[PATCH v2 0/3] drm/simpledrm: Apple M1 / DT platform support fixes

[PATCH v2 0/3] drm/simpledrm: Apple M1 / DT platform support fixes

[Hector Martin]

Hi DRM folks,

This short series makes simpledrm work on Apple M1 (including Pro/Max)
platforms the way simplefb already does, by adding XRGB2101010 support
and making it bind to framebuffers in /chosen the same way simplefb
does.

This avoids breaking the bootloader-provided framebuffer console when
simpledrm is selected to replace simplefb, as these FBs always seem to
be 10-bit (at least when a real screen is attached).

Changes since v1:
- Moved the OF platform device setup code from simplefb into common
  code, instead of duplicating it in simpledrm
- Rebased on drm-tip

Hector Martin (3):
  of: Move simple-framebuffer device handling from simplefb to of
  drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()
  drm/simpledrm: Add XRGB2101010 format

 drivers/gpu/drm/drm_format_helper.c | 62 +++++++++++++++++++++++++++++
 drivers/gpu/drm/tiny/simpledrm.c    |  2 +-
 drivers/of/platform.c               |  5 +++
 drivers/video/fbdev/simplefb.c      | 21 +---------
 include/drm/drm_format_helper.h     |  3 ++
 5 files changed, 72 insertions(+), 21 deletions(-)

-- 
2.33.0

[PATCH v2 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Hector Martin]

This code is required for both simplefb and simpledrm, so let's move it
into the OF core instead of having it as an ad-hoc initcall in the
drivers.

Signed-off-by: Hector Martin <marcan@marcan.st>
---
 drivers/of/platform.c          |  5 +++++
 drivers/video/fbdev/simplefb.c | 21 +--------------------
 2 files changed, 6 insertions(+), 20 deletions(-)

diff --git a/drivers/of/platform.c b/drivers/of/platform.c
index b3faf89744aa..e097f40b03c0 100644
--- a/drivers/of/platform.c
+++ b/drivers/of/platform.c
@@ -540,6 +540,11 @@ static int __init of_platform_default_populate_init(void)
 		of_node_put(node);
 	}
 
+	for_each_child_of_node(of_chosen, node) {
+		if (of_device_is_compatible(node, "simple-framebuffer"))
+			of_platform_device_create(node, NULL, NULL);
+	}
+
 	/* Populate everything else. */
 	of_platform_default_populate(NULL, NULL, NULL);
 
diff --git a/drivers/video/fbdev/simplefb.c b/drivers/video/fbdev/simplefb.c
index b63074fd892e..57541887188b 100644
--- a/drivers/video/fbdev/simplefb.c
+++ b/drivers/video/fbdev/simplefb.c
@@ -541,26 +541,7 @@ static struct platform_driver simplefb_driver = {
 	.remove = simplefb_remove,
 };
 
-static int __init simplefb_init(void)
-{
-	int ret;
-	struct device_node *np;
-
-	ret = platform_driver_register(&simplefb_driver);
-	if (ret)
-		return ret;
-
-	if (IS_ENABLED(CONFIG_OF_ADDRESS) && of_chosen) {
-		for_each_child_of_node(of_chosen, np) {
-			if (of_device_is_compatible(np, "simple-framebuffer"))
-				of_platform_device_create(np, NULL, NULL);
-		}
-	}
-
-	return 0;
-}
-
-fs_initcall(simplefb_init);
+module_platform_driver(simplefb_driver);
 
 MODULE_AUTHOR("Stephen Warren <swarren@wwwdotorg.org>");
 MODULE_DESCRIPTION("Simple framebuffer driver");
-- 
2.33.0

[PATCH v2 2/3] drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()

[Hector Martin]

Add XRGB8888 emulation support for devices that can only do XRGB2101010.

This is chiefly useful for simpledrm on Apple devices where the
bootloader-provided framebuffer is 10-bit.

Signed-off-by: Hector Martin <marcan@marcan.st>
---
 drivers/gpu/drm/drm_format_helper.c | 62 +++++++++++++++++++++++++++++
 include/drm/drm_format_helper.h     |  3 ++
 2 files changed, 65 insertions(+)

diff --git a/drivers/gpu/drm/drm_format_helper.c b/drivers/gpu/drm/drm_format_helper.c
index dbe3e830096e..edd611d3ab6a 100644
--- a/drivers/gpu/drm/drm_format_helper.c
+++ b/drivers/gpu/drm/drm_format_helper.c
@@ -409,6 +409,59 @@ void drm_fb_xrgb8888_to_rgb888_toio(void __iomem *dst, unsigned int dst_pitch,
 }
 EXPORT_SYMBOL(drm_fb_xrgb8888_to_rgb888_toio);
 
