* [PATCH tip/core/rcu 1/2] mm/slub: Fix backtrace of objects to handle redzone adjustment
2021-05-11 22:57 [PATCH tip/core/rcu 0/2] mem_dump_obj() updates for v5.14 Paul E. McKenney
@ 2021-05-11 22:58 ` Paul E. McKenney
2021-05-11 22:58 ` [PATCH tip/core/rcu 2/2] mm/slub: Add Support for free path information of an object Paul E. McKenney
1 sibling, 0 replies; 3+ messages in thread
From: Paul E. McKenney @ 2021-05-11 22:58 UTC (permalink / raw)
To: rcu
Cc: linux-kernel, kernel-team, mingo, jiangshanlai, akpm,
mathieu.desnoyers, josh, tglx, peterz, rostedt, dhowells,
edumazet, fweisbec, oleg, joel, Maninder Singh, Vlastimil Babka,
Vaneet Narang, Paul E . McKenney
From: Maninder Singh <maninder1.s@samsung.com>
This commit fixes commit 8e7f37f2aaa5 ("mm: Add mem_dump_obj() to print
source of memory block").
With current code, the backtrace of allocated object is incorrect:
/ # cat /proc/meminfo
[ 14.969843] slab kmalloc-64 start c8ab0140 data offset 64 pointer offset 0 size 64 allocated at 0x6b6b6b6b
[ 14.970635] 0x6b6b6b6b
[ 14.970794] 0x6b6b6b6b
[ 14.970932] 0x6b6b6b6b
[ 14.971077] 0x6b6b6b6b
[ 14.971202] 0x6b6b6b6b
[ 14.971317] 0x6b6b6b6b
[ 14.971423] 0x6b6b6b6b
[ 14.971635] 0x6b6b6b6b
[ 14.971740] 0x6b6b6b6b
[ 14.971871] 0x6b6b6b6b
[ 14.972229] 0x6b6b6b6b
[ 14.972363] 0x6b6b6b6b
[ 14.972505] 0xa56b6b6b
[ 14.972631] 0xbbbbbbbb
[ 14.972734] 0xc8ab0400
[ 14.972891] meminfo_proc_show+0x40/0x4fc
The reason is that the object address was not adjusted for the red zone.
With this fix, the backtrace is correct:
/ # cat /proc/meminfo
[ 14.870782] slab kmalloc-64 start c8ab0140 data offset 64 pointer offset 128 size 64 allocated at meminfo_proc_show+0x40/0x4f4
[ 14.871817] meminfo_proc_show+0x40/0x4f4
[ 14.872035] seq_read_iter+0x18c/0x4c4
[ 14.872229] proc_reg_read_iter+0x84/0xac
[ 14.872433] generic_file_splice_read+0xe8/0x17c
[ 14.872621] splice_direct_to_actor+0xb8/0x290
[ 14.872747] do_splice_direct+0xa0/0xe0
[ 14.872896] do_sendfile+0x2d0/0x438
[ 14.873044] sys_sendfile64+0x12c/0x140
[ 14.873229] ret_fast_syscall+0x0/0x58
[ 14.873372] 0xbe861de4
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Vaneet Narang <v.narang@samsung.com>
Signed-off-by: Maninder Singh <maninder1.s@samsung.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
mm/slub.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/mm/slub.c b/mm/slub.c
index feda53ae62ba..8f2d13508ec9 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -4002,6 +4002,7 @@ void kmem_obj_info(struct kmem_obj_info *kpp, void *object, struct page *page)
!(s->flags & SLAB_STORE_USER))
return;
#ifdef CONFIG_SLUB_DEBUG
+ objp = fixup_red_left(s, objp);
trackp = get_track(s, objp, TRACK_ALLOC);
kpp->kp_ret = (void *)trackp->addr;
#ifdef CONFIG_STACKTRACE
--
2.31.1.189.g2e36527f23
^ permalink raw reply [flat|nested] 3+ messages in thread* [PATCH tip/core/rcu 2/2] mm/slub: Add Support for free path information of an object
2021-05-11 22:57 [PATCH tip/core/rcu 0/2] mem_dump_obj() updates for v5.14 Paul E. McKenney
2021-05-11 22:58 ` [PATCH tip/core/rcu 1/2] mm/slub: Fix backtrace of objects to handle redzone adjustment Paul E. McKenney
@ 2021-05-11 22:58 ` Paul E. McKenney
1 sibling, 0 replies; 3+ messages in thread
From: Paul E. McKenney @ 2021-05-11 22:58 UTC (permalink / raw)
To: rcu
Cc: linux-kernel, kernel-team, mingo, jiangshanlai, akpm,
mathieu.desnoyers, josh, tglx, peterz, rostedt, dhowells,
edumazet, fweisbec, oleg, joel, Maninder Singh, Vlastimil Babka,
Vaneet Narang, Paul E . McKenney
From: Maninder Singh <maninder1.s@samsung.com>
This commit adds enables a stack dump for the last free of an object:
slab kmalloc-64 start c8ab0140 data offset 64 pointer offset 0 size 64 allocated at meminfo_proc_show+0x40/0x4fc
