[Intel-gfx] [PATCH v2] drm/probe_helper: sort out poll_running vs poll_enabled

[Dmitry Baryshkov]

On 20/01/2023 18:39, Laurent Pinchart wrote:
> Hi Dmitry,
> 
> Thank you for the patch.
> 
> On Wed, Jan 18, 2023 at 01:39:05AM +0200, Dmitry Baryshkov wrote:
>> There are two flags attemting to guard connector polling:
>> poll_enabled and poll_running. While poll_enabled semantics is clearly
>> defined and fully adhered (mark that drm_kms_helper_poll_init() was
>> called and not finalized by the _fini() call), the poll_running flag
>> doesn't have such clearliness.
>>
>> This flag is used only in drm_helper_probe_single_connector_modes() to
>> guard calling of drm_kms_helper_poll_enable, it doesn't guard the
>> drm_kms_helper_poll_fini(), etc. Change it to only be set if the polling
>> is actually running. Tie HPD enablement to this flag.
>>
>> This fixes the following warning reported after merging the HPD series:
>>
>> Hot plug detection already enabled
>> WARNING: CPU: 2 PID: 9 at drivers/gpu/drm/drm_bridge.c:1257 drm_bridge_hpd_enable+0x94/0x9c [drm]
>> Modules linked in: videobuf2_memops snd_soc_simple_card snd_soc_simple_card_utils fsl_imx8_ddr_perf videobuf2_common snd_soc_imx_spdif adv7511 etnaviv imx8m_ddrc imx_dcss mc cec nwl_dsi gov
>> CPU: 2 PID: 9 Comm: kworker/u8:0 Not tainted 6.2.0-rc2-15208-g25b283acd578 #6
>> Hardware name: NXP i.MX8MQ EVK (DT)
>> Workqueue: events_unbound deferred_probe_work_func
>> pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
>> pc : drm_bridge_hpd_enable+0x94/0x9c [drm]
>> lr : drm_bridge_hpd_enable+0x94/0x9c [drm]
>> sp : ffff800009ef3740
>> x29: ffff800009ef3740 x28: ffff000009331f00 x27: 0000000000001000
>> x26: 0000000000000020 x25: ffff800001148ed8 x24: ffff00000a8fe000
>> x23: 00000000fffffffd x22: ffff000005086348 x21: ffff800001133ee0
>> x20: ffff00000550d800 x19: ffff000005086288 x18: 0000000000000006
>> x17: 0000000000000000 x16: ffff8000096ef008 x15: 97ffff2891004260
>> x14: 2a1403e194000000 x13: 97ffff2891004260 x12: 2a1403e194000000
>> x11: 7100385f29400801 x10: 0000000000000aa0 x9 : ffff800008112744
>> x8 : ffff000000250b00 x7 : 0000000000000003 x6 : 0000000000000011
>> x5 : 0000000000000000 x4 : ffff0000bd986a48 x3 : 0000000000000001
>> x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff000000250000
>> Call trace:
>>   drm_bridge_hpd_enable+0x94/0x9c [drm]
>>   drm_bridge_connector_enable_hpd+0x2c/0x3c [drm_kms_helper]
>>   drm_kms_helper_poll_enable+0x94/0x10c [drm_kms_helper]
>>   drm_helper_probe_single_connector_modes+0x1a8/0x510 [drm_kms_helper]
>>   drm_client_modeset_probe+0x204/0x1190 [drm]
>>   __drm_fb_helper_initial_config_and_unlock+0x5c/0x4a4 [drm_kms_helper]
>>   drm_fb_helper_initial_config+0x54/0x6c [drm_kms_helper]
>>   drm_fbdev_client_hotplug+0xd0/0x140 [drm_kms_helper]
>>   drm_fbdev_generic_setup+0x90/0x154 [drm_kms_helper]
>>   dcss_kms_attach+0x1c8/0x254 [imx_dcss]
>>   dcss_drv_platform_probe+0x90/0xfc [imx_dcss]
>>   platform_probe+0x70/0xcc
>>   really_probe+0xc4/0x2e0
>>   __driver_probe_device+0x80/0xf0
>>   driver_probe_device+0xe0/0x164
>>   __device_attach_driver+0xc0/0x13c
>>   bus_for_each_drv+0x84/0xe0
>>   __device_attach+0xa4/0x1a0
>>   device_initial_probe+0x1c/0x30
>>   bus_probe_device+0xa4/0xb0
>>   deferred_probe_work_func+0x90/0xd0
>>   process_one_work+0x200/0x474
>>   worker_thread+0x74/0x43c
>>   kthread+0xfc/0x110
>>   ret_from_fork+0x10/0x20
>> ---[ end trace 0000000000000000 ]---
>>
>> Reported-by: Laurentiu Palcu <laurentiu.palcu at oss.nxp.com>
>> Fixes: c8268795c9a9 ("drm/probe-helper: enable and disable HPD on connectors")
>> Tested-by: Marek Szyprowski <m.szyprowski at samsung.com>
>> Tested-by: Chen-Yu Tsai <wenst at chromium.org>
>> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at linaro.org>
> 
> This fixes the issue on my Renesas boards.
> 
> Tested-by: Laurent Pinchart <laurent.pinchart+renesas at ideasonboard.com>
> 
> Please see below for a few comments.
> 

[skipped two small comments, will fix them in v3]


>> @@ -886,7 +897,12 @@ EXPORT_SYMBOL(drm_kms_helper_poll_init);
>>    */
>>   void drm_kms_helper_poll_fini(struct drm_device *dev)
>>   {
>> -	drm_kms_helper_poll_disable_fini(dev, true);
>> +	if (!dev->mode_config.poll_enabled)
>> +		return;
>> +
>> +	drm_kms_helper_poll_disable(dev);
>> +
>> +	dev->mode_config.poll_enabled = false;
> 
> The fact that poll_enabled guard init/fini and poll_running guards
> enable/disable is confusing :-S It would be nice if the poll_enable flag
> matched the enable/disable function names. One option could be
> 
> - drm_kms_helper_poll_enable() -> drm_kms_helper_poll_resume()
> - drm_kms_helper_poll_disable() -> drm_kms_helper_poll_suspend()
> - drm_kms_helper_poll_init() -> drm_kms_helper_poll_enable()
> - drm_kms_helper_poll_fini() -> drm_kms_helper_poll_disable()
> 
> This would align quite well with intended usage in drivers I think, the
> current enable/disable functions are documented as being used by drivers
> for suspend/resume, and the init()/fini() functions are documented as
> "init and enable" and "disable and cleanup".
> 
> This isn't a candidate for this patch of course, it can be done on top.

Yes, but I'd extremely dislike renaming a function to use the name of 
the function that existed previously. It can lead to very confusing issue.

What do you think about keeping poll_init()/_fini() as is, reaming 
dev->mode_config.poll_enabled to poll_initialized, and renaming 
poll_enable()/disable() to suspend()/resume() as you suggested?

> 
>>   }
>>   EXPORT_SYMBOL(drm_kms_helper_poll_fini);
>>   
> 

-- 
With best wishes
Dmitry