+static void drm_fb_xrgb8888_to_xrgb2101010_line(u32 *dbuf, const u32 *sbuf,
+						unsigned int pixels)
+{
+	unsigned int x;
+
+	for (x = 0; x < pixels; x++) {
+		*dbuf++ = ((sbuf[x] & 0x000000FF) << 2) |
+			  ((sbuf[x] & 0x0000FF00) << 4) |
+			  ((sbuf[x] & 0x00FF0000) << 6);
+	}
+}
+
+/**
+ * drm_fb_xrgb8888_to_xrgb2101010_toio - Convert XRGB8888 to XRGB2101010 clip
+ * buffer
+ * @dst: XRGB2101010 destination buffer (iomem)
+ * @dst_pitch: Number of bytes between two consecutive scanlines within dst
+ * @vaddr: XRGB8888 source buffer
+ * @fb: DRM framebuffer
+ * @clip: Clip rectangle area to copy
+ *
+ * Drivers can use this function for XRGB2101010 devices that don't natively
+ * support XRGB8888.
+ */
+void drm_fb_xrgb8888_to_xrgb2101010_toio(void __iomem *dst,
+					 unsigned int dst_pitch, const void *vaddr,
+					 const struct drm_framebuffer *fb,
+					 const struct drm_rect *clip)
+{
+	size_t linepixels = clip->x2 - clip->x1;
+	size_t dst_len = linepixels * sizeof(u32);
+	unsigned y, lines = clip->y2 - clip->y1;
+	void *dbuf;
+
+	if (!dst_pitch)
+		dst_pitch = dst_len;
+
+	dbuf = kmalloc(dst_len, GFP_KERNEL);
+	if (!dbuf)
+		return;
+
+	vaddr += clip_offset(clip, fb->pitches[0], sizeof(u32));
+	for (y = 0; y < lines; y++) {
+		drm_fb_xrgb8888_to_xrgb2101010_line(dbuf, vaddr, linepixels);
+		memcpy_toio(dst, dbuf, dst_len);
+		vaddr += fb->pitches[0];
+		dst += dst_pitch;
+	}
+
+	kfree(dbuf);
+}
+EXPORT_SYMBOL(drm_fb_xrgb8888_to_xrgb2101010_toio);
+
 /**
  * drm_fb_xrgb8888_to_gray8 - Convert XRGB8888 to grayscale
  * @dst: 8-bit grayscale destination buffer
@@ -500,6 +553,10 @@ int drm_fb_blit_toio(void __iomem *dst, unsigned int dst_pitch, uint32_t dst_for
 		fb_format = DRM_FORMAT_XRGB8888;
 	if (dst_format == DRM_FORMAT_ARGB8888)
 		dst_format = DRM_FORMAT_XRGB8888;
+	if (fb_format == DRM_FORMAT_ARGB2101010)
+		fb_format = DRM_FORMAT_XRGB2101010;
+	if (dst_format == DRM_FORMAT_ARGB2101010)
+		dst_format = DRM_FORMAT_XRGB2101010;
 
 	if (dst_format == fb_format) {
 		drm_fb_memcpy_toio(dst, dst_pitch, vmap, fb, clip);
@@ -515,6 +572,11 @@ int drm_fb_blit_toio(void __iomem *dst, unsigned int dst_pitch, uint32_t dst_for
 			drm_fb_xrgb8888_to_rgb888_toio(dst, dst_pitch, vmap, fb, clip);
 			return 0;
 		}
+	} else if (dst_format == DRM_FORMAT_XRGB2101010) {
+		if (fb_format == DRM_FORMAT_XRGB8888) {
+			drm_fb_xrgb8888_to_xrgb2101010_toio(dst, dst_pitch, vmap, fb, clip);
+			return 0;
+		}
 	}
 
 	return -EINVAL;
diff --git a/include/drm/drm_format_helper.h b/include/drm/drm_format_helper.h
index 97e4c3223af3..b30ed5de0a33 100644
--- a/include/drm/drm_format_helper.h
+++ b/include/drm/drm_format_helper.h
@@ -33,6 +33,9 @@ void drm_fb_xrgb8888_to_rgb888(void *dst, unsigned int dst_pitch, const void *sr
 void drm_fb_xrgb8888_to_rgb888_toio(void __iomem *dst, unsigned int dst_pitch,
 				    const void *vaddr, const struct drm_framebuffer *fb,
 				    const struct drm_rect *clip);
+void drm_fb_xrgb8888_to_xrgb2101010_toio(void __iomem *dst, unsigned int dst_pitch,
+					 const void *vaddr, const struct drm_framebuffer *fb,
+					 const struct drm_rect *clip);
 void drm_fb_xrgb8888_to_gray8(void *dst, unsigned int dst_pitch, const void *vaddr,
 			      const struct drm_framebuffer *fb, const struct drm_rect *clip);
 
-- 
2.33.0

[PATCH v2 3/3] drm/simpledrm: Add XRGB2101010 format

[Hector Martin]

This is the format used by the bootloader framebuffer on Apple ARM64
platforms.

Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Hector Martin <marcan@marcan.st>
---
 drivers/gpu/drm/tiny/simpledrm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/tiny/simpledrm.c b/drivers/gpu/drm/tiny/simpledrm.c
index 2f999915b9aa..edadfd9ee882 100644
--- a/drivers/gpu/drm/tiny/simpledrm.c
+++ b/drivers/gpu/drm/tiny/simpledrm.c
@@ -571,7 +571,7 @@ static const uint32_t simpledrm_default_formats[] = {
 	//DRM_FORMAT_XRGB1555,
 	//DRM_FORMAT_ARGB1555,
 	DRM_FORMAT_RGB888,
-	//DRM_FORMAT_XRGB2101010,
+	DRM_FORMAT_XRGB2101010,
 	//DRM_FORMAT_ARGB2101010,
 };
 
-- 
2.33.0

Re: [PATCH v2 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 2269 bytes --]

Hi

Am 07.12.21 um 08:29 schrieb Hector Martin:
> This code is required for both simplefb and simpledrm, so let's move it
> into the OF core instead of having it as an ad-hoc initcall in the
> drivers.
> 
> Signed-off-by: Hector Martin <marcan@marcan.st>

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>

This looks much better than before. Thank you.

> ---
>   drivers/of/platform.c          |  5 +++++
>   drivers/video/fbdev/simplefb.c | 21 +--------------------
>   2 files changed, 6 insertions(+), 20 deletions(-)
> 
> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> index b3faf89744aa..e097f40b03c0 100644
> --- a/drivers/of/platform.c
> +++ b/drivers/of/platform.c
> @@ -540,6 +540,11 @@ static int __init of_platform_default_populate_init(void)
>   		of_node_put(node);
>   	}
>   
> +	for_each_child_of_node(of_chosen, node) {
> +		if (of_device_is_compatible(node, "simple-framebuffer"))
> +			of_platform_device_create(node, NULL, NULL);
> +	}
> +
>   	/* Populate everything else. */
>   	of_platform_default_populate(NULL, NULL, NULL);
>   
> diff --git a/drivers/video/fbdev/simplefb.c b/drivers/video/fbdev/simplefb.c
> index b63074fd892e..57541887188b 100644
> --- a/drivers/video/fbdev/simplefb.c
> +++ b/drivers/video/fbdev/simplefb.c
> @@ -541,26 +541,7 @@ static struct platform_driver simplefb_driver = {
>   	.remove = simplefb_remove,
>   };
>   
> -static int __init simplefb_init(void)
> -{
> -	int ret;
> -	struct device_node *np;
> -
> -	ret = platform_driver_register(&simplefb_driver);
> -	if (ret)
> -		return ret;
> -
> -	if (IS_ENABLED(CONFIG_OF_ADDRESS) && of_chosen) {
> -		for_each_child_of_node(of_chosen, np) {
> -			if (of_device_is_compatible(np, "simple-framebuffer"))
> -				of_platform_device_create(np, NULL, NULL);
> -		}
> -	}
> -
> -	return 0;
> -}
> -
> -fs_initcall(simplefb_init);
> +module_platform_driver(simplefb_driver);
>   
>   MODULE_AUTHOR("Stephen Warren <swarren@wwwdotorg.org>");
>   MODULE_DESCRIPTION("Simple framebuffer driver");
> 

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

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

Re: [PATCH v2 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 2666 bytes --]



Am 07.12.21 um 10:02 schrieb Thomas Zimmermann:
> Hi
> 
> Am 07.12.21 um 08:29 schrieb Hector Martin:
>> This code is required for both simplefb and simpledrm, so let's move it
>> into the OF core instead of having it as an ad-hoc initcall in the
>> drivers.
>>
>> Signed-off-by: Hector Martin <marcan@marcan.st>
> 
> Acked-by: Thomas Zimmermann <tzimmermann@suse.de>

Well, please don't take this as a review. :)

