[PATCH 1/3] drm/meson: reorder driver deinit sequence to fix use-after-free bug

[Neil Armstrong]

On 19/09/2022 03:09, Adrián Larumbe wrote:
> Unloading the driver triggers the following KASAN warning:
>
> [  +0.006275] =============================================================
> [  +0.000029] BUG: KASAN: use-after-free in __list_del_entry_valid+0xe0/0x1a0
> [  +0.000026] Read of size 8 at addr ffff000020c395e0 by task rmmod/2695
>
> [  +0.000019] CPU: 5 PID: 2695 Comm: rmmod Tainted: G         C O      5.19.0-rc6-lrmbkasan+ #1
> [  +0.000013] Hardware name: Hardkernel ODROID-N2Plus (DT)
> [  +0.000008] Call trace:
> [  +0.000007]  dump_backtrace+0x1ec/0x280
> [  +0.000013]  show_stack+0x24/0x80
> [  +0.000008]  dump_stack_lvl+0x98/0xd4
> [  +0.000011]  print_address_description.constprop.0+0x80/0x520
> [  +0.000011]  print_report+0x128/0x260
> [  +0.000007]  kasan_report+0xb8/0xfc
> [  +0.000008]  __asan_report_load8_noabort+0x3c/0x50
> [  +0.000010]  __list_del_entry_valid+0xe0/0x1a0
> [  +0.000009]  drm_atomic_private_obj_fini+0x30/0x200 [drm]
> [  +0.000172]  drm_bridge_detach+0x94/0x260 [drm]
> [  +0.000145]  drm_encoder_cleanup+0xa4/0x290 [drm]
> [  +0.000144]  drm_mode_config_cleanup+0x118/0x740 [drm]
> [  +0.000143]  drm_mode_config_init_release+0x1c/0x2c [drm]
> [  +0.000144]  drm_managed_release+0x170/0x414 [drm]
> [  +0.000142]  drm_dev_put.part.0+0xc0/0x124 [drm]
> [  +0.000143]  drm_dev_put+0x20/0x30 [drm]
> [  +0.000142]  meson_drv_unbind+0x1d8/0x2ac [meson_drm]
> [  +0.000028]  take_down_aggregate_device+0xb0/0x160
> [  +0.000016]  component_del+0x18c/0x360
> [  +0.000009]  meson_dw_hdmi_remove+0x28/0x40 [meson_dw_hdmi]
> [  +0.000015]  platform_remove+0x64/0xb0
> [  +0.000009]  device_remove+0xb8/0x154
> [  +0.000009]  device_release_driver_internal+0x398/0x5b0
> [  +0.000009]  driver_detach+0xac/0x1b0
> [  +0.000009]  bus_remove_driver+0x158/0x29c
> [  +0.000009]  driver_unregister+0x70/0xb0
> [  +0.000008]  platform_driver_unregister+0x20/0x2c
> [  +0.000008]  meson_dw_hdmi_platform_driver_exit+0x1c/0x30 [meson_dw_hdmi]
> [  +0.000012]  __do_sys_delete_module+0x288/0x400
> [  +0.000011]  __arm64_sys_delete_module+0x5c/0x80
> [  +0.000009]  invoke_syscall+0x74/0x260
> [  +0.000009]  el0_svc_common.constprop.0+0xcc/0x260
> [  +0.000009]  do_el0_svc+0x50/0x70
> [  +0.000007]  el0_svc+0x68/0x1a0
> [  +0.000012]  el0t_64_sync_handler+0x11c/0x150
> [  +0.000008]  el0t_64_sync+0x18c/0x190
>
> [  +0.000018] Allocated by task 0:
> [  +0.000007] (stack is not available)
>
> [  +0.000011] Freed by task 2695:
> [  +0.000008]  kasan_save_stack+0x2c/0x5c
> [  +0.000011]  kasan_set_track+0x2c/0x40
> [  +0.000008]  kasan_set_free_info+0x28/0x50
> [  +0.000009]  ____kasan_slab_free+0x128/0x1d4
> [  +0.000008]  __kasan_slab_free+0x18/0x24
> [  +0.000007]  slab_free_freelist_hook+0x108/0x230
> [  +0.000011]  kfree+0x110/0x35c
> [  +0.000008]  release_nodes+0xf0/0x16c
> [  +0.000009]  devres_release_group+0x180/0x270
> [  +0.000008]  component_unbind+0x128/0x1e0
> [  +0.000010]  component_unbind_all+0x1b8/0x264
> [  +0.000009]  meson_drv_unbind+0x1a0/0x2ac [meson_drm]
> [  +0.000025]  take_down_aggregate_device+0xb0/0x160
> [  +0.000009]  component_del+0x18c/0x360
> [  +0.000009]  meson_dw_hdmi_remove+0x28/0x40 [meson_dw_hdmi]
> [  +0.000012]  platform_remove+0x64/0xb0
> [  +0.000008]  device_remove+0xb8/0x154
> [  +0.000009]  device_release_driver_internal+0x398/0x5b0
> [  +0.000009]  driver_detach+0xac/0x1b0
> [  +0.000009]  bus_remove_driver+0x158/0x29c
> [  +0.000008]  driver_unregister+0x70/0xb0
> [  +0.000008]  platform_driver_unregister+0x20/0x2c
