Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <b388f295-920a-b4fc-41ef-d090bdcd69e2@redhat.com>
Date:   Fri, 4 Feb 2022 20:31:36 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.5.0
Subject: Re: [PATCH v2 1/4] drm/format-helper: Add
 drm_fb_{xrgb8888,gray8}_to_mono_reversed()
Content-Language: en-US
To:     Thomas Zimmermann <tzimmermann@suse.de>,
        linux-kernel@vger.kernel.org
Cc:     Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
        Daniel Vetter <daniel.vetter@ffwll.ch>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        linux-fbdev@vger.kernel.org, Sam Ravnborg <sam@ravnborg.org>,
        dri-devel@lists.freedesktop.org,
        =?UTF-8?Q?Noralf_Tr=c3=b8nnes?= <noralf@tronnes.org>,
        Maxime Ripard <maxime@cerno.tech>,
        Daniel Vetter <daniel@ffwll.ch>,
        David Airlie <airlied@linux.ie>,
        Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
        Maxime Ripard <mripard@kernel.org>
References: <20220204134347.1187749-1-javierm@redhat.com>
 <20220204134347.1187749-2-javierm@redhat.com>
 <47100413-db63-1efa-45e9-028dfc430b7e@suse.de>
From:   Javier Martinez Canillas <javierm@redhat.com>
In-Reply-To: <47100413-db63-1efa-45e9-028dfc430b7e@suse.de>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hello Thomas,

Thanks a lot for your feedback.

On 2/4/22 16:52, Thomas Zimmermann wrote:

[snip]

>> +static void drm_fb_gray8_to_mono_reversed_line(u8 *dst, const u8 *src, size_t pixels)
>> +{
>> +	unsigned int xb, i;
>> +
>> +	for (xb = 0; xb < pixels / 8; xb++) {
> 
> In practice, all mode widths are multiples of 8 because VGA mandated it. 
> So it's ok-ish to assume this here. You should probably at least print a 
> warning somewhere if (pixels % 8 != 0)
>

Agreed.
 
[snip]

>> + * DRM doesn't have native monochrome or grayscale support.
>> + * Such drivers can announce the commonly supported XR24 format to userspace
>> + * and use drm_fb_xrgb8888_to_gray8() to convert to grayscale and then this
>> + * helper function to convert to the native format.
>> + */
>> +void drm_fb_gray8_to_mono_reversed(void *dst, unsigned int dst_pitch, const void *src,
>> +				   const struct drm_rect *clip)
> 
> There's a bug here. You want to pass in a drm_framebuffer as fourth 
> argument.
>
>> +{
>> +
>> +	size_t height = drm_rect_height(clip);
>> +	size_t width = drm_rect_width(clip);
>> +	unsigned int y;
>> +	const u8 *gray8 = src;
>> +	u8 *mono = dst;
>> +
>> +	if (!dst_pitch)
>> +		dst_pitch = width;
> 
> The dst_pitch is given in bytes. You have to device by 8. Here would be 
> a good place to warn if (width % 8 != 0).
>

Ok.
 
>> +
>> +	for (y = 0; y < height; y++) {
>> +		drm_fb_gray8_to_mono_reversed_line(mono, gray8, dst_pitch);
>> +		mono += (dst_pitch / 8);
> 
> The dst_pitch is already given in bytes.
>

Yes, I know but for reversed mono we want only 1/8 of the width since we
are converting from 8 bits per pixel greyscale to 1 bit per pixel mono.

Or am I misunderstanding what you meant ?

>> +		gray8 += dst_pitch;
> 
> 'gray8 += fb->pitches[0]' would be correct.
>

Ok.
 
[snip]

>> + */
>> +void drm_fb_xrgb8888_to_mono_reversed(void *dst, unsigned int dst_pitch, const void *src,
>> +				      const struct drm_framebuffer *fb,
>> +				      const struct drm_rect *clip)
>> +{
>> +	if (WARN_ON(fb->format->format != DRM_FORMAT_XRGB8888))
>> +		return;
>> +
>> +	if (!dst_pitch)
>> +		dst_pitch = drm_rect_width(clip);
>> +
>> +	drm_fb_xrgb8888_to_gray8(dst, dst_pitch, src, fb, clip);
>> +	drm_fb_gray8_to_mono_reversed(dst, dst_pitch, dst, fb, clip);
> 
> Converting from dst into dst can give incorrect results. At some point 
> we probably want to add restrict qualifiers to these pointers, to help 
> the compiler with optimizing.
> 
> A better approach here is to pull the per-line conversion from 
> drm_fb_xrgb8888_to_gray8() into a separate helper and implement a 
> line-by-line conversion here. something like this:
> 
>    drm_fb_xrgb8888_to_mono_reversed()
>    {
>      char *tmp = kmalloc(size of a single line of gray8)
> 
>      for (heigth) {
>         drm_fb_xrgb8888_to_gray8_line(tmp, ..., src, ...);
>         drm_fb_gray8_to_mono_reversed(dst, ..., tmp, ...);
> 
>         src += fb->pitches[0]
>         dst += dst_pitch;
>      }
> 
>      kfree(tmp);
>    }
>

I see. Yes, that sounds a much better approach. I'll change it in v3.
 
Best regards,
-- 
Javier Martinez Canillas
Linux Engineering
Red Hat