Hi everybody,
Here's the second version of the fbdev FOURCC-based format configuration API.
Compared to the previous version, I've removed the FB_VMODE_FOURCC bit (FOURCC
mode is now selected by a grayscale value > 1), reorganized the union in the
fb_var_screeninfo structure, and added a colorspace field used for YUV modes.
This patch set also contains an implementation of the FOURCC-based format API
for the sh_mobile_lcdc driver, based on top of the latest patches I've sent to
the list. You can find a consolidated version that includes this patch set at
http://git.linuxtv.org/pinchartl/fbdev.git/shortlog/refs/heads/fbdev-yuv.
YUV support when MERAM is enabled is currently broken, I'm working on fixing
that.
I've updated the fbdev-test tool to add FOURCC support. The code is available
in the fbdev-test yuv branch at
http://git.ideasonboard.org/?pûdev-test.git;a=shortlog;h=refs/heads/yuv.
Laurent Pinchart (3):
fbdev: Add FOURCC-based format configuration API
v4l: Add V4L2_PIX_FMT_NV24 and V4L2_PIX_FMT_NV42 formats
fbdev: sh_mobile_lcdc: Support FOURCC-based format API
Documentation/DocBook/media/v4l/pixfmt-nv24.xml | 128 +++++++++
Documentation/DocBook/media/v4l/pixfmt.xml | 1 +
Documentation/fb/api.txt | 299 ++++++++++++++++++++
arch/arm/mach-shmobile/board-ag5evm.c | 2 +-
arch/arm/mach-shmobile/board-ap4evb.c | 4 +-
arch/arm/mach-shmobile/board-mackerel.c | 4 +-
drivers/video/sh_mobile_lcdcfb.c | 342 +++++++++++++++--------
include/linux/fb.h | 27 ++-
include/linux/videodev2.h | 2 +
include/video/sh_mobile_lcdc.h | 4 +-
10 files changed, 681 insertions(+), 132 deletions(-)
create mode 100644 Documentation/DocBook/media/v4l/pixfmt-nv24.xml
create mode 100644 Documentation/fb/api.txt
--
Regards,
Laurent Pinchart
This API will be used to support YUV frame buffer formats in a standard
way.
Last but not least, create a much needed fbdev API documentation and
document the format setting APIs.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
---
Documentation/fb/api.txt | 299 ++++++++++++++++++++++++++++++++++++++++++++++
include/linux/fb.h | 27 ++++-
2 files changed, 320 insertions(+), 6 deletions(-)
create mode 100644 Documentation/fb/api.txt
@@ -0,0 +1,299 @@+ The Frame Buffer Device API+ ---------------------------++Last revised: June 21, 2011+++0. Introduction+---------------++This document describes the frame buffer API used by applications to interact+with frame buffer devices. In-kernel APIs between device drivers and the frame+buffer core are not described.++Due to a lack of documentation in the original frame buffer API, drivers+behaviours differ in subtle (and not so subtle) ways. This document describes+the recommended API implementation, but applications should be prepared to+deal with different behaviours.+++1. Capabilities+---------------++Device and driver capabilities are reported in the fixed screen information+capabilities field.++struct fb_fix_screeninfo {+ ...+ __u16 capabilities; /* see FB_CAP_* */+ ...+};++Application should use those capabilities to find out what features they can+expect from the device and driver.++- FB_CAP_FOURCC++The driver supports the four character code (FOURCC) based format setting API.+When supported, formats are configured using a FOURCC instead of manually+specifying color components layout.+++2. Types and visuals+--------------------++Pixels are stored in memory in hardware-dependent formats. Applications need+to be aware of the pixel storage format in order to write image data to the+frame buffer memory in the format expected by the hardware.++Formats are described by frame buffer types and visuals. Some visuals require+additional information, which are stored in the variable screen information+bits_per_pixel, grayscale, fourcc, red, green, blue and transp fields.++The following types and visuals are supported.++- FB_TYPE_PACKED_PIXELS++Color components (usually RGB or YUV) are packed together into macropixels+that are stored in a single plane. The exact color components layout is+described in a visual-dependent way.++Frame buffer visuals that don't use multiple color components per pixel+(such as monochrome and pseudo-color visuals) are reported as packed frame+buffer types, even though they don't stricly speaking pack color components+into macropixels.++- FB_TYPE_PLANES++Color components are stored in separate planes. Planes are located+contiguously in memory.++- FB_VISUAL_MONO01++Pixels are black or white and stored on one bit. A bit set to 1 represents a+black pixel and a bit set to 0 a white pixel. Pixels are packed together in+bytes with 8 pixels per byte.++FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.++- FB_VISUAL_MONO10++Pixels are black or white and stored on one bit. A bit set to 1 represents a+white pixel and a bit set to 0 a black pixel. Pixels are packed together in+bytes with 8 pixels per byte.++FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.++- FB_VISUAL_TRUECOLOR++Pixels are broken into red, green and blue components, and each component+indexes a read-only lookup table for the corresponding value. Lookup tables+are device-dependent, and provide linear or non-linear ramps.++Each component is stored in memory according to the variable screen+information red, green, blue and transp fields.++- FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR++Pixel values are encoded as indices into a colormap that stores red, green and+blue components. The colormap is read-only for FB_VISUAL_STATIC_PSEUDOCOLOR+and read-write for FB_VISUAL_PSEUDOCOLOR.++Each pixel value is stored in the number of bits reported by the variable+screen information bits_per_pixel field. Pixels are contiguous in memory.++FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR are used with+FB_TYPE_PACKED_PIXELS only.++- FB_VISUAL_DIRECTCOLOR++Pixels are broken into red, green and blue components, and each component+indexes a programmable lookup table for the corresponding value.++Each component is stored in memory according to the variable screen+information red, green, blue and transp fields.++- FB_VISUAL_FOURCC++Pixels are stored in memory as described by the format FOURCC identifier+stored in the variable screen information fourcc field.+++3. Screen information+---------------------++Screen information are queried by applications using the FBIOGET_FSCREENINFO+and FBIOGET_VSCREENINFO ioctls. Those ioctls take a pointer to a+fb_fix_screeninfo and fb_var_screeninfo structure respectively.++struct fb_fix_screeninfo stores device independent unchangeable information+about the frame buffer device and the current format. Those information can't+be directly modified by applications, but can be changed by the driver when an+application modifies the format.++struct fb_fix_screeninfo {+ char id[16]; /* identification string eg "TT Builtin" */+ unsigned long smem_start; /* Start of frame buffer mem */+ /* (physical address) */+ __u32 smem_len; /* Length of frame buffer mem */+ __u32 type; /* see FB_TYPE_* */+ __u32 type_aux; /* Interleave for interleaved Planes */+ __u32 visual; /* see FB_VISUAL_* */+ __u16 xpanstep; /* zero if no hardware panning */+ __u16 ypanstep; /* zero if no hardware panning */+ __u16 ywrapstep; /* zero if no hardware ywrap */+ __u32 line_length; /* length of a line in bytes */+ unsigned long mmio_start; /* Start of Memory Mapped I/O */+ /* (physical address) */+ __u32 mmio_len; /* Length of Memory Mapped I/O */+ __u32 accel; /* Indicate to driver which */+ /* specific chip/card we have */+ __u16 capabilities; /* see FB_CAP_* */+ __u16 reserved[2]; /* Reserved for future compatibility */+};++struct fb_var_screeninfo stores device independent changeable information+about a frame buffer device, its current format and video mode, as well as+other miscellaneous parameters.