> [  +0.000008]  meson_dw_hdmi_platform_driver_exit+0x1c/0x30 [meson_dw_hdmi]
> [  +0.000011]  __do_sys_delete_module+0x288/0x400
> [  +0.000010]  __arm64_sys_delete_module+0x5c/0x80
> [  +0.000008]  invoke_syscall+0x74/0x260
> [  +0.000008]  el0_svc_common.constprop.0+0xcc/0x260
> [  +0.000008]  do_el0_svc+0x50/0x70
> [  +0.000007]  el0_svc+0x68/0x1a0
> [  +0.000009]  el0t_64_sync_handler+0x11c/0x150
> [  +0.000009]  el0t_64_sync+0x18c/0x190
>
> [  +0.000014] The buggy address belongs to the object at ffff000020c39000
>                 which belongs to the cache kmalloc-4k of size 4096
> [  +0.000008] The buggy address is located 1504 bytes inside of
>                 4096-byte region [ffff000020c39000, ffff000020c3a000)
>
> [  +0.000016] The buggy address belongs to the physical page:
> [  +0.000009] page:fffffc0000830e00 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x20c38
> [  +0.000013] head:fffffc0000830e00 order:3 compound_mapcount:0 compound_pincount:0
> [  +0.000008] flags: 0xffff00000010200(slab|head|node=0|zone=0|lastcpupid=0xffff)
> [  +0.000019] raw: 0ffff00000010200 fffffc0000fd4808 fffffc0000126208 ffff000000002e80
> [  +0.000009] raw: 0000000000000000 0000000000020002 00000001ffffffff 0000000000000000
> [  +0.000008] page dumped because: kasan: bad access detected
>
> [  +0.000011] Memory state around the buggy address:
> [  +0.000008]  ffff000020c39480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> [  +0.000007]  ffff000020c39500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> [  +0.000007] >ffff000020c39580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> [  +0.000007]                                                        ^
> [  +0.000007]  ffff000020c39600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> [  +0.000007]  ffff000020c39680: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> [  +0.000006] ==================================================================
>
> The reason this is happening is unloading meson-dw-hdmi will cause the
> component API to take down the aggregate device, which in turn will cause
> all devres-managed memory to be freed, including the struct dw_hdmi
> allocated in dw_hdmi_probe. This struct embeds a struct drm_bridge that is
> added at the end of the function, and which is later on picked up in
> meson_encoder_hdmi_init.
>
> However, when attaching the bridge to the encoder created in
> meson_encoder_hdmi_init, it's linked to the encoder's bridge chain, from
> where it never leaves, even after devres_release_group is called when the
> driver's components are unbound and the embedding structure freed.
>
> Then, when calling drm_dev_put in the aggregate driver's unbind function,
> drm_bridge_detach is called for every single bridge linked to the encoder,
> including the one whose memory had already been deallocated.
>
> Fix by calling component_unbind_all after drm_dev_put.
>
> Signed-off-by: Adrián Larumbe <adrian.larumbe at collabora.com>
> ---
>   drivers/gpu/drm/meson/meson_drv.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/meson/meson_drv.c b/drivers/gpu/drm/meson/meson_drv.c
> index ef386d7b9450..8da454a17b77 100644
> --- a/drivers/gpu/drm/meson/meson_drv.c
> +++ b/drivers/gpu/drm/meson/meson_drv.c
> @@ -385,9 +385,9 @@ static void meson_drv_unbind(struct device *dev)
>   	drm_dev_unregister(drm);
>   	drm_kms_helper_poll_fini(drm);
>   	drm_atomic_helper_shutdown(drm);
> -	component_unbind_all(dev, drm);
>   	free_irq(priv->vsync_irq, drm);
>   	drm_dev_put(drm);
> +	component_unbind_all(dev, drm);
>   
>   	if (priv->afbcd.ops)
>   		priv->afbcd.ops->exit(priv);

Reviewed-by: Neil Armstrong <neil.armstrong at linaro.org>