[PATCH 3/3] kmemleak-test: drop __init to get better backtrace

Jim Cromie posted 3 patches 2 years, 9 months ago
[PATCH 3/3] kmemleak-test: drop __init to get better backtrace
Posted by Jim Cromie 2 years, 9 months ago
Drop the __init on kmemleak_test_init().  With it, the storage is
reclaimed, but then the symbol isn't available for "%pS" rendering,
and the backtrace gets a bare pointer where the actual leak happened.

unreferenced object 0xffff88800a2b0800 (size 1024):
  comm "modprobe", pid 413, jiffies 4294953430
  hex dump (first 32 bytes):
    73 02 00 00 75 01 00 68 02 00 00 01 00 00 00 04  s...u..h........
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace (ck 603070071):
    [<00000000fabad728>] kmalloc_trace+0x26/0x90
    [<00000000ef738764>] 0xffffffffc02350a2
    [<00000000004e5795>] do_one_initcall+0x43/0x210
    [<00000000d768905e>] do_init_module+0x4a/0x210
    [<0000000087135ab5>] __do_sys_finit_module+0x93/0xf0
    [<000000004fcb1fa2>] do_syscall_64+0x34/0x80
    [<00000000c73c8d9d>] entry_SYSCALL_64_after_hwframe+0x46/0xb0

with __init gone, that trace entry renders like:

    [<00000000ef738764>] kmemleak_test_init+<offset>/<size>
---
 samples/kmemleak/kmemleak-test.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/samples/kmemleak/kmemleak-test.c b/samples/kmemleak/kmemleak-test.c
index 7b476eb8285f..6ced5ddd99d4 100644
--- a/samples/kmemleak/kmemleak-test.c
+++ b/samples/kmemleak/kmemleak-test.c
@@ -32,7 +32,7 @@ static DEFINE_PER_CPU(void *, kmemleak_test_pointer);
  * Some very simple testing. This function needs to be extended for
  * proper testing.
  */
-static int __init kmemleak_test_init(void)
+static int kmemleak_test_init(void)
 {
 	struct test_node *elem;
 	int i;
-- 
2.40.0
Re: [PATCH 3/3] kmemleak-test: drop __init to get better backtrace
Posted by Catalin Marinas 2 years, 9 months ago
On Tue, Apr 25, 2023 at 04:24:46PM -0600, Jim Cromie wrote:
> Drop the __init on kmemleak_test_init().  With it, the storage is
> reclaimed, but then the symbol isn't available for "%pS" rendering,
> and the backtrace gets a bare pointer where the actual leak happened.
> 
> unreferenced object 0xffff88800a2b0800 (size 1024):
>   comm "modprobe", pid 413, jiffies 4294953430
>   hex dump (first 32 bytes):
>     73 02 00 00 75 01 00 68 02 00 00 01 00 00 00 04  s...u..h........
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>   backtrace (ck 603070071):
>     [<00000000fabad728>] kmalloc_trace+0x26/0x90
>     [<00000000ef738764>] 0xffffffffc02350a2
>     [<00000000004e5795>] do_one_initcall+0x43/0x210
>     [<00000000d768905e>] do_init_module+0x4a/0x210
>     [<0000000087135ab5>] __do_sys_finit_module+0x93/0xf0
>     [<000000004fcb1fa2>] do_syscall_64+0x34/0x80
>     [<00000000c73c8d9d>] entry_SYSCALL_64_after_hwframe+0x46/0xb0
> 
> with __init gone, that trace entry renders like:
> 
>     [<00000000ef738764>] kmemleak_test_init+<offset>/<size>

Acked-by: Catalin Marinas <catalin.marinas@arm.com>