++struct fb_var_screeninfo {+ __u32 xres; /* visible resolution */+ __u32 yres;+ __u32 xres_virtual; /* virtual resolution */+ __u32 yres_virtual;+ __u32 xoffset; /* offset from virtual to visible */+ __u32 yoffset; /* resolution */++ __u32 bits_per_pixel; /* guess what */+ union {+ struct { /* Legacy format API */+ __u32 grayscale; /* != 0 Graylevels instead of colors */+ /* bitfields in fb mem if true color, else only */+ /* length is significant */+ struct fb_bitfield red;+ struct fb_bitfield green;+ struct fb_bitfield blue;+ struct fb_bitfield transp; /* transparency */+ };+ struct { /* FOURCC-based format API */+ __u32 fourcc; /* FOURCC format */+ __u32 colorspace;+ __u32 reserved[11];+ } format;+ };++ struct fb_bitfield red; /* bitfield in fb mem if true color, */+ struct fb_bitfield green; /* else only length is significant */+ struct fb_bitfield blue;+ struct fb_bitfield transp; /* transparency */++ __u32 nonstd; /* != 0 Non standard pixel format */++ __u32 activate; /* see FB_ACTIVATE_* */++ __u32 height; /* height of picture in mm */+ __u32 width; /* width of picture in mm */++ __u32 accel_flags; /* (OBSOLETE) see fb_info.flags */++ /* Timing: All values in pixclocks, except pixclock (of course) */+ __u32 pixclock; /* pixel clock in ps (pico seconds) */+ __u32 left_margin; /* time from sync to picture */+ __u32 right_margin; /* time from picture to sync */+ __u32 upper_margin; /* time from sync to picture */+ __u32 lower_margin;+ __u32 hsync_len; /* length of horizontal sync */+ __u32 vsync_len; /* length of vertical sync */+ __u32 sync; /* see FB_SYNC_* */+ __u32 vmode; /* see FB_VMODE_* */+ __u32 rotate; /* angle we rotate counter clockwise */+ __u32 reserved[5]; /* Reserved for future compatibility */+};++To modify variable information, applications call the FBIOPUT_VSCREENINFO+ioctl with a pointer to a fb_var_screeninfo structure. If the call is+successful, the driver will update the fixed screen information accordingly.++Instead of filling the complete fb_var_screeninfo structure manually,+applications should call the FBIOGET_VSCREENINFO ioctl and modify only the+fields they care about.+++4. Format configuration+-----------------------++Frame buffer devices offer two ways to configure the frame buffer format: the+legacy API and the FOURCC-based API.+++The legacy API has been the only frame buffer format configuration API for a+long time and is thus widely used by application. It is the recommended API+for applications when using RGB and grayscale formats, as well as legacy+non-standard formats.++To select a format, applications set the fb_var_screeninfo bits_per_pixel field+to the desired frame buffer depth. Values up to 8 will usually map to+monochrome, grayscale or pseudocolor visuals, although this is not required.++- For grayscale formats, applications set the grayscale field to a non-zero+ value. The red, blue, green and transp fields must be set to 0 by+ applications and ignored by drivers. Drivers must fill the red, blue and+ green offsets to 0 and lengths to the bits_per_pixel value.++- For pseudocolor formats, applications set the grayscale field to a zero+ value. The red, blue, green and transp fields must be set to 0 by+ applications and ignored by drivers. Drivers must fill the red, blue and+ green offsets to 0 and lengths to the bits_per_pixel value.++- For truecolor and directcolor formats, applications set the grayscale field+ to a zero value, and the red, blue, green and transp fields to describe the+ layout of color components in memory.++struct fb_bitfield {+ __u32 offset; /* beginning of bitfield */+ __u32 length; /* length of bitfield */+ __u32 msb_right; /* != 0 : Most significant bit is */+ /* right */+};++ Pixel values are bits_per_pixel wide and are split in non-overlapping red,+ green, blue and alpha (transparency) components. Location and size of each+ component in the pixel value are described by the fb_bitfield offset and+ length fields. Offset are computed from the right.++ Pixels are always stored in an integer number of bytes. If the number of+ bits per pixel is not a multiple of 8, pixel values are padded to the next+ multiple of 8 bits.++Upon successful format configuration, drivers update the fb_fix_screeninfo+type, visual and line_length fields depending on the selected format.+++The FOURCC-based API replaces format descriptions by four character codes+(FOURCC). FOURCCs are abstract identifiers that uniquely define a format+without explicitly describing it. This is the only API that supports YUV+formats. Drivers are also encouraged to implement the FOURCC-based API for RGB+and grayscale formats.++Drivers that support the FOURCC-based API report this capability by setting+the FB_CAP_FOURCC bit in the fb_fix_screeninfo capabilities field.++FOURCC definitions are located in the linux/videodev2.h header. However, and+despite starting with the V4L2_PIX_FMT_prefix, they are not restricted to V4L2+and don't require usage of the V4L2 subsystem. FOURCC documentation is+available in Documentation/DocBook/v4l/pixfmt.xml.++To select a format, applications set the format.fourcc field to the desired+FOURCC. For YUV formats, they should also select the appropriate colorspace by+setting the format.colorspace field to one of the colorspaces listed in+linux/videodev2.h and documented in Documentation/DocBook/v4l/colorspaces.xml.++For forward compatibility reasons the format.reserved field must be set to 0 by+applications and ignored by drivers. Values other than 0 may get a meaning in+future extensions. Note that the grayscale, red, green, blue and transp field+share memory with the format field. Application must thus not touch those+fields when using the FOURCC-based API.++Upon successful format configuration, drivers update the fb_fix_screeninfo+type, visual and line_length fields depending on the selected format. The+visual field is set to FB_VISUAL_FOURCC.
@@ -171,7 +174,8 @@ struct fb_fix_screeninfo {__u32mmio_len;/* Length of Memory Mapped I/O */__u32accel;/* Indicate to driver which *//* specific chip/card we have */-__u16reserved[3];/* Reserved for future compatibility */+__u16capabilities;/* see FB_CAP_* */+__u16reserved[2];/* Reserved for future compatibility */};/* Interpretation of offset for color fields: All offsets are from the right,
@@ -246,12 +250,23 @@ struct fb_var_screeninfo {__u32yoffset;/* resolution */__u32bits_per_pixel;/* guess what */-__u32grayscale;/* != 0 Graylevels instead of colors */-structfb_bitfieldred;/* bitfield in fb mem if true color, */-structfb_bitfieldgreen;/* else only length is significant */-structfb_bitfieldblue;-structfb_bitfieldtransp;/* transparency */+union{+struct{/* Legacy format API */+__u32grayscale;/* != 0 Graylevels instead of colors */+/* bitfields in fb mem if true color, else only */+/* length is significant */+structfb_bitfieldred;+structfb_bitfieldgreen;+structfb_bitfieldblue;+structfb_bitfieldtransp;/* transparency */+};+struct{/* FOURCC-based format API */+__u32fourcc;/* FOURCC format */+__u32colorspace;+__u32reserved[11];+}format;+};__u32nonstd;/* != 0 Non standard pixel format */
@@ -434,7 +471,6 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv){structsh_mobile_lcdc_chan*ch;unsignedlongtmp;-intbpp=0;intk,m;/* Enable LCDC channels. Read data from external memory, avoid using the
@@ -518,7 +557,7 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)lcdc_write_chan(ch,LDDFR,tmp);lcdc_write_chan(ch,LDMLSR,ch->pitch);lcdc_write_chan(ch,LDSA1R,ch->base_addr_y);-if(ch->info->var.nonstd)+if(sh_mobile_format_yuv(&ch->info->var))lcdc_write_chan(ch,LDSA2R,ch->base_addr_c);/* When using deferred I/O mode, configure the LCDC for one-shot
@@ -535,21 +574,23 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)}/* Word and long word swap. */-if(priv->ch[0].info->var.nonstd)+switch(sh_mobile_format_fourcc(&priv->ch[0].info->var)){+caseV4L2_PIX_FMT_RGB565:+caseV4L2_PIX_FMT_NV21:+caseV4L2_PIX_FMT_NV61:+caseV4L2_PIX_FMT_NV42:+tmp=LDDDSR_LS|LDDDSR_WS;+break;+caseV4L2_PIX_FMT_BGR24:+caseV4L2_PIX_FMT_NV12:+caseV4L2_PIX_FMT_NV16:+caseV4L2_PIX_FMT_NV24:tmp=LDDDSR_LS|LDDDSR_WS|LDDDSR_BS;-else{-switch(bpp){-case16:-tmp=LDDDSR_LS|LDDDSR_WS;-break;-case24:-tmp=LDDDSR_LS|LDDDSR_WS|LDDDSR_BS;-break;-case32:-default:-tmp=LDDDSR_LS;-break;-}+break;+caseV4L2_PIX_FMT_BGR32:+default:+tmp=LDDDSR_LS;+break;}lcdc_write(priv,_LDDDSR,tmp);
@@ -621,12 +662,24 @@ static int sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)ch->meram_enabled=0;}-if(!ch->info->var.nonstd)-pixelformat=SH_MOBILE_MERAM_PF_RGB;-elseif(ch->info->var.bits_per_pixel=24)-pixelformat=SH_MOBILE_MERAM_PF_NV24;-else+switch(sh_mobile_format_fourcc(&ch->info->var)){+caseV4L2_PIX_FMT_NV12:+caseV4L2_PIX_FMT_NV21:+caseV4L2_PIX_FMT_NV16:+caseV4L2_PIX_FMT_NV61:pixelformat=SH_MOBILE_MERAM_PF_NV;+break;+caseV4L2_PIX_FMT_NV24:+caseV4L2_PIX_FMT_NV42:+pixelformat=SH_MOBILE_MERAM_PF_NV24;+break;+caseV4L2_PIX_FMT_RGB565:+caseV4L2_PIX_FMT_BGR24:+caseV4L2_PIX_FMT_BGR32:+default:+pixelformat=SH_MOBILE_MERAM_PF_RGB;+break;+}ret=mdev->ops->meram_register(mdev,cfg,ch->pitch,ch->info->var.yres,pixelformat,
