[igt-dev] [PATCH i-g-t] tests/kms_plane_scaling: Increase buffer size if driver doesn't support larger scale factors

[Jessica Zhang]

Hi Petri,

On 12/8/2021 6:06 AM, Petri Latvala wrote:
> On Fri, Dec 03, 2021 at 07:49:19PM -0800, Jessica Zhang wrote:
>> Catch edge cases where driver doesn't support larger scale factors or
>> pipe doesn't support scaling.
>>
>> Currently, a 20x20 framebuffer is passed in to be upscaled. However,
>> this will cause issues with other drivers as they may not support larger
>> scale factors or may not support scaling at all for certain planes.
>>
>> This avoids failures due to invalid scale factor by trying
>> the original 20x20 framebuffer commit, then trying to commit larger
>> framebuffers up to and including unity scale.
>>
>> Tested-on: Qualcomm RB5 (sdm845)
>>
>> Signed-off-by: Jessica Zhang <quic_jesszhan at quicinc.com>
>> ---
>>   tests/kms_plane_scaling.c | 52 ++++++++++++++++++++++++++++++---------
>>   1 file changed, 41 insertions(+), 11 deletions(-)
>>
>> diff --git a/tests/kms_plane_scaling.c b/tests/kms_plane_scaling.c
>> index 85db11ee..1f0143fc 100644
>> --- a/tests/kms_plane_scaling.c
>> +++ b/tests/kms_plane_scaling.c
>> @@ -1,5 +1,6 @@
>>   /*
>>    * Copyright © 2013,2014 Intel Corporation
>> + * Copyright (c) 2021 Qualcomm Innovation Center, Inc. All rights reserved.
>>    *
>>    * Permission is hereby granted, free of charge, to any person obtaining a
>>    * copy of this software and associated documentation files (the "Software"),
>> @@ -118,6 +119,30 @@ static void prepare_crtc(data_t *data, igt_output_t *output, enum pipe pipe,
>>   	igt_display_commit2(display, COMMIT_ATOMIC);
>>   }
>>   
>> +static int try_commit_with_fb_size(int width, int height, igt_rotation_t rot, igt_display_t *display, data_t *d, igt_plane_t *plane, drmModeModeInfo *mode)
>> +{
>> +	int ret;
>> +
>> +	cleanup_crtc(d);
>> +
>> +	/* Check min to full resolution upscaling */
>> +	igt_fb_set_position(&d->fb[0], plane, 0, 0);
>> +	igt_fb_set_size(&d->fb[0], plane, width, height);
>> +	igt_plane_set_position(plane, 0, 0);
>> +	igt_plane_set_size(plane, mode->hdisplay, mode->vdisplay);
>> +	igt_plane_set_rotation(plane, rot);
>> +
>> +	ret = igt_display_try_commit2(display, COMMIT_ATOMIC);
>> +
>> +	if (!ret) {
>> +		igt_display_commit2(display, COMMIT_ATOMIC);
>> +		igt_plane_set_fb(plane, NULL);
>> +		igt_plane_set_position(plane, 0, 0);
> 
> I don't understand what's done here. If doing a commit fails, do the
> commit again?
> 
In this method, we'll test if a commit will go through with a specific 
framebuffer (without actually doing the commit) using 
`igt_display_try_commit2()`. If this succeeds, then we'll actually go 
through with the commit. Otherwise, we'll propagate the error code.

This will allow us to catch errors (specifically -ERANGE) and try again 
with a larger framebuffer in case the scale factor is too large or the 
plane doesn't support scaling.

Thanks,
Jessica Zhang
>