> 
> This looks much better than before. Thank you.
> 
>> ---
>>   drivers/of/platform.c          |  5 +++++
>>   drivers/video/fbdev/simplefb.c | 21 +--------------------
>>   2 files changed, 6 insertions(+), 20 deletions(-)
>>
>> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
>> index b3faf89744aa..e097f40b03c0 100644
>> --- a/drivers/of/platform.c
>> +++ b/drivers/of/platform.c
>> @@ -540,6 +540,11 @@ static int __init 
>> of_platform_default_populate_init(void)
>>           of_node_put(node);
>>       }
>> +    for_each_child_of_node(of_chosen, node) {
>> +        if (of_device_is_compatible(node, "simple-framebuffer"))
>> +            of_platform_device_create(node, NULL, NULL);
>> +    }
>> +
>>       /* Populate everything else. */
>>       of_platform_default_populate(NULL, NULL, NULL);
>> diff --git a/drivers/video/fbdev/simplefb.c 
>> b/drivers/video/fbdev/simplefb.c
>> index b63074fd892e..57541887188b 100644
>> --- a/drivers/video/fbdev/simplefb.c
>> +++ b/drivers/video/fbdev/simplefb.c
>> @@ -541,26 +541,7 @@ static struct platform_driver simplefb_driver = {
>>       .remove = simplefb_remove,
>>   };
>> -static int __init simplefb_init(void)
>> -{
>> -    int ret;
>> -    struct device_node *np;
>> -
>> -    ret = platform_driver_register(&simplefb_driver);
>> -    if (ret)
>> -        return ret;
>> -
>> -    if (IS_ENABLED(CONFIG_OF_ADDRESS) && of_chosen) {
>> -        for_each_child_of_node(of_chosen, np) {
>> -            if (of_device_is_compatible(np, "simple-framebuffer"))
>> -                of_platform_device_create(np, NULL, NULL);
>> -        }
>> -    }
>> -
>> -    return 0;
>> -}
>> -
>> -fs_initcall(simplefb_init);
>> +module_platform_driver(simplefb_driver);
>>   MODULE_AUTHOR("Stephen Warren <swarren@wwwdotorg.org>");
>>   MODULE_DESCRIPTION("Simple framebuffer driver");
>>
> 

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

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

Re: [PATCH v2 3/3] drm/simpledrm: Add XRGB2101010 format

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 1460 bytes --]

Hi

Am 07.12.21 um 08:29 schrieb Hector Martin:
> This is the format used by the bootloader framebuffer on Apple ARM64
> platforms.
> 
> Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
> Signed-off-by: Hector Martin <marcan@marcan.st>
> ---
>   drivers/gpu/drm/tiny/simpledrm.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/tiny/simpledrm.c b/drivers/gpu/drm/tiny/simpledrm.c
> index 2f999915b9aa..edadfd9ee882 100644
> --- a/drivers/gpu/drm/tiny/simpledrm.c
> +++ b/drivers/gpu/drm/tiny/simpledrm.c
> @@ -571,7 +571,7 @@ static const uint32_t simpledrm_default_formats[] = {
>   	//DRM_FORMAT_XRGB1555,
>   	//DRM_FORMAT_ARGB1555,
>   	DRM_FORMAT_RGB888,
> -	//DRM_FORMAT_XRGB2101010,
> +	DRM_FORMAT_XRGB2101010,
>   	//DRM_FORMAT_ARGB2101010,

You should also enable DRM_FORMAT_ARGB2101010 here. You added the 
conversion function, so DRM can deal with it. Having an alpha channel 
isn't typically supported for primary planes, but the format is listed 
in SIMPLEFB_FORMATS. [1]

With this change:

Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>

Best regards

[1] 
https://elixir.bootlin.com/linux/latest/source/include/linux/platform_data/simplefb.h#L16

>   };
>   
> 

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

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

Re: [PATCH v2 2/3] drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 5077 bytes --]

Hi

Am 07.12.21 um 08:29 schrieb Hector Martin:
> Add XRGB8888 emulation support for devices that can only do XRGB2101010.
> 
> This is chiefly useful for simpledrm on Apple devices where the
> bootloader-provided framebuffer is 10-bit.
> 
> Signed-off-by: Hector Martin <marcan@marcan.st>
> ---
>   drivers/gpu/drm/drm_format_helper.c | 62 +++++++++++++++++++++++++++++
>   include/drm/drm_format_helper.h     |  3 ++
>   2 files changed, 65 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_format_helper.c b/drivers/gpu/drm/drm_format_helper.c
> index dbe3e830096e..edd611d3ab6a 100644
> --- a/drivers/gpu/drm/drm_format_helper.c
> +++ b/drivers/gpu/drm/drm_format_helper.c
> @@ -409,6 +409,59 @@ void drm_fb_xrgb8888_to_rgb888_toio(void __iomem *dst, unsigned int dst_pitch,
>   }
>   EXPORT_SYMBOL(drm_fb_xrgb8888_to_rgb888_toio);
>   
> +static void drm_fb_xrgb8888_to_xrgb2101010_line(u32 *dbuf, const u32 *sbuf,
> +						unsigned int pixels)
> +{
> +	unsigned int x;
> +
> +	for (x = 0; x < pixels; x++) {
> +		*dbuf++ = ((sbuf[x] & 0x000000FF) << 2) |
> +			  ((sbuf[x] & 0x0000FF00) << 4) |
> +			  ((sbuf[x] & 0x00FF0000) << 6);

