[PATCH] drm/tiny: Refactor framebuffer's size calculation

Mehdi Ben Hadj Khelifa posted 1 patch 3 months, 2 weeks ago
drivers/gpu/drm/tiny/repaper.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
[PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Mehdi Ben Hadj Khelifa 3 months, 2 weeks ago
Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
instead of directly multiplying width and height. This aligns with DRM
helpers for determining per-line byte size and avoids manual assumptions
about bytes per pixel.

Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
---
 drivers/gpu/drm/tiny/repaper.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/repaper.c
index 4824f863fdba..aeff49bc6ba7 100644
--- a/drivers/gpu/drm/tiny/repaper.c
+++ b/drivers/gpu/drm/tiny/repaper.c
@@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct drm_framebuffer *fb, const struct iosys_map *
 	unsigned int dst_pitch = 0;
 	struct iosys_map dst;
 	struct drm_rect clip;
+	const struct drm_format_info *info = fb->format;
+	size_t pitch;
 	int idx, ret = 0;
 	u8 *buf = NULL;
 
@@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct drm_framebuffer *fb, const struct iosys_map *
 	DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
 		  epd->factored_stage_time);
 
-	buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
+	pitch = drm_format_info_min_pitch(info, 0, fb->width);
+
+	buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
 	if (!buf) {
 		ret = -ENOMEM;
 		goto out_exit;
-- 
2.51.1.dirty
Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Thomas Zimmermann 3 months, 2 weeks ago
Hi

Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
> instead of directly multiplying width and height. This aligns with DRM
> helpers for determining per-line byte size and avoids manual assumptions
> about bytes per pixel.
>
> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
> ---
>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>   1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/repaper.c
> index 4824f863fdba..aeff49bc6ba7 100644
> --- a/drivers/gpu/drm/tiny/repaper.c
> +++ b/drivers/gpu/drm/tiny/repaper.c
> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct drm_framebuffer *fb, const struct iosys_map *
>   	unsigned int dst_pitch = 0;
>   	struct iosys_map dst;
>   	struct drm_rect clip;
> +	const struct drm_format_info *info = fb->format;

This is the wrong format. You're allocating the output buffer here, but 
you're using the input format. IIUC the output format is DRM_FORMAT_R1. 
The input is _XRGB8888.

Best regards
Thomas

> +	size_t pitch;
>   	int idx, ret = 0;
>   	u8 *buf = NULL;
>   
> @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct drm_framebuffer *fb, const struct iosys_map *
>   	DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>   		  epd->factored_stage_time);
>   
> -	buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
> +	pitch = drm_format_info_min_pitch(info, 0, fb->width);
> +
> +	buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
>   	if (!buf) {
>   		ret = -ENOMEM;
>   		goto out_exit;

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstrasse 146, 90461 Nuernberg, Germany
GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
HRB 36809 (AG Nuernberg)
Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Mehdi Ben Hadj Khelifa 3 months, 2 weeks ago
On 10/21/25 7:51 AM, Thomas Zimmermann wrote:
> Hi
> 
> Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
>> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
>> instead of directly multiplying width and height. This aligns with DRM
>> helpers for determining per-line byte size and avoids manual assumptions
>> about bytes per pixel.
>>
>> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
>> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
>> ---
>>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/ 
>> repaper.c
>> index 4824f863fdba..aeff49bc6ba7 100644
>> --- a/drivers/gpu/drm/tiny/repaper.c
>> +++ b/drivers/gpu/drm/tiny/repaper.c
>> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct drm_framebuffer 
>> *fb, const struct iosys_map *
>>       unsigned int dst_pitch = 0;
>>       struct iosys_map dst;
>>       struct drm_rect clip;
>> +    const struct drm_format_info *info = fb->format;
> 
> This is the wrong format. You're allocating the output buffer here, but 
> you're using the input format. IIUC the output format is DRM_FORMAT_R1. 
> The input is _XRGB8888.
> 
Ah. Thanks for clarification.I thought since it had the same output 
format. I will send a v3 shortly.> Best regards
> Thomas
> 
>> +    size_t pitch;
>>       int idx, ret = 0;
>>       u8 *buf = NULL;
>> @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct drm_framebuffer 
>> *fb, const struct iosys_map *
>>       DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>>             epd->factored_stage_time);
>> -    buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
>> +    pitch = drm_format_info_min_pitch(info, 0, fb->width);
>> +
>> +    buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
>>       if (!buf) {
>>           ret = -ENOMEM;
>>           goto out_exit;
> 

Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Thomas Zimmermann 3 months, 2 weeks ago
Hi