@@ -876,8 +930,9 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,unsignedlongnew_pan_offset;unsignedlongbase_addr_y,base_addr_c;unsignedlongc_offset;+boolyuv=sh_mobile_format_yuv(&info->var);-if(!info->var.nonstd)+if(!yuv)new_pan_offset=var->yoffset*info->fix.line_length+var->xoffset*(info->var.bits_per_pixel/8);else
@@ -891,7 +946,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,/* Set the source address for the next refresh */base_addr_y=ch->dma_handle+new_pan_offset;-if(info->var.nonstd){+if(yuv){/* Set y offset */c_offset=var->yoffset*info->fix.line_length*(info->var.bits_per_pixel-8)/8;
@@ -899,7 +954,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,+info->var.xres*info->var.yres_virtual+c_offset;/* Set x offset */-if(info->var.bits_per_pixel=24)+if(sh_mobile_format_fourcc(&info->var)=V4L2_PIX_FMT_NV24)base_addr_c+=2*var->xoffset;elsebase_addr_c+=var->xoffset;
@@ -923,7 +978,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,ch->base_addr_c=base_addr_c;lcdc_write_chan_mirror(ch,LDSA1R,base_addr_y);-if(info->var.nonstd)+if(yuv)lcdc_write_chan_mirror(ch,LDSA2R,base_addr_c);if(lcdc_chan_is_sublcd(ch))
@@ -1099,51 +1154,78 @@ static int sh_mobile_check_var(struct fb_var_screeninfo *var, struct fb_info *inif(var->yres_virtual<var->yres)var->yres_virtual=var->yres;-if(var->bits_per_pixel<=16){/* RGB 565 */-var->bits_per_pixel=16;-var->red.offset=11;-var->red.length=5;-var->green.offset=5;-var->green.length=6;-var->blue.offset=0;-var->blue.length=5;-var->transp.offset=0;-var->transp.length=0;-}elseif(var->bits_per_pixel<=24){/* RGB 888 */-var->bits_per_pixel=24;-var->red.offset=16;-var->red.length=8;-var->green.offset=8;-var->green.length=8;-var->blue.offset=0;-var->blue.length=8;-var->transp.offset=0;-var->transp.length=0;-}elseif(var->bits_per_pixel<=32){/* RGBA 888 */-var->bits_per_pixel=32;-var->red.offset=16;-var->red.length=8;-var->green.offset=8;-var->green.length=8;-var->blue.offset=0;-var->blue.length=8;-var->transp.offset=24;-var->transp.length=8;-}else-return-EINVAL;+if(var->format.fourcc>1){+switch(var->format.fourcc){+caseV4L2_PIX_FMT_NV12:+caseV4L2_PIX_FMT_NV21:+var->bits_per_pixel=12;+break;+caseV4L2_PIX_FMT_RGB565:+caseV4L2_PIX_FMT_NV16:+caseV4L2_PIX_FMT_NV61:+var->bits_per_pixel=16;+break;+caseV4L2_PIX_FMT_BGR24:+caseV4L2_PIX_FMT_NV24:+caseV4L2_PIX_FMT_NV42:+var->bits_per_pixel=24;+break;+caseV4L2_PIX_FMT_BGR32:+var->bits_per_pixel=32;+break;+default:+return-EINVAL;+}++memset(var->format.reserved,0,sizeof(var->format.reserved));+}else{+if(var->bits_per_pixel<=16){/* RGB 565 */+var->bits_per_pixel=16;+var->red.offset=11;+var->red.length=5;+var->green.offset=5;+var->green.length=6;+var->blue.offset=0;+var->blue.length=5;+var->transp.offset=0;+var->transp.length=0;+}elseif(var->bits_per_pixel<=24){/* RGB 888 */+var->bits_per_pixel=24;+var->red.offset=16;+var->red.length=8;+var->green.offset=8;+var->green.length=8;+var->blue.offset=0;+var->blue.length=8;+var->transp.offset=0;+var->transp.length=0;+}elseif(var->bits_per_pixel<=32){/* RGBA 888 */+var->bits_per_pixel=32;+var->red.offset=16;+var->red.length=8;+var->green.offset=8;+var->green.length=8;+var->blue.offset=0;+var->blue.length=8;+var->transp.offset=24;+var->transp.length=8;+}else+return-EINVAL;-var->red.msb_right=0;-var->green.msb_right=0;-var->blue.msb_right=0;-var->transp.msb_right=0;+var->red.msb_right=0;+var->green.msb_right=0;+var->blue.msb_right=0;+var->transp.msb_right=0;+}/* Make sure we don't exceed our allocated memory. */if(var->xres_virtual*var->yres_virtual*var->bits_per_pixel/8>info->fix.smem_len)return-EINVAL;-/* only accept the forced_bpp for dual channel configurations */-if(p->forced_bpp&&p->forced_bpp!=var->bits_per_pixel)+/* only accept the forced_fourcc for dual channel configurations */+if(p->forced_fourcc&&+p->forced_fourcc!=sh_mobile_format_fourcc(var))return-EINVAL;return0;
@@ -1157,7 +1239,7 @@ static int sh_mobile_set_par(struct fb_info *info)sh_mobile_lcdc_stop(ch->lcdc);-if(info->var.nonstd)+if(sh_mobile_format_yuv(&info->var))info->fix.line_length=info->var.xres;elseinfo->fix.line_length=info->var.xres
@@ -1169,6 +1251,11 @@ static int sh_mobile_set_par(struct fb_info *info)info->fix.line_length=line_length;}+if(info->var.format.fourcc>1)+info->fix.visual=FB_VISUAL_FOURCC;+else+info->fix.visual=FB_VISUAL_TRUECOLOR;+returnret;}
@@ -1463,9 +1550,9 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch,for(i=0,mode=cfg->lcd_cfg;i<cfg->num_cfg;i++,mode++){unsignedintsize=mode->yres*mode->xres;-/* NV12 buffers must have even number of lines */-if((cfg->nonstd)&&cfg->bpp=12&&-(mode->yres&0x1)){+/* NV12/NV21 buffers must have even number of lines */+if((cfg->fourcc=V4L2_PIX_FMT_NV12||+cfg->fourcc=V4L2_PIX_FMT_NV21)&&(mode->yres&0x1)){dev_err(dev,"yres must be multiple of 2 for YCbCr420 ""mode.\n");return-EINVAL;
@@ -1483,14 +1570,6 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch,dev_dbg(dev,"Found largest videomode %ux%u\n",max_mode->xres,max_mode->yres);-/* Initialize fixed screen information. Restrict pan to 2 lines steps-*forNV12.-*/-info->fix=sh_mobile_lcdc_fix;-info->fix.smem_len=max_size*2*cfg->bpp/8;-if(cfg->nonstd&&cfg->bpp=12)-info->fix.ypanstep=2;-/* Create the mode list. */if(cfg->lcd_cfg=NULL){mode=&default_720p;
@@ -1508,19 +1587,38 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch,*/var=&info->var;fb_videomode_to_var(var,mode);-var->bits_per_pixel=cfg->bpp;var->width=cfg->lcd_size_cfg.width;var->height=cfg->lcd_size_cfg.height;var->yres_virtual=var->yres*2;var->activate=FB_ACTIVATE_NOW;+switch(cfg->fourcc){+caseV4L2_PIX_FMT_RGB565:+var->bits_per_pixel=16;+break;+caseV4L2_PIX_FMT_BGR24:+var->bits_per_pixel=24;+break;+caseV4L2_PIX_FMT_BGR32:+var->bits_per_pixel=32;+break;+default:+var->format.fourcc=cfg->fourcc;+break;+}++/* Make sure the memory size check won't fail. smem_len is initialized+*laterbasedonvar.+*/+info->fix.smem_len=UINT_MAX;ret=sh_mobile_check_var(var,info);if(ret)returnret;+max_size*=var->bits_per_pixel/8*2;+/* Allocate frame buffer memory and color map. */-buf=dma_alloc_coherent(dev,info->fix.smem_len,&ch->dma_handle,-GFP_KERNEL);+buf=dma_alloc_coherent(dev,max_size,&ch->dma_handle,GFP_KERNEL);if(!buf){dev_err(dev,"unable to allocate buffer\n");return-ENOMEM;
@@ -1529,16 +1627,25 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch,ret=fb_alloc_cmap(&info->cmap,PALETTE_NR,0);if(ret<0){dev_err(dev,"unable to allocate cmap\n");-dma_free_coherent(dev,info->fix.smem_len,-buf,ch->dma_handle);+dma_free_coherent(dev,max_size,buf,ch->dma_handle);returnret;}+/* Initialize fixed screen information. Restrict pan to 2 lines steps+*forNV12andNV21.+*/+info->fix=sh_mobile_lcdc_fix;info->fix.smem_start=ch->dma_handle;-if(var->nonstd)+info->fix.smem_len=max_size*var->bits_per_pixel/8*2;+if(cfg->fourcc=V4L2_PIX_FMT_NV12||+cfg->fourcc=V4L2_PIX_FMT_NV21)+info->fix.ypanstep=2;++if(sh_mobile_format_yuv(var))info->fix.line_length=var->xres;else-info->fix.line_length=var->xres*(cfg->bpp/8);+info->fix.line_length=var->xres*var->bits_per_pixel+/8;info->screen_base=buf;info->device=dev;
@@ -1625,9 +1732,9 @@ static int __devinit sh_mobile_lcdc_probe(struct platform_device *pdev)gotoerr1;}-/* for dual channel LCDC (MAIN + SUB) force shared bpp setting */+/* for dual channel LCDC (MAIN + SUB) force shared format setting */if(num_channels=2)-priv->forced_bpp=pdata->ch[0].bpp;+priv->forced_fourcc=pdata->ch[0].fourcc;priv->base=ioremap_nocache(res->start,resource_size(res));if(!priv->base)
@@ -1674,13 +1781,10 @@ static int __devinit sh_mobile_lcdc_probe(struct platform_device *pdev)if(error<0)gotoerr1;-dev_info(info->dev,-"registered %s/%s as %dx%d %dbpp.\n",-pdev->name,-(ch->cfg.chan=LCDC_CHAN_MAINLCD)?-"mainlcd":"sublcd",-info->var.xres,info->var.yres,-ch->cfg.bpp);+dev_info(info->dev,"registered %s/%s as %dx%d %dbpp.\n",+pdev->name,(ch->cfg.chan=LCDC_CHAN_MAINLCD)?+"mainlcd":"sublcd",info->var.xres,info->var.yres,+info->var.bits_per_pixel);/* deferred io mode: disable clock to save power */if(info->fbdefio||info->state=FBINFO_STATE_SUSPENDED)
Hi Laurent,
hope we're close to the final thing now. Just a few minor issues.
On 08/19/2011 09:37 AM, Laurent Pinchart wrote:
This API will be used to support YUV frame buffer formats in a standard
way.