This isn't quite right. The lowest two destination bits in each 
component will always be zero. You have to do the shifting as above and 
for each component the two highest source bits have to be OR'ed into the 
two lowest destination bits. For example the source bits in a component 
are numbered 7 to 0

  | 7 6 5 4 3 2 1 0 |

then the destination bits should be

  | 7 6 5 4 3 2 1 0 7 6 |

Best regards
Thomas

> +	}
> +}
> +
> +/**
> + * drm_fb_xrgb8888_to_xrgb2101010_toio - Convert XRGB8888 to XRGB2101010 clip
> + * buffer
> + * @dst: XRGB2101010 destination buffer (iomem)
> + * @dst_pitch: Number of bytes between two consecutive scanlines within dst
> + * @vaddr: XRGB8888 source buffer
> + * @fb: DRM framebuffer
> + * @clip: Clip rectangle area to copy
> + *
> + * Drivers can use this function for XRGB2101010 devices that don't natively
> + * support XRGB8888.
> + */
> +void drm_fb_xrgb8888_to_xrgb2101010_toio(void __iomem *dst,
> +					 unsigned int dst_pitch, const void *vaddr,
> +					 const struct drm_framebuffer *fb,
> +					 const struct drm_rect *clip)
> +{
> +	size_t linepixels = clip->x2 - clip->x1;
> +	size_t dst_len = linepixels * sizeof(u32);
> +	unsigned y, lines = clip->y2 - clip->y1;
> +	void *dbuf;
> +
> +	if (!dst_pitch)
> +		dst_pitch = dst_len;
> +
> +	dbuf = kmalloc(dst_len, GFP_KERNEL);
> +	if (!dbuf)
> +		return;
> +
> +	vaddr += clip_offset(clip, fb->pitches[0], sizeof(u32));
> +	for (y = 0; y < lines; y++) {
> +		drm_fb_xrgb8888_to_xrgb2101010_line(dbuf, vaddr, linepixels);
> +		memcpy_toio(dst, dbuf, dst_len);
> +		vaddr += fb->pitches[0];
> +		dst += dst_pitch;
> +	}
> +
> +	kfree(dbuf);
> +}
> +EXPORT_SYMBOL(drm_fb_xrgb8888_to_xrgb2101010_toio);
> +
>   /**
>    * drm_fb_xrgb8888_to_gray8 - Convert XRGB8888 to grayscale
>    * @dst: 8-bit grayscale destination buffer
> @@ -500,6 +553,10 @@ int drm_fb_blit_toio(void __iomem *dst, unsigned int dst_pitch, uint32_t dst_for
>   		fb_format = DRM_FORMAT_XRGB8888;
>   	if (dst_format == DRM_FORMAT_ARGB8888)
>   		dst_format = DRM_FORMAT_XRGB8888;
> +	if (fb_format == DRM_FORMAT_ARGB2101010)
> +		fb_format = DRM_FORMAT_XRGB2101010;
> +	if (dst_format == DRM_FORMAT_ARGB2101010)
> +		dst_format = DRM_FORMAT_XRGB2101010;
>   
>   	if (dst_format == fb_format) {
>   		drm_fb_memcpy_toio(dst, dst_pitch, vmap, fb, clip);
> @@ -515,6 +572,11 @@ int drm_fb_blit_toio(void __iomem *dst, unsigned int dst_pitch, uint32_t dst_for
>   			drm_fb_xrgb8888_to_rgb888_toio(dst, dst_pitch, vmap, fb, clip);
>   			return 0;
>   		}
> +	} else if (dst_format == DRM_FORMAT_XRGB2101010) {
> +		if (fb_format == DRM_FORMAT_XRGB8888) {
> +			drm_fb_xrgb8888_to_xrgb2101010_toio(dst, dst_pitch, vmap, fb, clip);
> +			return 0;
> +		}
>   	}
>   
>   	return -EINVAL;
> diff --git a/include/drm/drm_format_helper.h b/include/drm/drm_format_helper.h
> index 97e4c3223af3..b30ed5de0a33 100644
> --- a/include/drm/drm_format_helper.h
> +++ b/include/drm/drm_format_helper.h
> @@ -33,6 +33,9 @@ void drm_fb_xrgb8888_to_rgb888(void *dst, unsigned int dst_pitch, const void *sr
>   void drm_fb_xrgb8888_to_rgb888_toio(void __iomem *dst, unsigned int dst_pitch,
>   				    const void *vaddr, const struct drm_framebuffer *fb,
>   				    const struct drm_rect *clip);
> +void drm_fb_xrgb8888_to_xrgb2101010_toio(void __iomem *dst, unsigned int dst_pitch,
> +					 const void *vaddr, const struct drm_framebuffer *fb,
> +					 const struct drm_rect *clip);
>   void drm_fb_xrgb8888_to_gray8(void *dst, unsigned int dst_pitch, const void *vaddr,
>   			      const struct drm_framebuffer *fb, const struct drm_rect *clip);
>   
> 

