Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | The Frame Buffer Device |
| 2 | ----------------------- |
| 3 | |
| 4 | Maintained by Geert Uytterhoeven <geert@linux-m68k.org> |
| 5 | Last revised: May 10, 2001 |
| 6 | |
| 7 | |
| 8 | 0. Introduction |
| 9 | --------------- |
| 10 | |
| 11 | The frame buffer device provides an abstraction for the graphics hardware. It |
| 12 | represents the frame buffer of some video hardware and allows application |
| 13 | software to access the graphics hardware through a well-defined interface, so |
| 14 | the software doesn't need to know anything about the low-level (hardware |
| 15 | register) stuff. |
| 16 | |
| 17 | The device is accessed through special device nodes, usually located in the |
| 18 | /dev directory, i.e. /dev/fb*. |
| 19 | |
| 20 | |
| 21 | 1. User's View of /dev/fb* |
| 22 | -------------------------- |
| 23 | |
| 24 | From the user's point of view, the frame buffer device looks just like any |
| 25 | other device in /dev. It's a character device using major 29; the minor |
| 26 | specifies the frame buffer number. |
| 27 | |
| 28 | By convention, the following device nodes are used (numbers indicate the device |
| 29 | minor numbers): |
| 30 | |
| 31 | 0 = /dev/fb0 First frame buffer |
| 32 | 1 = /dev/fb1 Second frame buffer |
| 33 | ... |
| 34 | 31 = /dev/fb31 32nd frame buffer |
| 35 | |
| 36 | For backwards compatibility, you may want to create the following symbolic |
| 37 | links: |
| 38 | |
| 39 | /dev/fb0current -> fb0 |
| 40 | /dev/fb1current -> fb1 |
| 41 | |
| 42 | and so on... |
| 43 | |
| 44 | The frame buffer devices are also `normal' memory devices, this means, you can |
| 45 | read and write their contents. You can, for example, make a screen snapshot by |
| 46 | |
| 47 | cp /dev/fb0 myfile |
| 48 | |
| 49 | There also can be more than one frame buffer at a time, e.g. if you have a |
| 50 | graphics card in addition to the built-in hardware. The corresponding frame |
| 51 | buffer devices (/dev/fb0 and /dev/fb1 etc.) work independently. |
| 52 | |
| 53 | Application software that uses the frame buffer device (e.g. the X server) will |
| 54 | use /dev/fb0 by default (older software uses /dev/fb0current). You can specify |
| 55 | an alternative frame buffer device by setting the environment variable |
| 56 | $FRAMEBUFFER to the path name of a frame buffer device, e.g. (for sh/bash |
| 57 | users): |
| 58 | |
| 59 | export FRAMEBUFFER=/dev/fb1 |
| 60 | |
| 61 | or (for csh users): |
| 62 | |
| 63 | setenv FRAMEBUFFER /dev/fb1 |
| 64 | |
| 65 | After this the X server will use the second frame buffer. |
| 66 | |
| 67 | |
| 68 | 2. Programmer's View of /dev/fb* |
| 69 | -------------------------------- |
| 70 | |
| 71 | As you already know, a frame buffer device is a memory device like /dev/mem and |
| 72 | it has the same features. You can read it, write it, seek to some location in |
| 73 | it and mmap() it (the main usage). The difference is just that the memory that |
| 74 | appears in the special file is not the whole memory, but the frame buffer of |
| 75 | some video hardware. |
| 76 | |
| 77 | /dev/fb* also allows several ioctls on it, by which lots of information about |
| 78 | the hardware can be queried and set. The color map handling works via ioctls, |
| 79 | too. Look into <linux/fb.h> for more information on what ioctls exist and on |
| 80 | which data structures they work. Here's just a brief overview: |
| 81 | |
| 82 | - You can request unchangeable information about the hardware, like name, |
| 83 | organization of the screen memory (planes, packed pixels, ...) and address |
| 84 | and length of the screen memory. |
| 85 | |
| 86 | - You can request and change variable information about the hardware, like |
| 87 | visible and virtual geometry, depth, color map format, timing, and so on. |
| 88 | If you try to change that information, the driver maybe will round up some |
