Re: [PATCH 0/8] drm: fb-helper/ssd130x: Add support for DRM_FORMAT_R1

[Javier Martinez Canillas]

Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> writes:

Hello Geert,

> Hi Javier,
>
> On Sun, Jul 16, 2023 at 3:30 PM Javier Martinez Canillas
> <javierm@xxxxxxxxxx> wrote:
>> Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> writes:
>> > The native display format of ssd1306 OLED displays is monochrome
>> > light-on-dark (R1).  This patch series adds support for the R1 buffer
>> > format to both the ssd130x DRM driver and the FB helpers, so monochrome
>> > applications (including fbdev emulation and the text console) can avoid
>> > the overhead of back-and-forth conversions between R1 and XR24.
>>
>> I've tested your series on a ssd1306 I2C OLED panel and fbcon did work for
>> me, but had some issues when trying to run your fbtest suite. Because the
>
> Thanks, due to the limited userspace environment on my RV32 test system,
> I didn't run fbtest myself.
>

You are welcome.

>> test005 gets killed with a SIGSEGV.
>>
>> $ ./fbtest -d
>> fb_init()
>> fb_open()
>> fb_get_var()
>> fb_get_fix()
>> fb_map()
>> fb_start = 0, fb_offset = 0, fb_len = 1000
>
>    [...]
>
>> Running test test005
>> Caught signal 11. Exiting
>
> Strange.
>
>> Maybe more tests are missing the minimum num_colors requirement? Also, the
>
> On monochrome, test005 should make the left half of the screen black,
> and the right half white.  It works on ARAnyM, and there don't seem
> to be off-by-one errors in the call to fill_rect().
> Can you please run this under gdb or valgrind?
>

Sure. I only spent my free time on these panels though so likely will do
during the week or more likely the weekend. I believe the bug is somewhere
in the test though and not in your kernel patches.

>> penguin in test004 is not displayed correctly. I was expecting that to be
>> working correctly since you mentioned to be using the Linux logo on boot.
>
> Linux has logos for displays using 2, 16, and 256 colors. Note that the
> default logos are 80x80, which is larger than your display, so no logo
> is drawn.
> Fbtest has only the full color logo, so it will look bad on a monochrome
> display.
>

I see. Should the test check for minimum num_colors and skip that test then?

>> Another question, do you know if is possible to change the default format?
>
> AFAIK DRM does not support that.
>
>> I believe that fbset won't work because the DRM fbdev emulation layer does
>> not implement mode settings but I thought that changing the mode using the
>
> Correct.
>

[...]

>> Maybe I'm doing something wrong or misunderstading about how should work?
>
> Do you need the "-d" option (drop master after mode set) of modetest?
> Still, that would only impact DRM.  The fbdev emulation on top of DRM has
> already been initialized before, and is never reinitialized.
>

Ah, that explains and makes sense now. I tested other user-space fbdev
programs that only support XRGB8888 and those were working correctly so I
guess that user-space that supports R1 be defaulted to that and not able
to change is an acceptable behaviour.

When you post v2, feel free to add:

Tested-by: Javier Martinez Canillas <javierm@xxxxxxxxxx>

-- 
Best regards,

Javier Martinez Canillas
Core Platforms
Red Hat