone. kmemleak: Cannot allocate a kmemleak_object structure kmemleak: Kernel memory leak detector disabled kmemleak: Automatic memory scanning thread ended RIP: 0010:__alloc_pages_nodemask+0x242a/0x2ab0 Call Trace: alloc_pages_current+0xdb/0x1c0 allocate_slab+0x4d9/0x930 new_slab+0x46/0x70 ___slab_alloc+0x5d3/0x9c0 __slab_alloc+0x12/0x20 kmem_cache_alloc+0x30a/0x360 create_object+0x96/0x9a0 kmemleak_alloc+0x71/0xa0 kmem_cache_alloc+0x254/0x360 mempool_alloc_slab+0x3f/0x60 mempool_alloc+0x120/0x329 bio_alloc_bioset+0x1a8/0x510 get_swap_bio+0x107/0x470 __swap_writepage+0xab4/0x1650 swap_writepage+0x86/0xe0 Signed-off-by: Qian Cai <cai@lca.pw> --- v5: Move everything into gfp_kmemleak_mask(). Use PREEMPT_COUNT to catch irq unsafe spinlocks held. v4: Update the commit log. Fix a typo in comments per Christ. Consolidate the allocation. v3: Update the commit log. Simplify the code inspired by graph_trace_open() from ftrace. v2: Remove the needless checking for NULL objects in slab_post_alloc_hook() per Catalin. mm/kmemleak.c | 29 ++++++++++++++++++++++++----- 1 file changed, 24 insertions(+), 5 deletions(-)