Re: [PATCH 07/11] drm/fbdevdrm: Add DRM <-> fbdev pixel-format conversion

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Tue, Mar 26, 2019 at 10:17:40AM +0100, Thomas Zimmermann wrote:
> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
> ---
>  drivers/gpu/drm/fbdevdrm/Makefile          |   1 +
>  drivers/gpu/drm/fbdevdrm/fbdevdrm_format.c | 441 +++++++++++++++++++++
>  drivers/gpu/drm/fbdevdrm/fbdevdrm_format.h |  26 ++
>  3 files changed, 468 insertions(+)
>  create mode 100644 drivers/gpu/drm/fbdevdrm/fbdevdrm_format.c
>  create mode 100644 drivers/gpu/drm/fbdevdrm/fbdevdrm_format.h
> 
> diff --git a/drivers/gpu/drm/fbdevdrm/Makefile b/drivers/gpu/drm/fbdevdrm/Makefile
> index b8fab9d52faa..aef60d0f4888 100644
> --- a/drivers/gpu/drm/fbdevdrm/Makefile
> +++ b/drivers/gpu/drm/fbdevdrm/Makefile
> @@ -2,6 +2,7 @@ ccflags-y = -Iinclude/drm
>  fbdevdrm-y := fbdevdrm_bo.o \
>  	      fbdevdrm_device.o \
>  	      fbdevdrm_drv.o \
> +	      fbdevdrm_format.o \
>  	      fbdevdrm_modeset.o \
>  	      fbdevdrm_ttm.o
>  
> diff --git a/drivers/gpu/drm/fbdevdrm/fbdevdrm_format.c b/drivers/gpu/drm/fbdevdrm/fbdevdrm_format.c
> new file mode 100644
> index 000000000000..208f7c60e525
> --- /dev/null
> +++ b/drivers/gpu/drm/fbdevdrm/fbdevdrm_format.c
> @@ -0,0 +1,441 @@
> +/* SPDX-License-Identifier: GPL-2.0-or-later
> + *
> + * One purpose of this driver is to allow for easy conversion of framebuffer
> + * drivers to DRM. As a special exception to the GNU GPL, you are allowed to
> + * relicense this file under the terms of a license of your choice if you're
> + * porting a framebuffer driver. In order to do so, update the SPDX license
> + * identifier to the new license and remove this exception.
> + *
> + * If you add code to this file, please ensure that it's compatible with the
> + * stated exception.
> + */
> +
> +#include "fbdevdrm_format.h"
> +#include <asm/byteorder.h>
> +#include <linux/fb.h>
> +
> +#if defined __BIG_ENDIAN
> +#define HOST_FUNC(_func) \
> +	_func ## _be
> +#elif defined __LITTLE_ENDIAN
> +#define HOST_FUNC(_func) \
> +	_func ## _le
> +#else
> +#error "Unsupported endianess"
> +#endif
> +
> +static bool is_c8(const struct fb_info* fb_info)
> +{
> +	return fb_info->var.bits_per_pixel = 8;
> +}
> +
> +static bool is_rgb565_be(const struct fb_info* fb_info)
> +{
> +	return (fb_info->var.bits_per_pixel = 16) &&
> +	       (fb_info->var.red.offset = 0) &&
> +	       (fb_info->var.red.length = 5) &&
> +	       (fb_info->var.green.offset = 5) &&
> +	       (fb_info->var.green.length = 6) &&
> +	       (fb_info->var.blue.offset = 11) &&
> +	       (fb_info->var.blue.length = 5);
> +}

You can't distinguish LE vs. BE like this.

Maybe FBINFO_FOREIGN_ENDIAN is trustworthy?

-- 
Ville Syrjälä
Intel