Last but not least, create a much needed fbdev API documentation and
document the format setting APIs.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
---
Documentation/fb/api.txt | 299 ++++++++++++++++++++++++++++++++++++++++++++++
include/linux/fb.h | 27 ++++-
2 files changed, 320 insertions(+), 6 deletions(-)
create mode 100644 Documentation/fb/api.txt
@@ -171,7 +174,8 @@ struct fb_fix_screeninfo {__u32mmio_len;/* Length of Memory Mapped I/O */__u32accel;/* Indicate to driver which *//* specific chip/card we have */-__u16reserved[3];/* Reserved for future compatibility */+__u16capabilities;/* see FB_CAP_* */+__u16reserved[2];/* Reserved for future compatibility */};/* Interpretation of offset for color fields: All offsets are from the right,
@@ -246,12 +250,23 @@ struct fb_var_screeninfo {__u32yoffset;/* resolution */__u32bits_per_pixel;/* guess what */-__u32grayscale;/* != 0 Graylevels instead of colors */-structfb_bitfieldred;/* bitfield in fb mem if true color, */-structfb_bitfieldgreen;/* else only length is significant */-structfb_bitfieldblue;-structfb_bitfieldtransp;/* transparency */+union{+struct{/* Legacy format API */+__u32grayscale;/* != 0 Graylevels instead of colors */
You should adjust the comment as well, to avoid misleading crazy people ;)
Needs also be fixed in the documentation at some places.
+ /* bitfields in fb mem if true color, else only */
+ /* length is significant */
+ struct fb_bitfield red;
+ struct fb_bitfield green;
+ struct fb_bitfield blue;
+ struct fb_bitfield transp; /* transparency */
+ };
+ struct { /* FOURCC-based format API */
+ __u32 fourcc; /* FOURCC format */
+ __u32 colorspace;
So we have again fields that are not always used. Okay, as we still have 11 left
that shouldn't be a big problem, I think.
+ __u32 reserved[11];
+ } format;
Ugh, if you want this union to have a name I suggest 'fourcc' and not 'format'
as the other struct contains format information as well and who knows, maybe in
10 or 20 years we'll have yet another format description that can do things none
of the existing can do.
+ };
__u32 nonstd; /* != 0 Non standard pixel format */
@@ -434,7 +471,6 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv){structsh_mobile_lcdc_chan*ch;unsignedlongtmp;-intbpp=0;intk,m;/* Enable LCDC channels. Read data from external memory, avoid using the
@@ -518,7 +557,7 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)lcdc_write_chan(ch,LDDFR,tmp);lcdc_write_chan(ch,LDMLSR,ch->pitch);lcdc_write_chan(ch,LDSA1R,ch->base_addr_y);-if(ch->info->var.nonstd)+if(sh_mobile_format_yuv(&ch->info->var))lcdc_write_chan(ch,LDSA2R,ch->base_addr_c);/* When using deferred I/O mode, configure the LCDC for one-shot
@@ -535,21 +574,23 @@ static void __sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)}/* Word and long word swap. */-if(priv->ch[0].info->var.nonstd)+switch(sh_mobile_format_fourcc(&priv->ch[0].info->var)){+caseV4L2_PIX_FMT_RGB565:+caseV4L2_PIX_FMT_NV21:+caseV4L2_PIX_FMT_NV61:+caseV4L2_PIX_FMT_NV42:+tmp=LDDDSR_LS|LDDDSR_WS;+break;+caseV4L2_PIX_FMT_BGR24:+caseV4L2_PIX_FMT_NV12:+caseV4L2_PIX_FMT_NV16:+caseV4L2_PIX_FMT_NV24:tmp=LDDDSR_LS|LDDDSR_WS|LDDDSR_BS;-else{-switch(bpp){-case16:-tmp=LDDDSR_LS|LDDDSR_WS;-break;-case24:-tmp=LDDDSR_LS|LDDDSR_WS|LDDDSR_BS;-break;-case32:-default:-tmp=LDDDSR_LS;-break;-}+break;+caseV4L2_PIX_FMT_BGR32:+default:+tmp=LDDDSR_LS;+break;}lcdc_write(priv,_LDDDSR,tmp);
@@ -621,12 +662,24 @@ static int sh_mobile_lcdc_start(struct sh_mobile_lcdc_priv *priv)ch->meram_enabled=0;}-if(!ch->info->var.nonstd)-pixelformat=SH_MOBILE_MERAM_PF_RGB;-elseif(ch->info->var.bits_per_pixel=24)-pixelformat=SH_MOBILE_MERAM_PF_NV24;-else+switch(sh_mobile_format_fourcc(&ch->info->var)){+caseV4L2_PIX_FMT_NV12:+caseV4L2_PIX_FMT_NV21:+caseV4L2_PIX_FMT_NV16:+caseV4L2_PIX_FMT_NV61:pixelformat=SH_MOBILE_MERAM_PF_NV;+break;+caseV4L2_PIX_FMT_NV24:+caseV4L2_PIX_FMT_NV42:+pixelformat=SH_MOBILE_MERAM_PF_NV24;+break;+caseV4L2_PIX_FMT_RGB565:+caseV4L2_PIX_FMT_BGR24:+caseV4L2_PIX_FMT_BGR32:+default:+pixelformat=SH_MOBILE_MERAM_PF_RGB;+break;+}ret=mdev->ops->meram_register(mdev,cfg,ch->pitch,ch->info->var.yres,pixelformat,
@@ -876,8 +930,9 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,unsignedlongnew_pan_offset;unsignedlongbase_addr_y,base_addr_c;unsignedlongc_offset;+boolyuv=sh_mobile_format_yuv(&info->var);-if(!info->var.nonstd)+if(!yuv)new_pan_offset=var->yoffset*info->fix.line_length+var->xoffset*(info->var.bits_per_pixel/8);else
@@ -891,7 +946,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,/* Set the source address for the next refresh */base_addr_y=ch->dma_handle+new_pan_offset;-if(info->var.nonstd){+if(yuv){/* Set y offset */c_offset=var->yoffset*info->fix.line_length*(info->var.bits_per_pixel-8)/8;
@@ -899,7 +954,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,+info->var.xres*info->var.yres_virtual+c_offset;/* Set x offset */-if(info->var.bits_per_pixel=24)+if(sh_mobile_format_fourcc(&info->var)=V4L2_PIX_FMT_NV24)base_addr_c+=2*var->xoffset;elsebase_addr_c+=var->xoffset;
@@ -923,7 +978,7 @@ static int sh_mobile_fb_pan_display(struct fb_var_screeninfo *var,ch->base_addr_c=base_addr_c;lcdc_write_chan_mirror(ch,LDSA1R,base_addr_y);-if(info->var.nonstd)+if(yuv)lcdc_write_chan_mirror(ch,LDSA2R,base_addr_c);if(lcdc_chan_is_sublcd(ch))
If we decide to use another of the reserved area this won't have the desired
behavior as the behavior of this driver will change even if it does not support
the new field. Probably the best thing is to get the desired behavior is zeroing
the whole struct and setting the supported fields to the actual values.
You should check and adjust colorspace here as well.
@@ -1463,9 +1550,9 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch, for (i = 0, mode = cfg->lcd_cfg; i < cfg->num_cfg; i++, mode++) { unsigned int size = mode->yres * mode->xres;- /* NV12 buffers must have even number of lines */- if ((cfg->nonstd) && cfg->bpp = 12 &&- (mode->yres & 0x1)) {+ /* NV12/NV21 buffers must have even number of lines */+ if ((cfg->fourcc = V4L2_PIX_FMT_NV12 ||+ cfg->fourcc = V4L2_PIX_FMT_NV21) && (mode->yres & 0x1)) { dev_err(dev, "yres must be multiple of 2 for YCbCr420 " "mode.\n"); return -EINVAL;
@@ -1483,14 +1570,6 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch, dev_dbg(dev, "Found largest videomode %ux%u\n", max_mode->xres, max_mode->yres);- /* Initialize fixed screen information. Restrict pan to 2 lines steps- * for NV12.- */- info->fix = sh_mobile_lcdc_fix;- info->fix.smem_len = max_size * 2 * cfg->bpp / 8;- if (cfg->nonstd && cfg->bpp = 12)- info->fix.ypanstep = 2;- /* Create the mode list. */ if (cfg->lcd_cfg = NULL) { mode = &default_720p;
@@ -1508,19 +1587,38 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch, */ var = &info->var; fb_videomode_to_var(var, mode);- var->bits_per_pixel = cfg->bpp; var->width = cfg->lcd_size_cfg.width; var->height = cfg->lcd_size_cfg.height; var->yres_virtual = var->yres * 2; var->activate = FB_ACTIVATE_NOW;+ switch (cfg->fourcc) {+ case V4L2_PIX_FMT_RGB565:+ var->bits_per_pixel = 16;+ break;+ case V4L2_PIX_FMT_BGR24:+ var->bits_per_pixel = 24;+ break;+ case V4L2_PIX_FMT_BGR32:+ var->bits_per_pixel = 32;+ break;+ default:+ var->format.fourcc = cfg->fourcc;+ break;+ }++ /* Make sure the memory size check won't fail. smem_len is initialized+ * later based on var.+ */+ info->fix.smem_len = UINT_MAX; ret = sh_mobile_check_var(var, info); if (ret) return ret;+ max_size *= var->bits_per_pixel / 8 * 2;+ /* Allocate frame buffer memory and color map. */- buf = dma_alloc_coherent(dev, info->fix.smem_len, &ch->dma_handle,- GFP_KERNEL);+ buf = dma_alloc_coherent(dev, max_size, &ch->dma_handle, GFP_KERNEL); if (!buf) { dev_err(dev, "unable to allocate buffer\n"); return -ENOMEM;
@@ -1529,16 +1627,25 @@ static int __devinit sh_mobile_lcdc_channel_init(struct sh_mobile_lcdc_chan *ch, ret = fb_alloc_cmap(&info->cmap, PALETTE_NR, 0); if (ret < 0) { dev_err(dev, "unable to allocate cmap\n");- dma_free_coherent(dev, info->fix.smem_len,- buf, ch->dma_handle);+ dma_free_coherent(dev, max_size, buf, ch->dma_handle); return ret; }+ /* Initialize fixed screen information. Restrict pan to 2 lines steps+ * for NV12 and NV21.+ */+ info->fix = sh_mobile_lcdc_fix; info->fix.smem_start = ch->dma_handle;- if (var->nonstd)+ info->fix.smem_len = max_size * var->bits_per_pixel / 8 * 2;+ if (cfg->fourcc = V4L2_PIX_FMT_NV12 ||+ cfg->fourcc = V4L2_PIX_FMT_NV21)+ info->fix.ypanstep = 2;++ if (sh_mobile_format_yuv(var)) info->fix.line_length = var->xres; else- info->fix.line_length = var->xres * (cfg->bpp / 8);+ info->fix.line_length = var->xres * var->bits_per_pixel+ / 8; info->screen_base = buf; info->device = dev;