| 89 | values to meet the hardware's capabilities (or return EINVAL if that isn't |
| 90 | possible). |
| 91 | |
| 92 | - You can get and set parts of the color map. Communication is done with 16 |
| 93 | bits per color part (red, green, blue, transparency) to support all |
| 94 | existing hardware. The driver does all the computations needed to apply |
| 95 | it to the hardware (round it down to less bits, maybe throw away |
| 96 | transparency). |
| 97 | |
| 98 | All this hardware abstraction makes the implementation of application programs |
| 99 | easier and more portable. E.g. the X server works completely on /dev/fb* and |
| 100 | thus doesn't need to know, for example, how the color registers of the concrete |
| 101 | hardware are organized. XF68_FBDev is a general X server for bitmapped, |
| 102 | unaccelerated video hardware. The only thing that has to be built into |
| 103 | application programs is the screen organization (bitplanes or chunky pixels |
| 104 | etc.), because it works on the frame buffer image data directly. |
| 105 | |
| 106 | For the future it is planned that frame buffer drivers for graphics cards and |
| 107 | the like can be implemented as kernel modules that are loaded at runtime. Such |
| 108 | a driver just has to call register_framebuffer() and supply some functions. |
| 109 | Writing and distributing such drivers independently from the kernel will save |
| 110 | much trouble... |
| 111 | |
| 112 | |
| 113 | 3. Frame Buffer Resolution Maintenance |
| 114 | -------------------------------------- |
| 115 | |
| 116 | Frame buffer resolutions are maintained using the utility `fbset'. It can |
| 117 | change the video mode properties of a frame buffer device. Its main usage is |
| 118 | to change the current video mode, e.g. during boot up in one of your /etc/rc.* |
| 119 | or /etc/init.d/* files. |
| 120 | |
| 121 | Fbset uses a video mode database stored in a configuration file, so you can |
| 122 | easily add your own modes and refer to them with a simple identifier. |
| 123 | |
| 124 | |
| 125 | 4. The X Server |
| 126 | --------------- |
| 127 | |
| 128 | The X server (XF68_FBDev) is the most notable application program for the frame |
| 129 | buffer device. Starting with XFree86 release 3.2, the X server is part of |
| 130 | XFree86 and has 2 modes: |
| 131 | |
| 132 | - If the `Display' subsection for the `fbdev' driver in the /etc/XF86Config |
| 133 | file contains a |
| 134 | |
| 135 | Modes "default" |
| 136 | |
| 137 | line, the X server will use the scheme discussed above, i.e. it will start |
| 138 | up in the resolution determined by /dev/fb0 (or $FRAMEBUFFER, if set). You |
| 139 | still have to specify the color depth (using the Depth keyword) and virtual |
| 140 | resolution (using the Virtual keyword) though. This is the default for the |
| 141 | configuration file supplied with XFree86. It's the most simple |
| 142 | configuration, but it has some limitations. |
| 143 | |
| 144 | - Therefore it's also possible to specify resolutions in the /etc/XF86Config |
| 145 | file. This allows for on-the-fly resolution switching while retaining the |
| 146 | same virtual desktop size. The frame buffer device that's used is still |
| 147 | /dev/fb0current (or $FRAMEBUFFER), but the available resolutions are |
| 148 | defined by /etc/XF86Config now. The disadvantage is that you have to |
| 149 | specify the timings in a different format (but `fbset -x' may help). |
| 150 | |
| 151 | To tune a video mode, you can use fbset or xvidtune. Note that xvidtune doesn't |
| 152 | work 100% with XF68_FBDev: the reported clock values are always incorrect. |
| 153 | |
| 154 | |
| 155 | 5. Video Mode Timings |
| 156 | --------------------- |
| 157 | |
| 158 | A monitor draws an image on the screen by using an electron beam (3 electron |