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

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

Re: [PATCH v2 2/3] drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()

[Hector Martin]

Hi, thanks for the review!

On 07/12/2021 18.40, Thomas Zimmermann wrote:
> Hi
> 
> Am 07.12.21 um 08:29 schrieb Hector Martin:
>> Add XRGB8888 emulation support for devices that can only do XRGB2101010.
>>
>> This is chiefly useful for simpledrm on Apple devices where the
>> bootloader-provided framebuffer is 10-bit.
>>
>> Signed-off-by: Hector Martin <marcan@marcan.st>
>> ---
>>    drivers/gpu/drm/drm_format_helper.c | 62 +++++++++++++++++++++++++++++
>>    include/drm/drm_format_helper.h     |  3 ++
>>    2 files changed, 65 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/drm_format_helper.c b/drivers/gpu/drm/drm_format_helper.c
>> index dbe3e830096e..edd611d3ab6a 100644
>> --- a/drivers/gpu/drm/drm_format_helper.c
>> +++ b/drivers/gpu/drm/drm_format_helper.c
>> @@ -409,6 +409,59 @@ void drm_fb_xrgb8888_to_rgb888_toio(void __iomem *dst, unsigned int dst_pitch,
>>    }
>>    EXPORT_SYMBOL(drm_fb_xrgb8888_to_rgb888_toio);
>>    
>> +static void drm_fb_xrgb8888_to_xrgb2101010_line(u32 *dbuf, const u32 *sbuf,
>> +						unsigned int pixels)
>> +{
>> +	unsigned int x;
>> +
>> +	for (x = 0; x < pixels; x++) {
>> +		*dbuf++ = ((sbuf[x] & 0x000000FF) << 2) |
>> +			  ((sbuf[x] & 0x0000FF00) << 4) |
>> +			  ((sbuf[x] & 0x00FF0000) << 6);
> 
> This isn't quite right. The lowest two destination bits in each
> component will always be zero. You have to do the shifting as above and
> for each component the two highest source bits have to be OR'ed into the
> two lowest destination bits. For example the source bits in a component
> are numbered 7 to 0
> 
>    | 7 6 5 4 3 2 1 0 |
> 
> then the destination bits should be
> 
>    | 7 6 5 4 3 2 1 0 7 6 |
> 

I think both approaches have pros and cons. Leaving the two LSBs always 
at 0 yields a fully linear transfer curve with no discontinuities, but 
means the maximum brightness is slightly less than full. Setting them 
fully maps the brightness range, but creates 4 double wide steps in the 
transfer curve (also it's potentially slightly slower CPU-wise).

If you prefer the latter I'll do that for v2.

-- 
Hector Martin (marcan@marcan.st)
Public Key: https://mrcn.st/pub

Re: [PATCH v2 2/3] drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 3117 bytes --]

