2013-09-30 17:51:33

by Larry Finger

Subject: 3.12-rc3 memory leak in module memstick

On 3.12-rc3, I detected a memory leak in module memstick. The kmemleak results
are listed below. The second output is after memstick was unloaded, which shows
that it is not a false positive from kmemleak.

larrylap:~ # echo scan > /sys/kernel/debug/kmemleak
larrylap:~ # cat /sys/kernel/debug/kmemleak
unreferenced object 0xffff8800ae85c190 (size 16):
comm "kworker/u4:3", pid 685, jiffies 4294916336 (age 2831.760s)
hex dump (first 16 bytes):
6d 65 6d 73 74 69 63 6b 30 00 00 00 00 00 00 00 memstick0.......
backtrace:
[<ffffffff8146a0d1>] kmemleak_alloc+0x21/0x50
[<ffffffff81160720>] __kmalloc_track_caller+0x160/0x2f0
[<ffffffff81237b9b>] kvasprintf+0x5b/0x90
[<ffffffff8122c0c1>] kobject_set_name_vargs+0x21/0x60
[<ffffffff812e7f5c>] dev_set_name+0x3c/0x40
[<ffffffffa02bf918>] memstick_check+0xb8/0x340 [memstick]
[<ffffffff81069862>] process_one_work+0x1d2/0x670
[<ffffffff8106a88a>] worker_thread+0x11a/0x370
[<ffffffff81072ea6>] kthread+0xd6/0xe0
[<ffffffff81478bbc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff
larrylap:~ # modprobe -rv r592
rmmod /lib/modules/3.12.0-rc3-wl+/kernel/drivers/memstick/host/r592.ko
rmmod /lib/modules/3.12.0-rc3-wl+/kernel/drivers/memstick/core/memstick.ko
larrylap:~ # echo scan > /sys/kernel/debug/kmemleak
larrylap:~ # cat /sys/kernel/debug/kmemleak
unreferenced object 0xffff8800ae85c190 (size 16):
comm "kworker/u4:3", pid 685, jiffies 4294916336 (age 2947.576s)
hex dump (first 16 bytes):
6d 65 6d 73 74 69 63 6b 30 00 00 00 00 00 00 00 memstick0.......
backtrace:
[<ffffffff8146a0d1>] kmemleak_alloc+0x21/0x50
[<ffffffff81160720>] __kmalloc_track_caller+0x160/0x2f0
[<ffffffff81237b9b>] kvasprintf+0x5b/0x90
[<ffffffff8122c0c1>] kobject_set_name_vargs+0x21/0x60
[<ffffffff812e7f5c>] dev_set_name+0x3c/0x40
[<ffffffffa02bf918>] 0xffffffffa02bf918
[<ffffffff81069862>] process_one_work+0x1d2/0x670
[<ffffffff8106a88a>] worker_thread+0x11a/0x370
[<ffffffff81072ea6>] kthread+0xd6/0xe0
[<ffffffff81478bbc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff
larrylap:~ #

As I have not had this driver enabled in my configuration for very long, I'm not
sure when this behavior started, but it is definitely present in 3.11.0.

Thanks,

Larry

2013-10-01 14:05:34

by Catalin Marinas

Subject: Re: 3.12-rc3 memory leak in module memstick

On 30 September 2013 18:51, Larry Finger <[email protected]> wrote:
> On 3.12-rc3, I detected a memory leak in module memstick. The kmemleak
> results are listed below. The second output is after memstick was unloaded,
> which shows that it is not a false positive from kmemleak.
>
> larrylap:~ # echo scan > /sys/kernel/debug/kmemleak
> larrylap:~ # cat /sys/kernel/debug/kmemleak
> unreferenced object 0xffff8800ae85c190 (size 16):
> comm "kworker/u4:3", pid 685, jiffies 4294916336 (age 2831.760s)
> hex dump (first 16 bytes):
> 6d 65 6d 73 74 69 63 6b 30 00 00 00 00 00 00 00 memstick0.......
> backtrace:
> [<ffffffff8146a0d1>] kmemleak_alloc+0x21/0x50
> [<ffffffff81160720>] __kmalloc_track_caller+0x160/0x2f0
> [<ffffffff81237b9b>] kvasprintf+0x5b/0x90
> [<ffffffff8122c0c1>] kobject_set_name_vargs+0x21/0x60
> [<ffffffff812e7f5c>] dev_set_name+0x3c/0x40
> [<ffffffffa02bf918>] memstick_check+0xb8/0x340 [memstick]
> [<ffffffff81069862>] process_one_work+0x1d2/0x670
> [<ffffffff8106a88a>] worker_thread+0x11a/0x370
> [<ffffffff81072ea6>] kthread+0xd6/0xe0
> [<ffffffff81478bbc>] ret_from_fork+0x7c/0xb0

The leak is most likely coming from memstick_alloc_card() on the error
path. Commit 0252c3b4f0 (memstick: struct device - replace bus_id with
dev_name(), dev_set_name()) sets allocates the device name but 'card'
is freed on the error path without freeing the name.

--
Catalin

2013-10-01 17:03:38

by Larry Finger

Subject: Re: 3.12-rc3 memory leak in module memstick

On 10/01/2013 09:04 AM, Catalin Marinas wrote:
> On 30 September 2013 18:51, Larry Finger <[email protected]> wrote:
>> On 3.12-rc3, I detected a memory leak in module memstick. The kmemleak
>> results are listed below. The second output is after memstick was unloaded,
>> which shows that it is not a false positive from kmemleak.
>>
>> larrylap:~ # echo scan > /sys/kernel/debug/kmemleak
>> larrylap:~ # cat /sys/kernel/debug/kmemleak
>> unreferenced object 0xffff8800ae85c190 (size 16):
>> comm "kworker/u4:3", pid 685, jiffies 4294916336 (age 2831.760s)
>> hex dump (first 16 bytes):
>> 6d 65 6d 73 74 69 63 6b 30 00 00 00 00 00 00 00 memstick0.......
>> backtrace:
>> [<ffffffff8146a0d1>] kmemleak_alloc+0x21/0x50
>> [<ffffffff81160720>] __kmalloc_track_caller+0x160/0x2f0
>> [<ffffffff81237b9b>] kvasprintf+0x5b/0x90
>> [<ffffffff8122c0c1>] kobject_set_name_vargs+0x21/0x60
>> [<ffffffff812e7f5c>] dev_set_name+0x3c/0x40
>> [<ffffffffa02bf918>] memstick_check+0xb8/0x340 [memstick]
>> [<ffffffff81069862>] process_one_work+0x1d2/0x670
>> [<ffffffff8106a88a>] worker_thread+0x11a/0x370
>> [<ffffffff81072ea6>] kthread+0xd6/0xe0
>> [<ffffffff81478bbc>] ret_from_fork+0x7c/0xb0
>
> The leak is most likely coming from memstick_alloc_card() on the error
> path. Commit 0252c3b4f0 (memstick: struct device - replace bus_id with
> dev_name(), dev_set_name()) sets allocates the device name but 'card'
> is freed on the error path without freeing the name.

Thanks for the analysis. From that I am quite sure I can devise a patch.

Larry