[懷舊] 聊聊 svgalib

the 1st edition: 20121024 (3)
the 2nd edition: 20190402 (3)

svgalib 是我想在 linux console 裡頭使用繪圖模式時候找到的 library。在很久很久以前 X 還沒有現在這麼好用, 我有很長一段時間是在 linux console 下作業 - 主要是寫程式。而想要練習在圖型介面下的程式, 類似 turbo c 的 bgi 或是 bios call 0x10 就會需要 svgalib。

那時候找了一些 linux console graphic library, ggi, Allegro, 當然還有 linux framebuffer, 不過最後我還是用了 svgalib, 至於為什麼不用相容性更好的 linux framebuffer? 因為我想要有文字模式, 我只想要在我需要的時候才切換到繪圖模式, linux framebuffer 不知道為什麼一定要在開機就決定, 不能在需要的時候才載入? 有什麼技術上的問題嗎？我似乎搞錯了, 有兩種模式, 開機時指定 vga 參數, 以及開機後手動載入 fb module, 不過一旦載入 fb module 就無法移除了。但是在 arm/linux 上, 就只能選擇 linux framebuffer了。

jmcce 中文終端機就是用這個寫的 (也支援 linux framebuffer), 我也曾經一度很迷中文終端機的程式, 不過在 X 環境愈來愈好用的現在, 已經沒這麼迫切的需求了。

svgalib 似乎沒在維護了, 在新的顯示卡上, 無法使用最新功能, 不過沒關係, 只要使用 vga 或是 vesa driver 就可以有基本的顯示效果, 而這也應該是所有顯示卡都會支援的模式。通常我比較需要 640X480X24bit color。

svgalib 有 2 個版本
1.4.3
1.9.25
https://mirrors.slackware.com/slackware/slackware-current/source/l/svgalib/svgalib-1.9.25.tar.bz2.mirrorlist
https://ftp.yzu.edu.tw/Linux/Slackware/slackware-current/source/l/svgalib/svgalib-1.9.25.tar.bz2

1.9.25 一直都是測試版本, 需要編譯出 kernel module, 比較麻煩, 所以一般都用 1.4.3, 我也只有很少的次數成功編譯出 1.9.25 的 kernel module。

安裝 svgalib

apt-get install libsvga1-dev

20150724 補充

deb http://ftp.tw.debian.org/debian wheezy main contrib non-free
現在需要使用 wheezy 才能抓到 svgalib

20190402 補充

目前似乎只有 debian Jessie 有保留 svgalib-1.4.3 package

我使用了 debian 的 patch, 並做一些適度的修改, 可以在 linux 4.16 環境上編譯, 32bit/64bit 都可以編譯。

github source: https://github.com/descent/svgalib-1.4.3
branch 32bit 可以編譯出 32bit 版本

debian patch downloads

File	Size	SHA-256 Checksum
svgalib_1.4.3-33.dsc	1.3 KiB	59aeddf6d516a5c9efc2180133883bdbc2148f662132d572c5bac6f3ac311ba0
svgalib_1.4.3.orig.tar.gz	787.8 KiB	3f2b962a63e140435f8ad575ed41c7e3e84e2de1150690d9f24a67cce5c96db0
svgalib_1.4.3-33.debian.tar.gz	64.0 KiB	c988bda8563643cefc55f747c31580d1f11e4afa87a08b576cff23dad6b1de07

Available diffs

• diff from 1:1.4.3-31 (in Ubuntu) to 1:1.4.3-33 (7.0 KiB)

No changes file available.

debian patch
for file in ../debian/patches/*.patch; do patch -p1 < $file; done /* 循环打补丁，

apt-get install libx86-dev

編譯 svgalib
linux控制台下的svgalib绘图函数库的使用

這個東西已經過時了, 在 uefi pc 上, 似乎無法有任何的發揮, 我在 uefi pc 上, 雖然可以執行 vgatest 但程式無法正常切換顯示模式。應該只能在 legacy bios pc 上使用。

[eeepc901/linux 測試]

先將 /lib/modules/4.16.0-1-amd64/kernel/drivers/gpu/drm/i915/i915.ko 改名, 重開機之後, 就不會載入 i915.ko。

apt-get install v86d
modprobe uvesafb

這樣就可以讓 svgalib 使用 vesa driver。 v86d , uvesafb (linux-4.4.110/drivers/video/fbdev/uvesafb.c) 這 2 個軟體很神奇, 需要相互搭配。 v86d provides a backend for kernel drivers that need to execute x86 BIOS code. uvesafb 則是呼叫 v86d 來執行 x86 Video BIOS code。

libx86 則是另外一個神奇的軟體, make real-mode x86 BIOS calls from userland, 在 linux 可以執行 real-mode x86 BIOS 呼叫。

【問題】UEFI? BIOS? Legacy? 淺談主機板UEFI觀念與迷思

GOP與VBIOS:
在UEFI與BIOS介面中，如何顯示出圖形介面的呢?
即是靠GOP與VBIOS。
簡單的說，純UEFI介面(無CSM模擬BIOS的UEFI)，是由顯示卡(或內顯)的GOP來產生的。
而傳統BIOS就是用VBIOS產生的，所以，你的顯示卡如果較新，就會有GOP功能，才能以純UEFI啟動(關掉CSM)，那如果你是較舊的顯示卡，則會以CSM模擬BIOS再以顯示卡的VBIOS啟動。

那麼哪些顯卡支援GOP呢?
1.AMD：只要是HD 7000系列以後的顯示卡，7850,7950,7970等等。
2.Nvidia：只要是GeForce 600系列以後，660, 660Ti, 670, 680…等等。
但是，就算你是AMD的7系列或者NVIDIA的6系列，有可能還是不支援UEFI GOP!
這時則可以透過後天刷新顯示卡韌體刷成GOP。

ref:
VGA support
[Windows 8 | UEFI]_UEFI與Windows 8的親密整合番外篇-顯示卡UEFI GOP化
BIOS and Firmware

支援 uefi 的顯示卡, 已經不會有 video bios 了, 綜合上述, 我才會說在 uefi pc 上, svgalib 不能用了。

[編輯 /etc/vga/libvga.config]

/etc/vga/libvga.config

 # Configuration file for svgalib. Default location is /etc/vga.
 # Other config file locations: ~/.svgalibrc
 #     where SVGALIB_CONFIG_FILE points
 # Lines starting with '#' are ignored.
 