Hi

Am 07.12.21 um 10:54 schrieb Hector Martin:
> Hi, thanks for the review!
> 
> On 07/12/2021 18.40, Thomas Zimmermann wrote:
>> Hi
>>
>> Am 07.12.21 um 08:29 schrieb Hector Martin:
>>> Add XRGB8888 emulation support for devices that can only do XRGB2101010.
>>>
>>> This is chiefly useful for simpledrm on Apple devices where the
>>> bootloader-provided framebuffer is 10-bit.
>>>
>>> Signed-off-by: Hector Martin <marcan@marcan.st>
>>> ---
>>>    drivers/gpu/drm/drm_format_helper.c | 62 
>>> +++++++++++++++++++++++++++++
>>>    include/drm/drm_format_helper.h     |  3 ++
>>>    2 files changed, 65 insertions(+)
>>>
>>> diff --git a/drivers/gpu/drm/drm_format_helper.c 
>>> b/drivers/gpu/drm/drm_format_helper.c
>>> index dbe3e830096e..edd611d3ab6a 100644
>>> --- a/drivers/gpu/drm/drm_format_helper.c
>>> +++ b/drivers/gpu/drm/drm_format_helper.c
>>> @@ -409,6 +409,59 @@ void drm_fb_xrgb8888_to_rgb888_toio(void __iomem 
>>> *dst, unsigned int dst_pitch,
>>>    }
>>>    EXPORT_SYMBOL(drm_fb_xrgb8888_to_rgb888_toio);
>>> +static void drm_fb_xrgb8888_to_xrgb2101010_line(u32 *dbuf, const u32 
>>> *sbuf,
>>> +                        unsigned int pixels)
>>> +{
>>> +    unsigned int x;
>>> +
>>> +    for (x = 0; x < pixels; x++) {
>>> +        *dbuf++ = ((sbuf[x] & 0x000000FF) << 2) |
>>> +              ((sbuf[x] & 0x0000FF00) << 4) |
>>> +              ((sbuf[x] & 0x00FF0000) << 6);
>>
>> This isn't quite right. The lowest two destination bits in each
>> component will always be zero. You have to do the shifting as above and
>> for each component the two highest source bits have to be OR'ed into the
>> two lowest destination bits. For example the source bits in a component
>> are numbered 7 to 0
>>
>>    | 7 6 5 4 3 2 1 0 |
>>
>> then the destination bits should be
>>
>>    | 7 6 5 4 3 2 1 0 7 6 |
>>
> 
> I think both approaches have pros and cons. Leaving the two LSBs always 
> at 0 yields a fully linear transfer curve with no discontinuities, but 
> means the maximum brightness is slightly less than full. Setting them 
> fully maps the brightness range, but creates 4 double wide steps in the 
> transfer curve (also it's potentially slightly slower CPU-wise).
> 
> If you prefer the latter I'll do that for v2.

We don't give guarantees for color output unless color spaces are 
involved. But the lack of LSB bits can be more visible than larger steps 
in the curve. With the current formats here, it's probably a non-issue. 
But there can be conversions, such as RGB444 to RGB88, where these 
missing LSBs make a visible difference.

Therefore, please change the algorithm. It produces more consistent 
results over a variety of format conversion. It's better to have the 
same (default) algorithm for all of them.

Best regards
Thomas

> 

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

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

Re: [PATCH v2 2/3] drm/format-helper: Add drm_fb_xrgb8888_to_xrgb2101010_toio()

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 3439 bytes --]



Am 07.12.21 um 11:20 schrieb Thomas Zimmermann:
> Hi
> 
> Am 07.12.21 um 10:54 schrieb Hector Martin:
>> Hi, thanks for the review!
>>
>> On 07/12/2021 18.40, Thomas Zimmermann wrote:
>>> Hi
>>>
>>> Am 07.12.21 um 08:29 schrieb Hector Martin:
>>>> Add XRGB8888 emulation support for devices that can only do 
>>>> XRGB2101010.
>>>>
>>>> This is chiefly useful for simpledrm on Apple devices where the
>>>> bootloader-provided framebuffer is 10-bit.
>>>>
>>>> Signed-off-by: Hector Martin <marcan@marcan.st>
>>>> ---
>>>>    drivers/gpu/drm/drm_format_helper.c | 62 
>>>> +++++++++++++++++++++++++++++
>>>>    include/drm/drm_format_helper.h     |  3 ++
>>>>    2 files changed, 65 insertions(+)
>>>>
>>>> diff --git a/drivers/gpu/drm/drm_format_helper.c 
>>>> b/drivers/gpu/drm/drm_format_helper.c
>>>> index dbe3e830096e..edd611d3ab6a 100644
>>>> --- a/drivers/gpu/drm/drm_format_helper.c
>>>> +++ b/drivers/gpu/drm/drm_format_helper.c
>>>> @@ -409,6 +409,59 @@ void drm_fb_xrgb8888_to_rgb888_toio(void 
>>>> __iomem *dst, unsigned int dst_pitch,
>>>>    }
>>>>    EXPORT_SYMBOL(drm_fb_xrgb8888_to_rgb888_toio);
>>>> +static void drm_fb_xrgb8888_to_xrgb2101010_line(u32 *dbuf, const 
>>>> u32 *sbuf,
>>>> +                        unsigned int pixels)
>>>> +{
>>>> +    unsigned int x;
>>>> +
>>>> +    for (x = 0; x < pixels; x++) {
>>>> +        *dbuf++ = ((sbuf[x] & 0x000000FF) << 2) |
>>>> +              ((sbuf[x] & 0x0000FF00) << 4) |
>>>> +              ((sbuf[x] & 0x00FF0000) << 6);
>>>
>>> This isn't quite right. The lowest two destination bits in each
>>> component will always be zero. You have to do the shifting as above and
>>> for each component the two highest source bits have to be OR'ed into the
>>> two lowest destination bits. For example the source bits in a component
>>> are numbered 7 to 0
>>>
>>>    | 7 6 5 4 3 2 1 0 |
>>>
>>> then the destination bits should be
>>>
>>>    | 7 6 5 4 3 2 1 0 7 6 |
>>>
>>
>> I think both approaches have pros and cons. Leaving the two LSBs 
>> always at 0 yields a fully linear transfer curve with no 
>> discontinuities, but means the maximum brightness is slightly less 
>> than full. Setting them fully maps the brightness range, but creates 4 
>> double wide steps in the transfer curve (also it's potentially 
>> slightly slower CPU-wise).
>>
>> If you prefer the latter I'll do that for v2.
> 
> We don't give guarantees for color output unless color spaces are 
> involved. But the lack of LSB bits can be more visible than larger steps 
> in the curve. With the current formats here, it's probably a non-issue. 
> But there can be conversions, such as RGB444 to RGB88, where these 
> missing LSBs make a visible difference.
> 
> Therefore, please change the algorithm. It produces more consistent 
> results over a variety of format conversion. It's better to have the 
> same (default) algorithm for all of them.

