Re: [PATCH 1/3] drm/connector: hdmi: Fix memory leak in drm_display_mode_from_cea_vic()

[Maxime Ripard]

On Mon, Oct 14, 2024 at 03:16:30PM GMT, Jinjie Ruan wrote:
> modprobe drm_connector_test and then rmmod drm_connector_test,
> the following memory leak occurs.
> 
> The `mode` allocated in drm_mode_duplicate() called by
> drm_display_mode_from_cea_vic() is not freed, which cause the memory leak:
> 
> 	unreferenced object 0xffffff80cb0ee400 (size 128):
> 	  comm "kunit_try_catch", pid 1948, jiffies 4294950339
> 	  hex dump (first 32 bytes):
> 	    14 44 02 00 80 07 d8 07 04 08 98 08 00 00 38 04  .D............8.
> 	    3c 04 41 04 65 04 00 00 05 00 00 00 00 00 00 00  <.A.e...........
> 	  backtrace (crc 90e9585c):
> 	    [<00000000ec42e3d7>] kmemleak_alloc+0x34/0x40
> 	    [<00000000d0ef055a>] __kmalloc_cache_noprof+0x26c/0x2f4
> 	    [<00000000c2062161>] drm_mode_duplicate+0x44/0x19c
> 	    [<00000000f96c74aa>] drm_display_mode_from_cea_vic+0x88/0x98
> 	    [<00000000d8f2c8b4>] 0xffffffdc982a4868
> 	    [<000000005d164dbc>] kunit_try_run_case+0x13c/0x3ac
> 	    [<000000006fb23398>] kunit_generic_run_threadfn_adapter+0x80/0xec
> 	    [<000000006ea56ca0>] kthread+0x2e8/0x374
> 	    [<000000000676063f>] ret_from_fork+0x10/0x20
> 	......
> 
> Free `mode` by calling drm_mode_destroy() to fix it.
> 
> Cc: stable@xxxxxxxxxxxxxxx
> Fixes: abb6f74973e2 ("drm/tests: Add HDMI TDMS character rate tests")
> Signed-off-by: Jinjie Ruan <ruanjinjie@xxxxxxxxxx>
> ---
>  drivers/gpu/drm/tests/drm_connector_test.c | 24 ++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
> 
> diff --git a/drivers/gpu/drm/tests/drm_connector_test.c b/drivers/gpu/drm/tests/drm_connector_test.c
> index 15e36a8db685..9c94d26b3486 100644
> --- a/drivers/gpu/drm/tests/drm_connector_test.c
> +++ b/drivers/gpu/drm/tests/drm_connector_test.c
> @@ -1004,6 +1004,8 @@ static void drm_test_drm_hdmi_compute_mode_clock_rgb(struct kunit *test)
>  	rate = drm_hdmi_compute_mode_clock(mode, 8, HDMI_COLORSPACE_RGB);
>  	KUNIT_ASSERT_GT(test, rate, 0);
>  	KUNIT_EXPECT_EQ(test, mode->clock * 1000ULL, rate);
> +
> +	drm_mode_destroy(drm, mode);
>  }

If KUNIT_ASSERT_GT triggers, then we would end up leaking the mode as well.

I think we should create a kunit_drm_display_mode_from_cea_vic()
function that registers a kunit action to free the mode when the test is
done.

Maxime

Attachment: signature.asc
Description: PGP signature