| 159 | beams for color models, 1 electron beam for monochrome monitors). The front of |
| 160 | the screen is covered by a pattern of colored phosphors (pixels). If a phosphor |
| 161 | is hit by an electron, it emits a photon and thus becomes visible. |
| 162 | |
| 163 | The electron beam draws horizontal lines (scanlines) from left to right, and |
| 164 | from the top to the bottom of the screen. By modifying the intensity of the |
| 165 | electron beam, pixels with various colors and intensities can be shown. |
| 166 | |
| 167 | After each scanline the electron beam has to move back to the left side of the |
| 168 | screen and to the next line: this is called the horizontal retrace. After the |
| 169 | whole screen (frame) was painted, the beam moves back to the upper left corner: |
| 170 | this is called the vertical retrace. During both the horizontal and vertical |
| 171 | retrace, the electron beam is turned off (blanked). |
| 172 | |
| 173 | The speed at which the electron beam paints the pixels is determined by the |
| 174 | dotclock in the graphics board. For a dotclock of e.g. 28.37516 MHz (millions |
| 175 | of cycles per second), each pixel is 35242 ps (picoseconds) long: |
| 176 | |
| 177 | 1/(28.37516E6 Hz) = 35.242E-9 s |
| 178 | |
| 179 | If the screen resolution is 640x480, it will take |
| 180 | |
| 181 | 640*35.242E-9 s = 22.555E-6 s |
| 182 | |
| 183 | to paint the 640 (xres) pixels on one scanline. But the horizontal retrace |
| 184 | also takes time (e.g. 272 `pixels'), so a full scanline takes |
| 185 | |
| 186 | (640+272)*35.242E-9 s = 32.141E-6 s |
| 187 | |
| 188 | We'll say that the horizontal scanrate is about 31 kHz: |
| 189 | |
| 190 | 1/(32.141E-6 s) = 31.113E3 Hz |
| 191 | |
| 192 | A full screen counts 480 (yres) lines, but we have to consider the vertical |
| 193 | retrace too (e.g. 49 `lines'). So a full screen will take |
| 194 | |
| 195 | (480+49)*32.141E-6 s = 17.002E-3 s |
| 196 | |
| 197 | The vertical scanrate is about 59 Hz: |
| 198 | |
| 199 | 1/(17.002E-3 s) = 58.815 Hz |
| 200 | |
| 201 | This means the screen data is refreshed about 59 times per second. To have a |
| 202 | stable picture without visible flicker, VESA recommends a vertical scanrate of |
| 203 | at least 72 Hz. But the perceived flicker is very human dependent: some people |
| 204 | can use 50 Hz without any trouble, while I'll notice if it's less than 80 Hz. |
| 205 | |
| 206 | Since the monitor doesn't know when a new scanline starts, the graphics board |
| 207 | will supply a synchronization pulse (horizontal sync or hsync) for each |
| 208 | scanline. Similarly it supplies a synchronization pulse (vertical sync or |
| 209 | vsync) for each new frame. The position of the image on the screen is |
| 210 | influenced by the moments at which the synchronization pulses occur. |
| 211 | |
| 212 | The following picture summarizes all timings. The horizontal retrace time is |
| 213 | the sum of the left margin, the right margin and the hsync length, while the |
| 214 | vertical retrace time is the sum of the upper margin, the lower margin and the |
| 215 | vsync length. |
| 216 | |
| 217 | +----------+---------------------------------------------+----------+-------+ |
| 218 | | | ^ | | | |
| 219 | | | |upper_margin | | | |
| 220 | | | ¥ | | | |
| 221 | +----------###############################################----------+-------+ |
| 222 | | # ^ # | | |
| 223 | | # | # | | |
| 224 | | # | # | | |
| 225 | | # | # | | |
| 226 | | left # | # right | hsync | |
| 227 | | margin # | xres # margin | len | |
| 228 | |<-------->#<---------------+--------------------------->#<-------->|<----->| |
| 229 | | # | # | | |
| 230 | | # | # | | |
| 231 | | # | # | | |
| 232 | | # |yres # | | |
| 233 | | # | # | | |
| 234 | | # | # | | |
| 235 | | # | # | | |
| 236 | | # | # | | |
| 237 | | # | # | | |
| 238 | | # | # | | |
| 239 | | # | # | | |
| 240 | | # | # | | |