@@ -1625,9 +1732,9 @@ static int __devinit sh_mobile_lcdc_probe(struct platform_device *pdev) goto err1; }- /* for dual channel LCDC (MAIN + SUB) force shared bpp setting */+ /* for dual channel LCDC (MAIN + SUB) force shared format setting */ if (num_channels = 2)- priv->forced_bpp = pdata->ch[0].bpp;+ priv->forced_fourcc = pdata->ch[0].fourcc; priv->base = ioremap_nocache(res->start, resource_size(res)); if (!priv->base)
@@ -1674,13 +1781,10 @@ static int __devinit sh_mobile_lcdc_probe(struct platform_device *pdev) if (error < 0) goto err1;- dev_info(info->dev,- "registered %s/%s as %dx%d %dbpp.\n",- pdev->name,- (ch->cfg.chan = LCDC_CHAN_MAINLCD) ?- "mainlcd" : "sublcd",- info->var.xres, info->var.yres,- ch->cfg.bpp);+ dev_info(info->dev, "registered %s/%s as %dx%d %dbpp.\n",+ pdev->name, (ch->cfg.chan = LCDC_CHAN_MAINLCD) ?+ "mainlcd" : "sublcd", info->var.xres, info->var.yres,+ info->var.bits_per_pixel); /* deferred io mode: disable clock to save power */ if (info->fbdefio || info->state = FBINFO_STATE_SUSPENDED)
@@ -1099,51 +1154,78 @@ static int sh_mobile_check_var(struct
[snip]
quoted
+ if (var->format.fourcc > 1) {
+ switch (var->format.fourcc) {
+ case V4L2_PIX_FMT_NV12:
+ case V4L2_PIX_FMT_NV21:
+ var->bits_per_pixel = 12;
+ break;
+ case V4L2_PIX_FMT_RGB565:
+ case V4L2_PIX_FMT_NV16:
+ case V4L2_PIX_FMT_NV61:
+ var->bits_per_pixel = 16;
+ break;
+ case V4L2_PIX_FMT_BGR24:
+ case V4L2_PIX_FMT_NV24:
+ case V4L2_PIX_FMT_NV42:
+ var->bits_per_pixel = 24;
+ break;
+ case V4L2_PIX_FMT_BGR32:
+ var->bits_per_pixel = 32;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ memset(var->format.reserved, 0, sizeof(var->format.reserved));
If we decide to use another of the reserved area this won't have the
desired behavior as the behavior of this driver will change even if it
does not support the new field. Probably the best thing is to get the
desired behavior is zeroing the whole struct and setting the supported
fields to the actual values. You should check and adjust colorspace here
as well.
Agreed. I'll fix the patch accordingly.
--
Regards,
Laurent Pinchart
Hi Laurent, thanks for the patch!
Since you're changing the LCDC platform data please make sure you also
update the 5 boards using the LCDC under arch/sh.
Thanks,
/ magnus
+2. Types and visuals
+--------------------
+
+Pixels are stored in memory in hardware-dependent formats. Applications need
+to be aware of the pixel storage format in order to write image data to the
+frame buffer memory in the format expected by the hardware.
+
+Formats are described by frame buffer types and visuals. Some visuals require
+additional information, which are stored in the variable screen information
+bits_per_pixel, grayscale, fourcc, red, green, blue and transp fields.
+
+The following types and visuals are supported.
+
+- FB_TYPE_PACKED_PIXELS
+
+Color components (usually RGB or YUV) are packed together into macropixels
+that are stored in a single plane. The exact color components layout is
+described in a visual-dependent way.
+
+Frame buffer visuals that don't use multiple color components per pixel
+(such as monochrome and pseudo-color visuals) are reported as packed frame
+buffer types, even though they don't stricly speaking pack color components
+into macropixels.
That's because the "packing" is not about the color components, but about the
bits that represent a single pixel.
I.e. the bits that make up the pixel (the macropixel) are stored next
to each other
in memory.
+- FB_TYPE_PLANES
+
+Color components are stored in separate planes. Planes are located
+contiguously in memory.
The bits that make up a pixel are stored in separate planes. Planes are located
contiguously in memory.
- FB_TYPE_INTERLEAVED_PLANES
The bits that make up a pixel are stored in separate planes. Planes
are interleaved.
The interleave factor (the distance in bytes between the planes in
memory) is stored
in the type_aux field.
+- FB_VISUAL_MONO01
+
+Pixels are black or white and stored on one bit. A bit set to 1 represents a
+black pixel and a bit set to 0 a white pixel. Pixels are packed together in
+bytes with 8 pixels per byte.
Actually we do have drivers that use 8 bits per pixel for a monochrome visual.
Hence:
"Pixels are black or white. A black pixel is represented by all
(typically one) bits
set to ones, a white pixel by all bits set to zeroes."
+FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.
... so this may also not be true (but it is for all current drivers, IIRC).
There's a strict orthogonality between type (how is a pixel stored in memory)
and visual (how the bits that represent the pixel are interpreted and converted
to a color value).
Same comments for FB_VISUAL_MONO10
+- FB_VISUAL_TRUECOLOR
+
+Pixels are broken into red, green and blue components, and each component
+indexes a read-only lookup table for the corresponding value. Lookup tables
+are device-dependent, and provide linear or non-linear ramps.
+
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable screen
information red, green, blue and transp fields."
Storage format in memory is determined by the FB_TYPE_* value.
+- FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR
+
+Pixel values are encoded as indices into a colormap that stores red, green and
+blue components. The colormap is read-only for FB_VISUAL_STATIC_PSEUDOCOLOR
+and read-write for FB_VISUAL_PSEUDOCOLOR.
+
+Each pixel value is stored in the number of bits reported by the variable
+screen information bits_per_pixel field. Pixels are contiguous in memory.
Whether pixels are contiguous in memory or not is determined by the
FB_TYPE_* value.
+FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR are used with
+FB_TYPE_PACKED_PIXELS only.
Not true. Several drivers use bit planes or interleaved bitplanes.
+- FB_VISUAL_DIRECTCOLOR
+
+Pixels are broken into red, green and blue components, and each component
+indexes a programmable lookup table for the corresponding value.
+
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable screen
information red, green, blue and transp fields."
+- FB_VISUAL_FOURCC
+
+Pixels are stored in memory as described by the format FOURCC identifier
+stored in the variable screen information fourcc field.
... stored in memory and interpreted ...
+struct fb_var_screeninfo {
+ __u32 xres; /* visible resolution */
+ __u32 yres;
+ __u32 xres_virtual; /* virtual resolution */
+ __u32 yres_virtual;
+ __u32 xoffset; /* offset from virtual to visible */
+ __u32 yoffset; /* resolution */
+
+ __u32 bits_per_pixel; /* guess what */
+ union {
+ struct { /* Legacy format API */
+ __u32 grayscale; /* != 0 Graylevels instead of colors */
+ /* bitfields in fb mem if true color, else only */
+ /* length is significant */
+ struct fb_bitfield red;
+ struct fb_bitfield green;
+ struct fb_bitfield blue;
+ struct fb_bitfield transp; /* transparency */
+ };
+ struct { /* FOURCC-based format API */
+ __u32 fourcc; /* FOURCC format */
+ __u32 colorspace;
+ __u32 reserved[11];
+ } format;
+ };
+
+ struct fb_bitfield red; /* bitfield in fb mem if true color, */
+ struct fb_bitfield green; /* else only length is significant */
+ struct fb_bitfield blue;
+ struct fb_bitfield transp; /* transparency */
These four are duplicated, cfr. the union above.
+ Pixel values are bits_per_pixel wide and are split in non-overlapping red,
+ green, blue and alpha (transparency) components. Location and size of each
+ component in the pixel value are described by the fb_bitfield offset and
+ length fields. Offset are computed from the right.
Offsets
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Hi Laurent, thanks for the patch!
Since you're changing the LCDC platform data please make sure you also
update the 5 boards using the LCDC under arch/sh.
Sure. Sorry for forgetting about that.
--
Regards,
Laurent Pinchart
Hi Geert,
Thanks for the review.
On Monday 29 August 2011 10:13:07 Geert Uytterhoeven wrote:
On Fri, Aug 19, 2011 at 11:37, Laurent Pinchart wrote:
[snip]
quoted
+- FB_TYPE_PACKED_PIXELS
+
+Color components (usually RGB or YUV) are packed together into
macropixels +that are stored in a single plane. The exact color
components layout is +described in a visual-dependent way.
+
+Frame buffer visuals that don't use multiple color components per pixel
+(such as monochrome and pseudo-color visuals) are reported as packed
frame +buffer types, even though they don't stricly speaking pack color
components +into macropixels.
That's because the "packing" is not about the color components, but about
the bits that represent a single pixel.
I.e. the bits that make up the pixel (the macropixel) are stored next
to each other
in memory.
OK, I've modified that last sentence to read
"Frame buffer visuals that don't use multiple color components per pixel (such
as monochrome and pseudo-color visuals) are also reported as packed frame
buffer types, as the bits that make up individual pixels are packed next to
each other in memory."
quoted
+- FB_TYPE_PLANES
+
+Color components are stored in separate planes. Planes are located
+contiguously in memory.