FTR, I just tested this in a painting program. I can see a difference 
between ffffff and fcfcfc iff both are next to each other. f8f8f8 is 
obviously gray.

> 
> Best regards
> Thomas
> 
>>
> 

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

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

Re: [PATCH v2 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Rob Herring]

On Tue, Dec 7, 2021 at 1:31 AM Hector Martin <marcan@marcan.st> wrote:
>
> This code is required for both simplefb and simpledrm, so let's move it
> into the OF core instead of having it as an ad-hoc initcall in the
> drivers.
>
> Signed-off-by: Hector Martin <marcan@marcan.st>
> ---
>  drivers/of/platform.c          |  5 +++++
>  drivers/video/fbdev/simplefb.c | 21 +--------------------
>  2 files changed, 6 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> index b3faf89744aa..e097f40b03c0 100644
> --- a/drivers/of/platform.c
> +++ b/drivers/of/platform.c
> @@ -540,6 +540,11 @@ static int __init of_platform_default_populate_init(void)
>                 of_node_put(node);
>         }
>
> +       for_each_child_of_node(of_chosen, node) {
> +               if (of_device_is_compatible(node, "simple-framebuffer"))

node = of_get_compatible_child(of_chosen, "simple-framebuffer");
of_platform_device_create(node, NULL, NULL);
of_node_put(node);

Please Cc the DT list. Looks like this patch can be applied
independently. (Better get the other 2 into drm-misc soon or it will
miss 5.17).

Rob

Re: [PATCH v2 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Thomas Zimmermann]

[-- Attachment #1.1: Type: text/plain, Size: 1599 bytes --]

Hi

Am 08.12.21 um 18:49 schrieb Rob Herring:
> On Tue, Dec 7, 2021 at 1:31 AM Hector Martin <marcan@marcan.st> wrote:
>>
>> This code is required for both simplefb and simpledrm, so let's move it
>> into the OF core instead of having it as an ad-hoc initcall in the
>> drivers.
>>
>> Signed-off-by: Hector Martin <marcan@marcan.st>
>> ---
>>   drivers/of/platform.c          |  5 +++++
>>   drivers/video/fbdev/simplefb.c | 21 +--------------------
>>   2 files changed, 6 insertions(+), 20 deletions(-)
>>
>> diff --git a/drivers/of/platform.c b/drivers/of/platform.c
>> index b3faf89744aa..e097f40b03c0 100644
>> --- a/drivers/of/platform.c
>> +++ b/drivers/of/platform.c
>> @@ -540,6 +540,11 @@ static int __init of_platform_default_populate_init(void)
>>                  of_node_put(node);
>>          }
>>
>> +       for_each_child_of_node(of_chosen, node) {
>> +               if (of_device_is_compatible(node, "simple-framebuffer"))
> 
> node = of_get_compatible_child(of_chosen, "simple-framebuffer");
> of_platform_device_create(node, NULL, NULL);
> of_node_put(node);
> 
> Please Cc the DT list. Looks like this patch can be applied
> independently. (Better get the other 2 into drm-misc soon or it will
> miss 5.17).

Can we merge the whole patchset through drm-misc? Patches 2 and 3 are 
useless without the first one.

Best regards
Thomas

> 
> Rob
> 

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

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