The patch titled
Subject: mm/slab.c: kmemleak no scan alien caches
has been added to the -mm tree. Its filename is
slab-kmemleak-no-scan-alien-caches.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/slab-kmemleak-no-scan-alien-caches.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/slab-kmemleak-no-scan-alien-caches.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Qian Cai <cai@xxxxxx>
Subject: mm/slab.c: kmemleak no scan alien caches
Kmemleak throws endless warnings during boot due to in
__alloc_alien_cache(),
alc = kmalloc_node(memsize, gfp, node);
init_arraycache(&alc->ac, entries, batch);
kmemleak_no_scan(ac);
Kmemleak does not track the array cache (alc->ac) but the alien cache
(alc) instead, so let it track the later by lifting kmemleak_no_scan() out
of init_arraycache().
There is another place that calls init_arraycache(), but
alloc_kmem_cache_cpus() uses the percpu allocation where will never be
considered as a leak.
[ 32.258841] kmemleak: Found object by alias at 0xffff8007b9aa7e38
[ 32.258847] CPU: 190 PID: 1 Comm: swapper/0 Not tainted 5.0.0-rc2+ #2
[ 32.258851] Call trace:
[ 32.258858] dump_backtrace+0x0/0x168
[ 32.258863] show_stack+0x24/0x30
[ 32.258868] dump_stack+0x88/0xb0
[ 32.258873] lookup_object+0x84/0xac
[ 32.258877] find_and_get_object+0x84/0xe4
[ 32.258882] kmemleak_no_scan+0x74/0xf4
[ 32.258887] setup_kmem_cache_node+0x2b4/0x35c
[ 32.258892] __do_tune_cpucache+0x250/0x2d4
[ 32.258896] do_tune_cpucache+0x4c/0xe4
[ 32.258901] enable_cpucache+0xc8/0x110
[ 32.258905] setup_cpu_cache+0x40/0x1b8
[ 32.258909] __kmem_cache_create+0x240/0x358
[ 32.258913] create_cache+0xc0/0x198
[ 32.258918] kmem_cache_create_usercopy+0x158/0x20c
[ 32.258922] kmem_cache_create+0x50/0x64
[ 32.258928] fsnotify_init+0x58/0x6c
[ 32.258932] do_one_initcall+0x194/0x388
[ 32.258937] kernel_init_freeable+0x668/0x688
[ 32.258941] kernel_init+0x18/0x124
[ 32.258946] ret_from_fork+0x10/0x18
[ 32.258950] kmemleak: Object 0xffff8007b9aa7e00 (size 256):
[ 32.258954] kmemleak: comm "swapper/0", pid 1, jiffies 4294697137
[ 32.258958] kmemleak: min_count = 1
[ 32.258962] kmemleak: count = 0
[ 32.258965] kmemleak: flags = 0x1
[ 32.258969] kmemleak: checksum = 0
[ 32.258972] kmemleak: backtrace:
[ 32.258977] kmemleak_alloc+0x84/0xb8
[ 32.258982] kmem_cache_alloc_node_trace+0x31c/0x3a0
[ 32.258987] __kmalloc_node+0x58/0x78
[ 32.258991] setup_kmem_cache_node+0x26c/0x35c
[ 32.258996] __do_tune_cpucache+0x250/0x2d4
[ 32.259001] do_tune_cpucache+0x4c/0xe4
[ 32.259005] enable_cpucache+0xc8/0x110
[ 32.259010] setup_cpu_cache+0x40/0x1b8
[ 32.259014] __kmem_cache_create+0x240/0x358
[ 32.259018] create_cache+0xc0/0x198
[ 32.259022] kmem_cache_create_usercopy+0x158/0x20c
[ 32.259026] kmem_cache_create+0x50/0x64
[ 32.259031] fsnotify_init+0x58/0x6c
[ 32.259035] do_one_initcall+0x194/0x388
[ 32.259039] kernel_init_freeable+0x668/0x688
[ 32.259043] kernel_init+0x18/0x124
[ 32.259048] kmemleak: Not scanning unknown object at 0xffff8007b9aa7e38
