Re: [PATCH 3/7] drm/mipi-dbi: Use framebuffer dma-buf helpers

[Thomas Zimmermann <tzimmermann@suse.de>]

[-- Attachment #1.1: Type: text/plain, Size: 3191 bytes --]

Hi

Am 20.07.21 um 16:03 schrieb Daniel Vetter:
> On Fri, Jul 16, 2021 at 04:07:57PM +0200, Thomas Zimmermann wrote:
>> Replace dma_buf_begin_cpu_access() with drm_gem_fb_begin_cpu_access();
>> same for _end_cpu_access(). Remove some boiler-plate code. No functional
>> changes.
>>
>> There's one left-over reference to the imported attachment that we
>> keep. GEM BOs with imported attachment are considered uncached and
>> enables special handling within the drm_fb_swab().
>>
>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
>> ---
>>   drivers/gpu/drm/drm_mipi_dbi.c | 20 +++++++-------------
>>   1 file changed, 7 insertions(+), 13 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/drm_mipi_dbi.c b/drivers/gpu/drm/drm_mipi_dbi.c
>> index 10b4e59384ae..71b646c4131f 100644
>> --- a/drivers/gpu/drm/drm_mipi_dbi.c
>> +++ b/drivers/gpu/drm/drm_mipi_dbi.c
>> @@ -7,7 +7,6 @@
>>   
>>   #include <linux/debugfs.h>
>>   #include <linux/delay.h>
>> -#include <linux/dma-buf.h>
>>   #include <linux/gpio/consumer.h>
>>   #include <linux/module.h>
>>   #include <linux/regulator/consumer.h>
>> @@ -202,21 +201,17 @@ int mipi_dbi_buf_copy(void *dst, struct drm_framebuffer *fb,
>>   {
>>   	struct drm_gem_object *gem = drm_gem_fb_get_obj(fb, 0);
>>   	struct drm_gem_cma_object *cma_obj = to_drm_gem_cma_obj(gem);
>> -	struct dma_buf_attachment *import_attach = gem->import_attach;
>>   	void *src = cma_obj->vaddr;
>> -	int ret = 0;
>> +	int ret;
>>   
>> -	if (import_attach) {
>> -		ret = dma_buf_begin_cpu_access(import_attach->dmabuf,
>> -					       DMA_FROM_DEVICE);
>> -		if (ret)
>> -			return ret;
>> -	}
>> +	ret = drm_gem_fb_begin_cpu_access(fb, DMA_FROM_DEVICE);
>> +	if (ret)
>> +		return ret;
>>   
>>   	switch (fb->format->format) {
>>   	case DRM_FORMAT_RGB565:
>>   		if (swap)
>> -			drm_fb_swab(dst, src, fb, clip, !import_attach);
>> +			drm_fb_swab(dst, src, fb, clip, !gem->import_attach);
> 
> I freaked out about this because for dma-buf WC vs WB is undefined, but
> it's purely a perf optimization. So it's fine. Plus we're not even
> bothering with the iomem vs normal memory distinction here.

Indeed. Because of the line above, I made a proto-patchset to store 
caching flags within struct dma_buf_map. But it's a lot of churn, so I 
kept the current heuristic for now.

> 
> Anyway, that aside, all looks good. On the series:
> 
> Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Thanks.

Best regards
Thomas

> 
>>   		else
>>   			drm_fb_memcpy(dst, src, fb, clip);
>>   		break;
>> @@ -229,9 +224,8 @@ int mipi_dbi_buf_copy(void *dst, struct drm_framebuffer *fb,
>>   		return -EINVAL;
>>   	}
>>   
>> -	if (import_attach)
>> -		ret = dma_buf_end_cpu_access(import_attach->dmabuf,
>> -					     DMA_FROM_DEVICE);
>> +	drm_gem_fb_end_cpu_access(fb, DMA_FROM_DEVICE);
>> +
>>   	return ret;
>>   }
>>   EXPORT_SYMBOL(mipi_dbi_buf_copy);
>> -- 
>> 2.32.0
>>
> 

-- 
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 840 bytes --]