| 241 | | # ¥ # | | |
| 242 | +----------###############################################----------+-------+ |
| 243 | | | ^ | | | |
| 244 | | | |lower_margin | | | |
| 245 | | | ¥ | | | |
| 246 | +----------+---------------------------------------------+----------+-------+ |
| 247 | | | ^ | | | |
| 248 | | | |vsync_len | | | |
| 249 | | | ¥ | | | |
| 250 | +----------+---------------------------------------------+----------+-------+ |
| 251 | |
| 252 | The frame buffer device expects all horizontal timings in number of dotclocks |
| 253 | (in picoseconds, 1E-12 s), and vertical timings in number of scanlines. |
| 254 | |
| 255 | |
| 256 | 6. Converting XFree86 timing values info frame buffer device timings |
| 257 | -------------------------------------------------------------------- |
| 258 | |
| 259 | An XFree86 mode line consists of the following fields: |
| 260 | "800x600" 50 800 856 976 1040 600 637 643 666 |
| 261 | < name > DCF HR SH1 SH2 HFL VR SV1 SV2 VFL |
| 262 | |
| 263 | The frame buffer device uses the following fields: |
| 264 | |
| 265 | - pixclock: pixel clock in ps (pico seconds) |
| 266 | - left_margin: time from sync to picture |
| 267 | - right_margin: time from picture to sync |
| 268 | - upper_margin: time from sync to picture |
| 269 | - lower_margin: time from picture to sync |
| 270 | - hsync_len: length of horizontal sync |
| 271 | - vsync_len: length of vertical sync |
| 272 | |
| 273 | 1) Pixelclock: |
| 274 | xfree: in MHz |
| 275 | fb: in picoseconds (ps) |
| 276 | |
| 277 | pixclock = 1000000 / DCF |
| 278 | |
| 279 | 2) horizontal timings: |
| 280 | left_margin = HFL - SH2 |
| 281 | right_margin = SH1 - HR |
| 282 | hsync_len = SH2 - SH1 |
| 283 | |
| 284 | 3) vertical timings: |
| 285 | upper_margin = VFL - SV2 |
| 286 | lower_margin = SV1 - VR |
| 287 | vsync_len = SV2 - SV1 |
| 288 | |
| 289 | Good examples for VESA timings can be found in the XFree86 source tree, |
| 290 | under "xc/programs/Xserver/hw/xfree86/doc/modeDB.txt". |
| 291 | |
| 292 | |
| 293 | 7. References |
| 294 | ------------- |
| 295 | |
| 296 | For more specific information about the frame buffer device and its |
| 297 | applications, please refer to the Linux-fbdev website: |
| 298 | |
| 299 | http://linux-fbdev.sourceforge.net/ |
| 300 | |
| 301 | and to the following documentation: |
| 302 | |
| 303 | - The manual pages for fbset: fbset(8), fb.modes(5) |
| 304 | - The manual pages for XFree86: XF68_FBDev(1), XF86Config(4/5) |
| 305 | - The mighty kernel sources: |
| 306 | o linux/drivers/video/ |
| 307 | o linux/include/linux/fb.h |
| 308 | o linux/include/video/ |
| 309 | |
| 310 | |
| 311 | |
| 312 | 8. Mailing list |
| 313 | --------------- |
| 314 | |
| 315 | There are several frame buffer device related mailing lists at SourceForge: |
| 316 | - linux-fbdev-announce@lists.sourceforge.net, for announcements, |
| 317 | - linux-fbdev-user@lists.sourceforge.net, for generic user support, |
| 318 | - linux-fbdev-devel@lists.sourceforge.net, for project developers. |
| 319 | |
| 320 | Point your web browser to http://sourceforge.net/projects/linux-fbdev/ for |
| 321 | subscription information and archive browsing. |
| 322 | |
| 323 | |
| 324 | 9. Downloading |
| 325 | -------------- |
| 326 | |
| 327 | All necessary files can be found at |
| 328 | |
| 329 | ftp://ftp.uni-erlangen.de/pub/Linux/LOCAL/680x0/ |
| 330 | |
| 331 | and on its mirrors. |
| 332 | |
| 333 | The latest version of fbset can be found at |
| 334 | |
| 335 | http://home.tvd.be/cr26864/Linux/fbdev/ |
| 336 | |
| 337 | |
| 338 | 10. Credits |
| 339 | ---------- |
| 340 | |
| 341 | This readme was written by Geert Uytterhoeven, partly based on the original |
| 342 | `X-framebuffer.README' by Roman Hodek and Martin Schaller. Section 6 was |
| 343 | provided by Frank Neumann. |
| 344 | |
| 345 | The frame buffer device abstraction was designed by Martin Schaller. |