[Intel-gfx] LOOKS GOOD: Possible regression in drm/i915 driver: memleak

[Mirsad Todorovac]

On 12/20/22 16:52, Tvrtko Ursulin wrote:

> On 20/12/2022 15:22, srinivas pandruvada wrote:
>> +Added DRM mailing list and maintainers
>>
>> On Tue, 2022-12-20 at 15:33 +0100, Mirsad Todorovac wrote:
>>> Hi all,
>>>
>>> I have been unsuccessful to find any particular Intel i915 maintainer
>>> emails, so my best bet is to post here, as you will must assuredly
>>> already know them.
> 
> For future reference you can use ${kernel_dir}/scripts/get_maintainer.pl 
> -f ...
> 
>>> The problem is a kernel memory leak that is repeatedly occurring
>>> triggered during the execution of Chrome browser under the latest
>>> 6.1.0+
>>> kernel of this morning and Almalinux 8.6 on a Lenovo desktop box
>>> with Intel(R) Core(TM) i5-8400 CPU @ 2.80GHz CPU.
>>>
>>> The build is with KMEMLEAK, KASAN and MGLRU turned on during the
>>> build,
>>> on a vanilla mainline kernel from Mr. Torvalds' tree.
>>>
>>> The leaks look like this one:
>>>
>>> unreferenced object 0xffff888131754880 (size 64):
>>>     comm "chrome", pid 13058, jiffies 4298568878 (age 3708.084s)
>>>     hex dump (first 32 bytes):
>>>       01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>>> ................
>>>       00 00 00 00 00 00 00 00 00 80 1e 3e 83 88 ff ff
>>> ...........>....
>>>     backtrace:
>>>       [<ffffffff9e9b5542>] slab_post_alloc_hook+0xb2/0x340
>>>       [<ffffffff9e9bbf5f>] __kmem_cache_alloc_node+0x1bf/0x2c0
>>>       [<ffffffff9e8f767a>] kmalloc_trace+0x2a/0xb0
>>>       [<ffffffffc08dfde5>] drm_vma_node_allow+0x45/0x150 [drm]
>>>       [<ffffffffc0b33315>] __assign_mmap_offset_handle+0x615/0x820
>>> [i915]
>>>       [<ffffffffc0b34057>] i915_gem_mmap_offset_ioctl+0x77/0x110
>>> [i915]
>>>       [<ffffffffc08bc5e1>] drm_ioctl_kernel+0x181/0x280 [drm]
>>>       [<ffffffffc08bc9cd>] drm_ioctl+0x2dd/0x6a0 [drm]
>>>       [<ffffffff9ea54744>] __x64_sys_ioctl+0xc4/0x100
>>>       [<ffffffff9fbc0178>] do_syscall_64+0x58/0x80
>>>       [<ffffffff9fc000aa>] entry_SYSCALL_64_after_hwframe+0x72/0xdc
>>>
>>> The complete list of leaks in attachment, but they seem similar or
>>> the same.
>>>
>>> Please find attached lshw and kernel build config file.
>>>
>>> I will probably check the same parms on my laptop at home, which is
>>> also
>>> Lenovo, but a different hw config and Ubuntu 22.10.
> 
> Could you try the below patch?
> 
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c 
> b/drivers/gpu/drm/i915/gem/i915_gem_mman.c
> index c3ea243d414d..0b07534c203a 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c
> @@ -679,9 +679,10 @@ mmap_offset_attach(struct drm_i915_gem_object *obj,
>   insert:
>          mmo = insert_mmo(obj, mmo);
>          GEM_BUG_ON(lookup_mmo(obj, mmap_type) != mmo);
> -out:
> +
>          if (file)
>                  drm_vma_node_allow(&mmo->vma_node, file);
> +out:
>          return mmo;
> 
>   err:
> 
> Maybe it is not the best fix but curious to know if it will make the 
> leak go away.

Hi,

After 27 minutes uptime with the patched kernel it looks promising.
It is much longer than it took for the buggy kernel to leak slabs.

Here is the output:

[root at pc-mtodorov marvin]# echo scan > /sys/kernel/debug/kmemleak
[root at pc-mtodorov marvin]# cat !$
cat /sys/kernel/debug/kmemleak
unreferenced object 0xffff888105028d80 (size 16):
   comm "kworker/u12:5", pid 359, jiffies 4294902898 (age 1620.144s)
   hex dump (first 16 bytes):
     6d 65 6d 73 74 69 63 6b 30 00 00 00 00 00 00 00  memstick0.......
   backtrace:
     [<ffffffffb6bb5542>] slab_post_alloc_hook+0xb2/0x340
     [<ffffffffb6bbbf5f>] __kmem_cache_alloc_node+0x1bf/0x2c0
     [<ffffffffb6af8175>] __kmalloc_node_track_caller+0x55/0x160
     [<ffffffffb6ae34a6>] kstrdup+0x36/0x60
     [<ffffffffb6ae3508>] kstrdup_const+0x28/0x30
     [<ffffffffb70d0757>] kvasprintf_const+0x97/0xd0
     [<ffffffffb7c9cdf4>] kobject_set_name_vargs+0x34/0xc0
     [<ffffffffb750289b>] dev_set_name+0x9b/0xd0
     [<ffffffffc12d9201>] memstick_check+0x181/0x639 [memstick]
     [<ffffffffb676e1d6>] process_one_work+0x4e6/0x7e0
     [<ffffffffb676e556>] worker_thread+0x76/0x770
     [<ffffffffb677b468>] kthread+0x168/0x1a0
     [<ffffffffb6604c99>] ret_from_fork+0x29/0x50
[root at pc-mtodorov marvin]# w
  20:27:35 up 27 min,  2 users,  load average: 0.83, 1.15, 1.19
USER     TTY      FROM             LOGIN@   IDLE   JCPU   PCPU WHAT
marvin   tty2     tty2             20:01   27:10  10:12   2.09s 
/opt/google/chrome/chrome --type=utility --utility-sub-type=audio.m
marvin   pts/1    -                20:01    0.00s  2:00   0.38s sudo bash
[root at pc-mtodorov marvin]# uname -rms
Linux 6.1.0-b6bb9676f216-mglru-kmemlk-kasan+ x86_64
[root at pc-mtodorov marvin]#

2. On the Ubuntu 22.10 with Debian build I did not reproduce the error 
thus far.

This looks to me like fixed, but if it doesn't leak anything until 
Thursday morning when I will see this desktop box next time, then we'll 
know with more certainty.

Hope this helps. (My $0.02 .)

Kudos for the quick fix :)

Kind regards,
Mirsad

-- 
Mirsad Goran Todorovac
Sistem inženjer
Grafički fakultet | Akademija likovnih umjetnosti
Sveučilište u Zagrebu
-- 
System engineer
Faculty of Graphic Arts | Academy of Fine Arts
University of Zagreb, Republic of Croatia