[ 20.192078] meminfo_proc_show+0x40/0x4fc
[ 20.192263] seq_read_iter+0x18c/0x4c4
[ 20.192430] proc_reg_read_iter+0x84/0xac
[ 20.192617] generic_file_splice_read+0xe8/0x17c
[ 20.192816] splice_direct_to_actor+0xb8/0x290
[ 20.193008] do_splice_direct+0xa0/0xe0
[ 20.193185] do_sendfile+0x2d0/0x438
[ 20.193345] sys_sendfile64+0x12c/0x140
[ 20.193523] ret_fast_syscall+0x0/0x58
[ 20.193695] 0xbeeacde4
[ 20.193822] Free path:
[ 20.193935] meminfo_proc_show+0x5c/0x4fc
[ 20.194115] seq_read_iter+0x18c/0x4c4
[ 20.194285] proc_reg_read_iter+0x84/0xac
[ 20.194475] generic_file_splice_read+0xe8/0x17c
[ 20.194685] splice_direct_to_actor+0xb8/0x290
[ 20.194870] do_splice_direct+0xa0/0xe0
[ 20.195014] do_sendfile+0x2d0/0x438
[ 20.195174] sys_sendfile64+0x12c/0x140
[ 20.195336] ret_fast_syscall+0x0/0x58
[ 20.195491] 0xbeeacde4
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Co-developed-by: Vaneet Narang <v.narang@samsung.com>
Signed-off-by: Vaneet Narang <v.narang@samsung.com>
Signed-off-by: Maninder Singh <maninder1.s@samsung.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
mm/slab.h | 1 +
mm/slab_common.c | 12 +++++++++++-
mm/slub.c | 7 +++++++
mm/util.c | 2 +-
4 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/mm/slab.h b/mm/slab.h
index 18c1927cd196..7189daa0c586 100644
--- a/mm/slab.h
+++ b/mm/slab.h
@@ -640,6 +640,7 @@ struct kmem_obj_info {
struct kmem_cache *kp_slab_cache;
void *kp_ret;
void *kp_stack[KS_ADDRS_COUNT];
+ void *kp_free_stack[KS_ADDRS_COUNT];
};
void kmem_obj_info(struct kmem_obj_info *kpp, void *object, struct page *page);
#endif
diff --git a/mm/slab_common.c b/mm/slab_common.c
index f8833d3e5d47..92e3aa78bb4d 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -564,7 +564,7 @@ EXPORT_SYMBOL_GPL(kmem_valid_obj);
* depends on the type of object and on how much debugging is enabled.
* For a slab-cache object, the fact that it is a slab object is printed,
* and, if available, the slab name, return address, and stack trace from
- * the allocation of that object.
+ * the allocation and last free path of that object.
*
* This function will splat if passed a pointer to a non-slab object.
* If you are not sure what type of object you have, you should instead
@@ -609,6 +609,16 @@ void kmem_dump_obj(void *object)
break;
pr_info(" %pS\n", kp.kp_stack[i]);
}
+
+ if (kp.kp_free_stack[0])
+ pr_cont(" Free path:\n");
+
+ for (i = 0; i < ARRAY_SIZE(kp.kp_free_stack); i++) {
+ if (!kp.kp_free_stack[i])
+ break;
+ pr_info(" %pS\n", kp.kp_free_stack[i]);
+ }
+
}
EXPORT_SYMBOL_GPL(kmem_dump_obj);
#endif
diff --git a/mm/slub.c b/mm/slub.c
index 8f2d13508ec9..deec894a1345 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -4011,6 +4011,13 @@ void kmem_obj_info(struct kmem_obj_info *kpp, void *object, struct page *page)
if (!kpp->kp_stack[i])
break;
}
+
+ trackp = get_track(s, objp, TRACK_FREE);
+ for (i = 0; i < KS_ADDRS_COUNT && i < TRACK_ADDRS_COUNT; i++) {
+ kpp->kp_free_stack[i] = (void *)trackp->addrs[i];
+ if (!kpp->kp_free_stack[i])
+ break;
+ }
#endif
#endif
}
diff --git a/mm/util.c b/mm/util.c
index a8bf17f18a81..0b6dd9d81da7 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -983,7 +983,7 @@ int __weak memcmp_pages(struct page *page1, struct page *page2)
* depends on the type of object and on how much debugging is enabled.
* For example, for a slab-cache object, the slab name is printed, and,
* if available, the return address and stack trace from the allocation
- * of that object.
+ * and last free path of that object.
*/
void mem_dump_obj(void *object)
{
--
2.31.1.189.g2e36527f23
^ permalink raw reply [flat|nested] 3+ messages in thread