The bits that make up a pixel are stored in separate planes. Planes are
located contiguously in memory.
I'm not sure to agree with this. You make it sounds like FB_TYPE_PLANES stores
each bit in a different plane. Is that really the case ?
- FB_TYPE_INTERLEAVED_PLANES
The bits that make up a pixel are stored in separate planes. Planes
are interleaved.
The interleave factor (the distance in bytes between the planes in
memory) is stored in the type_aux field.
That's a bit unclear to me. How are they interleaved ?
quoted
+- FB_VISUAL_MONO01
+
+Pixels are black or white and stored on one bit. A bit set to 1
represents a +black pixel and a bit set to 0 a white pixel. Pixels are
packed together in +bytes with 8 pixels per byte.
Actually we do have drivers that use 8 bits per pixel for a monochrome
visual. Hence:
"Pixels are black or white. A black pixel is represented by all
(typically one) bits set to ones, a white pixel by all bits set to zeroes."
OK. I've rephrased it as
"Pixels are black or white and stored on a number of bits (typically one)
specified by the variable screen information bpp field.
Black pixels are represented by all bits set to 1 and white pixels by all bits
set to 0. When the number of bits per pixel is smaller than 8, several pixels
are packed together in a byte."
quoted
+FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.
... so this may also not be true (but it is for all current drivers, IIRC).
There's a strict orthogonality between type (how is a pixel stored in
memory) and visual (how the bits that represent the pixel are interpreted
and converted to a color value).
What about
"FB_VISUAL_MONO01 is currently used with FB_TYPE_PACKED_PIXELS only." ?
Same comments for FB_VISUAL_MONO10
Fixed the same way.
quoted
+- FB_VISUAL_TRUECOLOR
+
+Pixels are broken into red, green and blue components, and each
component +indexes a read-only lookup table for the corresponding value.
Lookup tables +are device-dependent, and provide linear or non-linear
ramps.
+
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable screen
information red, green, blue and transp fields."
Storage format in memory is determined by the FB_TYPE_* value.
How so ? With FB_TYPE_PLANES and FB_VISUAL_TRUECOLOR for an RGB format, how
are the R, G and B planes ordered ? Are color components packed or padded
inside a plane ? I understand that the design goal was to have orthogonal
FB_TYPE_* and FB_VISUAL_* values, but we're missing too much information for
that to be truly generic.
quoted
+- FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR
+
+Pixel values are encoded as indices into a colormap that stores red,
green and +blue components. The colormap is read-only for
FB_VISUAL_STATIC_PSEUDOCOLOR +and read-write for FB_VISUAL_PSEUDOCOLOR.
+
+Each pixel value is stored in the number of bits reported by the
variable +screen information bits_per_pixel field. Pixels are contiguous
in memory.
Whether pixels are contiguous in memory or not is determined by the
FB_TYPE_* value.
How can they not be contiguous in memory ? Can you please give an example ?
quoted
+FB_VISUAL_PSEUDOCOLOR and FB_VISUAL_STATIC_PSEUDOCOLOR are used with
+FB_TYPE_PACKED_PIXELS only.
Not true. Several drivers use bit planes or interleaved bitplanes.
How does that work ?
quoted
+- FB_VISUAL_DIRECTCOLOR
+
+Pixels are broken into red, green and blue components, and each
component +indexes a programmable lookup table for the corresponding
value. +
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable screen
information red, green, blue and transp fields."
quoted
+- FB_VISUAL_FOURCC
+
+Pixels are stored in memory as described by the format FOURCC identifier
+stored in the variable screen information fourcc field.
... stored in memory and interpreted ...
quoted
+struct fb_var_screeninfo {
+ __u32 xres; /* visible resolution
*/ + __u32 yres;
+ __u32 xres_virtual; /* virtual resolution
*/ + __u32 yres_virtual;
+ __u32 xoffset; /* offset from virtual to visible
*/ + __u32 yoffset; /* resolution
*/ +
+ __u32 bits_per_pixel; /* guess what
*/ + union {
+ struct { /* Legacy format API
*/ + __u32 grayscale; /* != 0 Graylevels instead
of colors */ + /* bitfields in fb mem if true
color, else only */ + /* length is significant
*/ + struct fb_bitfield red;
+ struct fb_bitfield green;
+ struct fb_bitfield blue;
+ struct fb_bitfield transp; /* transparency
*/ + };
+ struct { /* FOURCC-based format API
*/ + __u32 fourcc; /* FOURCC format
*/ + __u32 colorspace;
+ __u32 reserved[11];
+ } format;
+ };
+
+ struct fb_bitfield red; /* bitfield in fb mem if true
color, */ + struct fb_bitfield green; /* else only length is
significant */ + struct fb_bitfield blue;
+ struct fb_bitfield transp; /* transparency
*/
Hi Laurent,
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart
[off-list ref] wrote:
On Monday 29 August 2011 10:13:07 Geert Uytterhoeven wrote:
quoted
On Fri, Aug 19, 2011 at 11:37, Laurent Pinchart wrote:
[snip]
quoted
quoted
+- FB_TYPE_PACKED_PIXELS
+
+Color components (usually RGB or YUV) are packed together into
macropixels +that are stored in a single plane. The exact color
components layout is +described in a visual-dependent way.
+
+Frame buffer visuals that don't use multiple color components per pixel
+(such as monochrome and pseudo-color visuals) are reported as packed
frame +buffer types, even though they don't stricly speaking pack color
components +into macropixels.
That's because the "packing" is not about the color components, but about
the bits that represent a single pixel.
I.e. the bits that make up the pixel (the macropixel) are stored next
to each other
in memory.
OK, I've modified that last sentence to read
"Frame buffer visuals that don't use multiple color components per pixel (such
as monochrome and pseudo-color visuals) are also reported as packed frame
buffer types, as the bits that make up individual pixels are packed next to
each other in memory."
Still not correct, as you don't grasp the concept of e.g. bitplanes
yet (see below).
The visual doesn't have anything to do with how the macropixels are stored in
memory.
quoted
quoted
+- FB_TYPE_PLANES
+
+Color components are stored in separate planes. Planes are located
+contiguously in memory.
The bits that make up a pixel are stored in separate planes. Planes are
located contiguously in memory.
I'm not sure to agree with this. You make it sounds like FB_TYPE_PLANES stores
each bit in a different plane. Is that really the case ?
- FB_TYPE_INTERLEAVED_PLANES
The bits that make up a pixel are stored in separate planes. Planes
are interleaved.
The interleave factor (the distance in bytes between the planes in
memory) is stored in the type_aux field.
That's a bit unclear to me. How are they interleaved ?
Instead of storing the same bits of all pixels contiguously into memory, they
are interleaved. Typically this is done per line (type_aux is the
length of a line in bytes),
or per word (Atari uses 2 bytes interleaving.
E.g. for a 320 x 200 display, with FB_TYPE_PLANES, you would store 320 x 200 64000 first bits, followed 64000 second bits, and so on.
With FB_TYPE_INTERLEAVED_PLANES and type_aux = 320 / 8 = 40,
you store the first line of the screen as 320 first bits, followed by
320 second bits, and so on.
Then the next line, as 320 first bits, followed by 320 second bits, and so on...
Hence the bits that make up a pixel are spread across memory. It was
useful in the days
computers couldn't show many colors, and allows things like 5 or 6
bits per pixels.
quoted
quoted
+- FB_VISUAL_MONO01
+
+Pixels are black or white and stored on one bit. A bit set to 1
represents a +black pixel and a bit set to 0 a white pixel. Pixels are
packed together in +bytes with 8 pixels per byte.
Actually we do have drivers that use 8 bits per pixel for a monochrome
visual. Hence:
"Pixels are black or white. A black pixel is represented by all
(typically one) bits set to ones, a white pixel by all bits set to zeroes."
OK. I've rephrased it as
"Pixels are black or white and stored on a number of bits (typically one)
specified by the variable screen information bpp field.
Black pixels are represented by all bits set to 1 and white pixels by all bits
set to 0. When the number of bits per pixel is smaller than 8, several pixels
are packed together in a byte."
OK.
quoted
quoted
+FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.
... so this may also not be true (but it is for all current drivers, IIRC).
There's a strict orthogonality between type (how is a pixel stored in
memory) and visual (how the bits that represent the pixel are interpreted
and converted to a color value).
What about
"FB_VISUAL_MONO01 is currently used with FB_TYPE_PACKED_PIXELS only." ?
stifb.c seems to use FB_TYPE_PLANES, but it uses bits_per_pixel is 1, so
FB_TYPE_PACKED_PIXELS, FB_TYPE_PLANES, and FB_TYPE_INTERLEAVED_PLANES
all degenerate to the same case anyway.
quoted
quoted
+- FB_VISUAL_TRUECOLOR
+
+Pixels are broken into red, green and blue components, and each
component +indexes a read-only lookup table for the corresponding value.
Lookup tables +are device-dependent, and provide linear or non-linear
ramps.
+
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable screen
information red, green, blue and transp fields."
Storage format in memory is determined by the FB_TYPE_* value.
How so ? With FB_TYPE_PLANES and FB_VISUAL_TRUECOLOR for an RGB format, how
are the R, G and B planes ordered ? Are color components packed or padded
That's specified by the fb_bitfield structs.
inside a plane ? I understand that the design goal was to have orthogonal
FB_TYPE_* and FB_VISUAL_* values, but we're missing too much information for
that to be truly generic.
The visual specifies how to interprete the fields that make up a pixel (as color
components, indices, ...).
The fb_bitfield structs specify how the fields are laid out in a pixel of size
bits_per_pixel.