Am 21.10.25 um 10:41 schrieb Mehdi Ben Hadj Khelifa:
> On 10/21/25 7:51 AM, Thomas Zimmermann wrote:
>> Hi
>>
>> Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
>>> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
>>> instead of directly multiplying width and height. This aligns with DRM
>>> helpers for determining per-line byte size and avoids manual 
>>> assumptions
>>> about bytes per pixel.
>>>
>>> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
>>> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
>>> ---
>>>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/ 
>>> repaper.c
>>> index 4824f863fdba..aeff49bc6ba7 100644
>>> --- a/drivers/gpu/drm/tiny/repaper.c
>>> +++ b/drivers/gpu/drm/tiny/repaper.c
>>> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct 
>>> drm_framebuffer *fb, const struct iosys_map *
>>>       unsigned int dst_pitch = 0;
>>>       struct iosys_map dst;
>>>       struct drm_rect clip;
>>> +    const struct drm_format_info *info = fb->format;
>>
>> This is the wrong format. You're allocating the output buffer here, 
>> but you're using the input format. IIUC the output format is 
>> DRM_FORMAT_R1. The input is _XRGB8888.
>>
> Ah. Thanks for clarification.I thought since it had the same output 
> format. I will send a v3 shortly.> Best regards

Maybe just don't do it. This is just churn with no clear goal.

Best regards
Thomas

>> Thomas
>>
>>> +    size_t pitch;
>>>       int idx, ret = 0;
>>>       u8 *buf = NULL;
>>> @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct 
>>> drm_framebuffer *fb, const struct iosys_map *
>>>       DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>>>             epd->factored_stage_time);
>>> -    buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
>>> +    pitch = drm_format_info_min_pitch(info, 0, fb->width);
>>> +
>>> +    buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
>>>       if (!buf) {
>>>           ret = -ENOMEM;
>>>           goto out_exit;
>>
>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstrasse 146, 90461 Nuernberg, Germany
GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
HRB 36809 (AG Nuernberg)


Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Mehdi Ben Hadj Khelifa 3 months, 2 weeks ago
On 10/21/25 9:32 AM, Thomas Zimmermann wrote:
> Hi
> 
> Am 21.10.25 um 10:41 schrieb Mehdi Ben Hadj Khelifa:
>> On 10/21/25 7:51 AM, Thomas Zimmermann wrote:
>>> Hi
>>>
>>> Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
>>>> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
>>>> instead of directly multiplying width and height. This aligns with DRM
>>>> helpers for determining per-line byte size and avoids manual 
>>>> assumptions
>>>> about bytes per pixel.
>>>>
>>>> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
>>>> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
>>>> ---
>>>>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>>>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/ 
>>>> repaper.c
>>>> index 4824f863fdba..aeff49bc6ba7 100644
>>>> --- a/drivers/gpu/drm/tiny/repaper.c
>>>> +++ b/drivers/gpu/drm/tiny/repaper.c
>>>> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct 
>>>> drm_framebuffer *fb, const struct iosys_map *
>>>>       unsigned int dst_pitch = 0;
>>>>       struct iosys_map dst;
>>>>       struct drm_rect clip;
>>>> +    const struct drm_format_info *info = fb->format;
>>>
>>> This is the wrong format. You're allocating the output buffer here, 
>>> but you're using the input format. IIUC the output format is 
>>> DRM_FORMAT_R1. The input is _XRGB8888.
>>>
>> Ah. Thanks for clarification.I thought since it had the same output 
>> format. I will send a v3 shortly.> Best regards
> 
> Maybe just don't do it. This is just churn with no clear goal.
> 
Okay,I will abort working on it.Though my goal was to remove the manual 
assumption that height is multiple of 8 and to align with other 
'correct' API used in other drm drivers as you suggested[1].

[1]:https://lore.kernel.org/all/cb0f0a36-0593-4d4c-8450-d086b9c99d87@suse.de/

Best Regards,
Mehdi Ben Hadj Khelifa> Best regards
> Thomas
> 
>>> Thomas
>>>
>>>> +    size_t pitch;
>>>>       int idx, ret = 0;
>>>>       u8 *buf = NULL;
>>>> @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct 
>>>> drm_framebuffer *fb, const struct iosys_map *
>>>>       DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>>>>             epd->factored_stage_time);
>>>> -    buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
>>>> +    pitch = drm_format_info_min_pitch(info, 0, fb->width);
>>>> +
>>>> +    buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
>>>>       if (!buf) {
>>>>           ret = -ENOMEM;
>>>>           goto out_exit;
>>>
>>
> 

Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Thomas Zimmermann 3 months, 2 weeks ago