[ 32.259052] CPU: 190 PID: 1 Comm: swapper/0 Not tainted 5.0.0-rc2+ #2
[ 32.259056] Call trace:
[ 32.259060] dump_backtrace+0x0/0x168
[ 32.259065] show_stack+0x24/0x30
[ 32.259070] dump_stack+0x88/0xb0
[ 32.259074] kmemleak_no_scan+0x90/0xf4
[ 32.259078] setup_kmem_cache_node+0x2b4/0x35c
[ 32.259083] __do_tune_cpucache+0x250/0x2d4
[ 32.259088] do_tune_cpucache+0x4c/0xe4
[ 32.259092] enable_cpucache+0xc8/0x110
[ 32.259096] setup_cpu_cache+0x40/0x1b8
[ 32.259100] __kmem_cache_create+0x240/0x358
[ 32.259104] create_cache+0xc0/0x198
[ 32.259108] kmem_cache_create_usercopy+0x158/0x20c
[ 32.259112] kmem_cache_create+0x50/0x64
[ 32.259116] fsnotify_init+0x58/0x6c
[ 32.259120] do_one_initcall+0x194/0x388
[ 32.259125] kernel_init_freeable+0x668/0x688
[ 32.259129] kernel_init+0x18/0x124
[ 32.259133] ret_from_fork+0x10/0x18
Link: http://lkml.kernel.org/r/20190129184518.39808-1-cai@xxxxxx
Fixes: 1fe00d50a9e8 (slab: factor out initialization of array cache)
Signed-off-by: Qian Cai <cai@xxxxxx>
Cc: Christoph Lameter <cl@xxxxxxxxx>
Cc: Pekka Enberg <penberg@xxxxxxxxxx>
Cc: David Rientjes <rientjes@xxxxxxxxxx>
Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/slab.c | 17 +++++++++--------
1 file changed, 9 insertions(+), 8 deletions(-)
--- a/mm/slab.c~slab-kmemleak-no-scan-alien-caches
+++ a/mm/slab.c
@@ -550,14 +550,6 @@ static void start_cpu_timer(int cpu)
static void init_arraycache(struct array_cache *ac, int limit, int batch)
{
- /*
- * The array_cache structures contain pointers to free object.
- * However, when such objects are allocated or transferred to another
- * cache the pointers are not cleared and they could be counted as
- * valid references during a kmemleak scan. Therefore, kmemleak must
- * not scan such objects.
- */
- kmemleak_no_scan(ac);
if (ac) {
ac->avail = 0;
ac->limit = limit;
@@ -573,6 +565,14 @@ static struct array_cache *alloc_arrayca
struct array_cache *ac = NULL;
ac = kmalloc_node(memsize, gfp, node);
+ /*
+ * The array_cache structures contain pointers to free object.
+ * However, when such objects are allocated or transferred to another
+ * cache the pointers are not cleared and they could be counted as
+ * valid references during a kmemleak scan. Therefore, kmemleak must
+ * not scan such objects.
+ */
+ kmemleak_no_scan(ac);
init_arraycache(ac, entries, batchcount);
return ac;
}
@@ -667,6 +667,7 @@ static struct alien_cache *__alloc_alien
alc = kmalloc_node(memsize, gfp, node);
if (alc) {
+ kmemleak_no_scan(alc);
init_arraycache(&alc->ac, entries, batch);
spin_lock_init(&alc->lock);
}
_
Patches currently in -mm which might be from cai@xxxxxx are
mm-page_owner-fix-for-deferred-struct-page-init.patch
x86_64-increase-stack-size-for-kasan_extra.patch
mm-hotplug-invalid-pfns-from-pfn_to_online_page.patch
slab-kmemleak-no-scan-alien-caches.patch
page_poison-plays-nicely-with-kasan.patch
signal-allow-the-null-signal-in-rt_sigqueueinfo.patch