The frame buffer type specifies how pixels are laid out in memory.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 11:36:07 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 10:13:07 Geert Uytterhoeven wrote:
quoted
On Fri, Aug 19, 2011 at 11:37, Laurent Pinchart wrote:
[snip]
quoted
quoted
+- FB_TYPE_PACKED_PIXELS
+
+Color components (usually RGB or YUV) are packed together into
macropixels +that are stored in a single plane. The exact color
components layout is +described in a visual-dependent way.
+
+Frame buffer visuals that don't use multiple color components per
pixel +(such as monochrome and pseudo-color visuals) are reported as
packed frame +buffer types, even though they don't stricly speaking
pack color components +into macropixels.
That's because the "packing" is not about the color components, but
about the bits that represent a single pixel.
I.e. the bits that make up the pixel (the macropixel) are stored next
to each other
in memory.
OK, I've modified that last sentence to read
"Frame buffer visuals that don't use multiple color components per pixel
(such as monochrome and pseudo-color visuals) are also reported as
packed frame buffer types, as the bits that make up individual pixels
are packed next to each other in memory."
Still not correct, as you don't grasp the concept of e.g. bitplanes
yet (see below).
The visual doesn't have anything to do with how the macropixels are stored
in memory.
quoted
quoted
quoted
+- FB_TYPE_PLANES
+
+Color components are stored in separate planes. Planes are located
+contiguously in memory.
The bits that make up a pixel are stored in separate planes. Planes are
located contiguously in memory.
I'm not sure to agree with this. You make it sounds like FB_TYPE_PLANES
stores each bit in a different plane. Is that really the case ?
Thank you. That's clearer now. Planar formats in V4L2 are different, hence my
initial confusion.
If my understanding is now correct, a V4L2 planar YUV type where Y, U and V
components are stored in separate byte-oriented planes, with each plane
storing Y, U or V components packed (such as http://linuxtv.org/downloads/v4l-
dvb-apis/V4L2-PIX-FMT-YUV422P.html), would be of neither FB_TYPE_PLANES nor
FB_TYPE_PACKED. The same would be true for an RGB format where each component
is stored in a separate plane with each plane sotring R, G or B packed.
If the above is correct, what FB_TYPE_* should a driver report when using
FB_VISUAL_FOURCC with V4L2_PIX_FMT_YUV422P (http://linuxtv.org/downloads/v4l-
dvb-apis/V4L2-PIX-FMT-YUV422P.html) or V4L2_PIX_FMT_NV12
(http://linuxtv.org/downloads/v4l-dvb-apis/re25.html) for instance ?
quoted
quoted
- FB_TYPE_INTERLEAVED_PLANES
The bits that make up a pixel are stored in separate planes. Planes
are interleaved.
The interleave factor (the distance in bytes between the planes in
memory) is stored in the type_aux field.
That's a bit unclear to me. How are they interleaved ?
Instead of storing the same bits of all pixels contiguously into memory,
they are interleaved. Typically this is done per line (type_aux is the
length of a line in bytes),
or per word (Atari uses 2 bytes interleaving.
E.g. for a 320 x 200 display, with FB_TYPE_PLANES, you would store 320 x
200 = 64000 first bits, followed 64000 second bits, and so on.
With FB_TYPE_INTERLEAVED_PLANES and type_aux = 320 / 8 = 40,
you store the first line of the screen as 320 first bits, followed by
320 second bits, and so on.
Then the next line, as 320 first bits, followed by 320 second bits, and so
on...
Hence the bits that make up a pixel are spread across memory. It was
useful in the days
computers couldn't show many colors, and allows things like 5 or 6
bits per pixels.
quoted
quoted
quoted
+- FB_VISUAL_MONO01
+
+Pixels are black or white and stored on one bit. A bit set to 1
represents a +black pixel and a bit set to 0 a white pixel. Pixels are
packed together in +bytes with 8 pixels per byte.
Actually we do have drivers that use 8 bits per pixel for a monochrome
visual. Hence:
"Pixels are black or white. A black pixel is represented by all
(typically one) bits set to ones, a white pixel by all bits set to
zeroes."
OK. I've rephrased it as
"Pixels are black or white and stored on a number of bits (typically one)
specified by the variable screen information bpp field.
Black pixels are represented by all bits set to 1 and white pixels by all
bits set to 0. When the number of bits per pixel is smaller than 8,
several pixels are packed together in a byte."
OK.
quoted
quoted
quoted
+FB_VISUAL_MONO01 is used with FB_TYPE_PACKED_PIXELS only.
... so this may also not be true (but it is for all current drivers,
IIRC). There's a strict orthogonality between type (how is a pixel
stored in memory) and visual (how the bits that represent the pixel are
interpreted and converted to a color value).
What about
"FB_VISUAL_MONO01 is currently used with FB_TYPE_PACKED_PIXELS only." ?
stifb.c seems to use FB_TYPE_PLANES, but it uses bits_per_pixel is 1, so
FB_TYPE_PACKED_PIXELS, FB_TYPE_PLANES, and FB_TYPE_INTERLEAVED_PLANES
all degenerate to the same case anyway.
quoted
quoted
quoted
+- FB_VISUAL_TRUECOLOR
+
+Pixels are broken into red, green and blue components, and each
component +indexes a read-only lookup table for the corresponding
value. Lookup tables +are device-dependent, and provide linear or
non-linear ramps.
+
+Each component is stored in memory according to the variable screen
+information red, green, blue and transp fields.
"Each component is stored in a macropixel according to the variable
screen information red, green, blue and transp fields."
Storage format in memory is determined by the FB_TYPE_* value.
How so ? With FB_TYPE_PLANES and FB_VISUAL_TRUECOLOR for an RGB format,
how are the R, G and B planes ordered ? Are color components packed or
padded
That's specified by the fb_bitfield structs.
quoted
inside a plane ? I understand that the design goal was to have orthogonal
FB_TYPE_* and FB_VISUAL_* values, but we're missing too much information
for that to be truly generic.
The visual specifies how to interprete the fields that make up a pixel (as
color components, indices, ...).
The fb_bitfield structs specify how the fields are laid out in a pixel of
size bits_per_pixel.
The frame buffer type specifies how pixels are laid out in memory.
That's fine if the FB device uses bitplanes, but not if it uses a format such
as the ones described above.
--
Regards,
Laurent Pinchart
Hi Laurent,
On Mon, Aug 29, 2011 at 12:09, Laurent Pinchart
[off-list ref] wrote:
On Monday 29 August 2011 11:36:07 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart wrote:
[...]
If my understanding is now correct, a V4L2 planar YUV type where Y, U and V
components are stored in separate byte-oriented planes, with each plane
storing Y, U or V components packed (such as http://linuxtv.org/downloads/v4l-
dvb-apis/V4L2-PIX-FMT-YUV422P.html), would be of neither FB_TYPE_PLANES nor
FB_TYPE_PACKED. The same would be true for an RGB format where each component
is stored in a separate plane with each plane sotring R, G or B packed.
We need new types for those. Or always use FOURCC for them.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 13:04:15 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 12:09, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 11:36:07 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart wrote:
[...]
quoted
If my understanding is now correct, a V4L2 planar YUV type where Y, U and
V components are stored in separate byte-oriented planes, with each
plane storing Y, U or V components packed (such as
http://linuxtv.org/downloads/v4l- dvb-apis/V4L2-PIX-FMT-YUV422P.html),
would be of neither FB_TYPE_PLANES nor FB_TYPE_PACKED. The same would be
true for an RGB format where each component is stored in a separate
plane with each plane sotring R, G or B packed.
On Mon, Aug 29, 2011 at 13:08, Laurent Pinchart
[off-list ref] wrote:
On Monday 29 August 2011 13:04:15 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 12:09, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 11:36:07 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart wrote:
[...]
quoted
If my understanding is now correct, a V4L2 planar YUV type where Y, U and
V components are stored in separate byte-oriented planes, with each
plane storing Y, U or V components packed (such as
http://linuxtv.org/downloads/v4l- dvb-apis/V4L2-PIX-FMT-YUV422P.html),
would be of neither FB_TYPE_PLANES nor FB_TYPE_PACKED. The same would be
true for an RGB format where each component is stored in a separate
plane with each plane sotring R, G or B packed.
We need new types for those. Or always use FOURCC for them.
My proposal currently defined FB_VISUAL_FOURCC. What about adding
FB_TYPE_FOURCC as well ?
That may make sense.
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy way,
- But what with modes that can be represented? Legacy software cannot
handle FB_{TYPE,VISUAL}_FOURCC.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 13:20:44 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 13:08, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 13:04:15 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 12:09, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 11:36:07 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 10:50, Laurent Pinchart wrote:
[...]
quoted
If my understanding is now correct, a V4L2 planar YUV type where Y, U
and V components are stored in separate byte-oriented planes, with
each plane storing Y, U or V components packed (such as
http://linuxtv.org/downloads/v4l- dvb-apis/V4L2-PIX-FMT-YUV422P.html),
would be of neither FB_TYPE_PLANES nor FB_TYPE_PACKED. The same would
be true for an RGB format where each component is stored in a
separate plane with each plane sotring R, G or B packed.
We need new types for those. Or always use FOURCC for them.
My proposal currently defined FB_VISUAL_FOURCC. What about adding
FB_TYPE_FOURCC as well ?
That may make sense.
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy way,
Definitely.
- But what with modes that can be represented? Legacy software cannot
handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is configured
using the FOURCC API. If FBIOPUT_VSCREENINFO is called with a non-FOURCC
format, the driver will report non-FOURCC types and visuals.