Am 21.10.25 um 08:51 schrieb Thomas Zimmermann:
> Hi
>
> Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
>> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
>> instead of directly multiplying width and height. This aligns with DRM
>> helpers for determining per-line byte size and avoids manual assumptions
>> about bytes per pixel.
>>
>> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
>> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
>> ---
>>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/tiny/repaper.c 
>> b/drivers/gpu/drm/tiny/repaper.c
>> index 4824f863fdba..aeff49bc6ba7 100644
>> --- a/drivers/gpu/drm/tiny/repaper.c
>> +++ b/drivers/gpu/drm/tiny/repaper.c
>> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct 
>> drm_framebuffer *fb, const struct iosys_map *
>>       unsigned int dst_pitch = 0;
>>       struct iosys_map dst;
>>       struct drm_rect clip;
>> +    const struct drm_format_info *info = fb->format;
>
> This is the wrong format. You're allocating the output buffer here, 
> but you're using the input format. IIUC the output format is 
> DRM_FORMAT_R1. The input is _XRGB8888.
>
> Best regards
> Thomas
>
>> +    size_t pitch;
>>       int idx, ret = 0;
>>       u8 *buf = NULL;
>>   @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct 
>> drm_framebuffer *fb, const struct iosys_map *
>>       DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>>             epd->factored_stage_time);
>>   -    buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
>> +    pitch = drm_format_info_min_pitch(info, 0, fb->width);
>> +
>> +    buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);

And TBH I'm not sure if we should really convert this call. It's not an 
array in the sense of that API.

Good candidates look like 'kmalloc(count * sizeof(<some type or variable>))'

>>       if (!buf) {
>>           ret = -ENOMEM;
>>           goto out_exit;
>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstrasse 146, 90461 Nuernberg, Germany
GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
HRB 36809 (AG Nuernberg)


Re: [PATCH] drm/tiny: Refactor framebuffer's size calculation
Posted by Mehdi Ben Hadj Khelifa 3 months, 2 weeks ago
On 10/21/25 7:57 AM, Thomas Zimmermann wrote:
> 
> 
> Am 21.10.25 um 08:51 schrieb Thomas Zimmermann:
>> Hi
>>
>> Am 20.10.25 um 13:57 schrieb Mehdi Ben Hadj Khelifa:
>>> Use drm_format_info_min_pitch() to calculate the framebuffer line pitch
>>> instead of directly multiplying width and height. This aligns with DRM
>>> helpers for determining per-line byte size and avoids manual assumptions
>>> about bytes per pixel.
>>>
>>> Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
>>> Signed-off-by: Mehdi Ben Hadj Khelifa <mehdi.benhadjkhelifa@gmail.com>
>>> ---
>>>   drivers/gpu/drm/tiny/repaper.c | 6 +++++-
>>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/tiny/repaper.c b/drivers/gpu/drm/tiny/ 
>>> repaper.c
>>> index 4824f863fdba..aeff49bc6ba7 100644
>>> --- a/drivers/gpu/drm/tiny/repaper.c
>>> +++ b/drivers/gpu/drm/tiny/repaper.c
>>> @@ -517,6 +517,8 @@ static int repaper_fb_dirty(struct 
>>> drm_framebuffer *fb, const struct iosys_map *
>>>       unsigned int dst_pitch = 0;
>>>       struct iosys_map dst;
>>>       struct drm_rect clip;
>>> +    const struct drm_format_info *info = fb->format;
>>
>> This is the wrong format. You're allocating the output buffer here, 
>> but you're using the input format. IIUC the output format is 
>> DRM_FORMAT_R1. The input is _XRGB8888.
>>
>> Best regards
>> Thomas
>>
>>> +    size_t pitch;
>>>       int idx, ret = 0;
>>>       u8 *buf = NULL;
>>>   @@ -534,7 +536,9 @@ static int repaper_fb_dirty(struct 
>>> drm_framebuffer *fb, const struct iosys_map *
>>>       DRM_DEBUG("Flushing [FB:%d] st=%ums\n", fb->base.id,
>>>             epd->factored_stage_time);
>>>   -    buf = kmalloc(fb->width * fb->height / 8, GFP_KERNEL);
>>> +    pitch = drm_format_info_min_pitch(info, 0, fb->width);
>>> +
>>> +    buf = kmalloc_array(fb->height, pitch, GFP_KERNEL);
> 
> And TBH I'm not sure if we should really convert this call. It's not an 
> array in the sense of that API.
> 
> Good candidates look like 'kmalloc(count * sizeof(<some type or 
> variable>))'
> 
I will leave kmalloc in v3. Do you want me to send deep testing 
information on this patch?
BR,
Mehdi>>>       if (!buf) {
>>>           ret = -ENOMEM;
>>>           goto out_exit;
>>
>