--
Regards,
Laurent Pinchart
Hi Laurent,
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart
[off-list ref] wrote:
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software cannot
handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is configured
using the FOURCC API. If FBIOPUT_VSCREENINFO is called with a non-FOURCC
format, the driver will report non-FOURCC types and visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
- Is there an easy way to force FOURCC reporting, so new apps don't have to
support parsing the legacy formats? This is useful for new apps that want to
support (a subset of) FOURCC modes only.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software cannot
handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called with a
non-FOURCC format, the driver will report non-FOURCC types and visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We have no
per-open fixed and variable screen info, and FB devices can be opened by
multiple applications at the same time.
- Is there an easy way to force FOURCC reporting, so new apps don't have
to support parsing the legacy formats? This is useful for new apps that
want to support (a subset of) FOURCC modes only.
Hi Laurent,
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart
[off-list ref] wrote:
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
 - When using a mode that cannot be represented in the legacy way,
Definitely.
quoted
 - But what with modes that can be represented? Legacy software cannot
  handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called with a
non-FOURCC format, the driver will report non-FOURCC types and visuals.
Hmm, two use cases:
 - The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
  Later the user runs a legacy application.
   => Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We have no
per-open fixed and variable screen info, and FB devices can be opened by
multiple applications at the same time.
quoted
 - Is there an easy way to force FOURCC reporting, so new apps don't have
to support parsing the legacy formats? This is useful for new apps that
want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Gr{oetje,eeting}s,
            Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                 -- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 16:14:38 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software
cannot handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called with
a non-FOURCC format, the driver will report non-FOURCC types and
visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We have
no per-open fixed and variable screen info, and FB devices can be opened
by multiple applications at the same time.
quoted
- Is there an easy way to force FOURCC reporting, so new apps don't
have to support parsing the legacy formats? This is useful for new apps
that want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Do you mean new ioctls to replace the FOURCC API proposal, or new ioctls for
the above two operations ?
--
Regards,
Laurent Pinchart
Hi Laurent,
On Mon, Aug 29, 2011 at 16:17, Laurent Pinchart
[off-list ref] wrote:
On Monday 29 August 2011 16:14:38 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
 - When using a mode that cannot be represented in the legacy way,
Definitely.
quoted
 - But what with modes that can be represented? Legacy software
cannot handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called with
a non-FOURCC format, the driver will report non-FOURCC types and
visuals.
Hmm, two use cases:
 - The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
  Later the user runs a legacy application.
   => Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We have
no per-open fixed and variable screen info, and FB devices can be opened
by multiple applications at the same time.
quoted
 - Is there an easy way to force FOURCC reporting, so new apps don't
have to support parsing the legacy formats? This is useful for new apps
that want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Do you mean new ioctls to replace the FOURCC API proposal, or new ioctls for
the above two operations ?
New ioctls to replace the FOURCC proposal.
Gr{oetje,eeting}s,
            Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                 -- Linus Torvalds
Hi Geert,
On Monday 29 August 2011 16:26:02 Geert Uytterhoeven wrote:
On Mon, Aug 29, 2011 at 16:17, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 16:14:38 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy
way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software
cannot handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called
with a non-FOURCC format, the driver will report non-FOURCC types
and visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We
have no per-open fixed and variable screen info, and FB devices can
be opened by multiple applications at the same time.
quoted
- Is there an easy way to force FOURCC reporting, so new apps don't
have to support parsing the legacy formats? This is useful for new
apps that want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Do you mean new ioctls to replace the FOURCC API proposal, or new ioctls
for the above two operations ?
New ioctls to replace the FOURCC proposal.
*sigh*...
I'd like other people's opinion on this before throwing everything away.
Florian, Magnus, Guennadi, others, what do you think ?
--
Regards,
Laurent Pinchart
Hi,
On 08/29/2011 02:32 PM, Laurent Pinchart wrote:
Hi Geert,
On Monday 29 August 2011 16:26:02 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 16:17, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 16:14:38 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy
way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software
cannot handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called
with a non-FOURCC format, the driver will report non-FOURCC types
and visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We
have no per-open fixed and variable screen info, and FB devices can
be opened by multiple applications at the same time.
quoted
- Is there an easy way to force FOURCC reporting, so new apps don't
have to support parsing the legacy formats? This is useful for new
apps that want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Do you mean new ioctls to replace the FOURCC API proposal, or new ioctls
for the above two operations ?
New ioctls to replace the FOURCC proposal.
*sigh*...
I'd like other people's opinion on this before throwing everything away.
Florian, Magnus, Guennadi, others, what do you think ?
So, your issue is that some formats can be represented in the new and the old way?
There are 2 simpler solutions I can think of:
(1) ignore it, just do it the way Laurent proposed. I understand that someone
might feel uneasy about applications that are trapped because they don't know
the new format but could work with the old one. But I think this is not a big
issue as many applications will just try to set their own mode. For those that
doesn't and rely on the previous mode that is set by fbset or similar, we could
change fbset to prefer the old format if available. But even if we don't do
this, I don't have a problem with a program failing because it sees an
unsuitable mode even if it supports the legacy mode. It's not a regression and
can be easily fixed in userspace.
(2) forbid it, just allow drivers to implement FOURCC for formats that cannot be
represented in the old scheme. This is my preferred solution if anyone has
problems with (1).
I don't see how IOCTLs would help here. The pixel format just belongs into var
and fix so it has to be represented there anyway and thus set through it. We
could do an IOCTL that always returns the FOURCC active at the moment, if such a
FOURCC exists, and always use the legacy API for representing it in var/fix, if
it exists. But as I see this is not what you thought about so please explain
what your IOCTLS would look like and how they would solve the problem.
And I don't think a in-kernel translation layer is a good idea. Yes, it sounds
interesting, but it's tricky and the result will be that the driver and
userspace will permanently see different var and fix structures. Seriously I
think changing every framebuffer driver out there would be easier and much saner
than trying to implement such a thing.
Best regards,
Florian Tobias Schandinat
Hi Florian,
On Monday 29 August 2011 18:41:03 Florian Tobias Schandinat wrote:
On 08/29/2011 02:32 PM, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 16:26:02 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 16:17, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 16:14:38 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 15:34, Laurent Pinchart wrote:
quoted
On Monday 29 August 2011 15:09:04 Geert Uytterhoeven wrote:
quoted
On Mon, Aug 29, 2011 at 14:55, Laurent Pinchart wrote:
quoted
quoted
When will the driver report FB_{TYPE,VISUAL}_FOURCC?
- When using a mode that cannot be represented in the legacy
way,
Definitely.
quoted
- But what with modes that can be represented? Legacy software
cannot handle FB_{TYPE,VISUAL}_FOURCC.
My idea was to use FB_{TYPE,VISUAL}_FOURCC only when the mode is
configured using the FOURCC API. If FBIOPUT_VSCREENINFO is called
with a non-FOURCC format, the driver will report non-FOURCC types
and visuals.
Hmm, two use cases:
- The video mode is configured using a FOURCC-aware tool ("fbset
on
steroids").
Later the user runs a legacy application.
=> Do not retain FOURCC across opening of /dev/fb*.
I know about that problem, but it's not that easy to work around. We
have no per-open fixed and variable screen info, and FB devices can
be opened by multiple applications at the same time.
quoted
- Is there an easy way to force FOURCC reporting, so new apps
don't
have to support parsing the legacy formats? This is useful for new
apps that want to support (a subset of) FOURCC modes only.
Not at the moment.
So perhaps we do need new ioctls instead...
That would also ease an in-kernel translation layer.
Do you mean new ioctls to replace the FOURCC API proposal, or new
ioctls for the above two operations ?
New ioctls to replace the FOURCC proposal.
*sigh*...
I'd like other people's opinion on this before throwing everything away.
Florian, Magnus, Guennadi, others, what do you think ?
So, your issue is that some formats can be represented in the new and the
old way? There are 2 simpler solutions I can think of:
(1) ignore it, just do it the way Laurent proposed. I understand that
someone might feel uneasy about applications that are trapped because they
don't know the new format but could work with the old one. But I think
this is not a big issue as many applications will just try to set their
own mode. For those that doesn't and rely on the previous mode that is set
by fbset or similar, we could change fbset to prefer the old format if
available. But even if we don't do this, I don't have a problem with a
program failing because it sees an unsuitable mode even if it supports the
legacy mode. It's not a regression and can be easily fixed in userspace.
I agree with you here. Issues should be fixed in userspace. I don't expect
many issues in practice, as fbdev is often used on systems where userspace
components are developed to work with each other (such as embedded systems).
If such a system is upgraded to use YUV support, all components will likely be
tested and upgraded.
(2) forbid it, just allow drivers to implement FOURCC for formats that
cannot be represented in the old scheme. This is my preferred solution if
anyone has problems with (1).
I'm tempted by that solution. Not that I have an issue with (1), but I'm
wondering if we should try to use the FOURCC API for all formats, or just for
formats that can't be represented by the current API. We currently have no
bitplane FOURCCs, and I'm not sure if it would be worth adding them.
BTW, do recent hardware still support planar (in the bitplane sense) frame
buffers ? If so, is that used in practice on recent devices ? If the answer is
no, (1) might be the best solution, with bit-planar formats supported by the
current API only, and all other formats supported by the FOURCC API.
I don't see how IOCTLs would help here. The pixel format just belongs into
var and fix so it has to be represented there anyway and thus set through
it. We could do an IOCTL that always returns the FOURCC active at the
moment, if such a FOURCC exists, and always use the legacy API for
representing it in var/fix, if it exists. But as I see this is not what
you thought about so please explain what your IOCTLS would look like and
how they would solve the problem.
And I don't think a in-kernel translation layer is a good idea. Yes, it
sounds interesting, but it's tricky and the result will be that the driver
and userspace will permanently see different var and fix structures.
Seriously I think changing every framebuffer driver out there would be
easier and much saner than trying to implement such a thing.