 # If you have two vga cards with the same pci vendor id, svgalib will try
 # to use the first one, even if the second one is active. In that case,
 # use PCIStart to force starting the search for a vga card only at a
 # specific bus and device numbers. For example, an AGP card is usually on
 # bus 1, while pci is on bus 0, so to use the AGP card, uncomment:
 
 # PCIStart 1 0
 
 # Have a deep look at README.config to see what can do here (especially
 # for mach32).
 
 # Mouse type:
 
 # mouse Microsoft # Microsoft
 # mouse MouseSystems # Mouse Systems
 # mouse MMSeries # Logitech MM Series
 # mouse Logitech # Logitech protocol (old, newer mice use Microsoft protocol)
 # mouse Busmouse # Bus mouse
 # mouse PS2  # PS/2 mouse
 # mouse MouseMan # Logitech MouseMan
 # mouse Spaceball # Spacetec Spaceball
 # mouse IntelliMouse # Microsoft IntelliMouse or Logitech MouseMan+ on serial port
 # mouse IMPS2  # Microsoft IntelliMouse or Logitech MouseMan+ on PS/2 port
 # mouse pnp  # plug'n'pray
 # mouse WacomGraphire   # Wacom Graphire tablet/mouse
 # mouse DRMOUSE4DS # Digital Research double-wheeled mouse
 # mouse none  # None
 
 mouse unconfigured
 
 # (DEBIAN NOTE: the mouse used to default to microsoft, but this was changed
 #  to fix bug #13458. If your mouse used to work fine, you can simply change
 #  it back to read "microsoft" again. If you are careful to change *only that
 #  one word*, and not to add or remove extra whitespace, the package
 #  installation will continue to update this file without requiring user
 #  intervention because of a modified config file.
 #  This applies to all mouse types, not just microsoft.)
 
 # Mouse/keyboard customisation by 101 (Attila Lendvai). If you have any good
 # ideas you can reach me at 101@kempelen.inf.bme.hu
 
 # mouse_accel_type normal # No acceleration while delta is less then
     # threshold but delta is multiplied by
     # mouse_accel_mult if more. Originally done by
     # Mike Chapman mike@paranoia.com
 
 mouse_accel_type power # The acceleration factor is a power function
     # of delta until it reaches m_accel_mult. It
     # starts from the coordinate 
     # [1, 1 + m_accel_offset] and goes to
     # [m_accel_thresh, m_accel_mult]. If delta
     # is bigger then m_accel_thresh it is a plain
     # constant (m_accel_mult). It is the f(delta)
     # function with which the delta itself will be
     # multiplied. m_accel_offset is 1 by default,
     # so for delta = 1 the accelerated delta will
     # remain 1 (You don't lose resolution). The
     # starting point of the f(delta) function
     # might be moved along the Y axis up/down with
     # m_accel_offset thus defining the initial
     # minimum acceleration (for delta = 1).
     # Basically it's like the normal mode but the
     # acceleration factor grows as you move your
     # mouse faster and faster, not just turns in
     # and out. Threshold is the point from where
     # the f(delta) function gets linear.
     # This is the one I use for *uaking... =)
 
 # mouse_accel_type off # No comment...
 
 
 mouse_accel_mult 60 # This is the number with which delta will
     # be multiplied. Basically it's the number
     # that defines how big the acceleration will
     # be
 
 mouse_accel_thresh 4 # This is the threshold. See description by
     # power
 
 mouse_accel_power 0.8 # This is the second parameter of the power
     # function used in power mode. Used only by
     # the power mode
 
 mouse_accel_offset 30 # This is the offset of the starting point
     # on the Y axis. With this you can define the
     # number that will multiply delta = 1 so it's
     # the initial acceleration.
 
 # mouse_accel_maxdelta 600 # This is an upper limit for delta after
     # the acceleration was applied
 
 # mouse_maxdelta 30 # This is an upper limit for the delta
     # before the acceleration. With this one you
     # can limit the biggest valid delta that
     # comes from the mouse.
 
 # mouse_force   # Force parameters even if they seem strange
     # By default svgalib prints an error if any
     # of the numbers are somhow out of the
     # reasonable limit, (Like a negative mult :)
     # and uses the default that's in vgamouse.h
 
 # The default device is /dev/input/mice.
 # However, esp. with the Spacetec Spaceball you may
 # want to specify a different device for svgalib to use
 
 # mdev /dev/ttyS0 # mouse is at /dev/ttyS0
 
 # Some multiprotocol mice will need one of the following:
 
 # setRTS   # set the RTS wire.
 # clearRTS # clear the RTS wire.
 # leaveRTS # leave the RTS wire alone (default) (Wire is usually set)
 # setDTR   # set the DTR wire.
 # clearDTR # clear the DTR wire.
 # leaveDTR # leave the DTR wire alone (default) (Wire is usually set)
 
 # On mice such as the Microsoft IntelliMouse and Logitech MouseMan+, turning
 # the wheel rotates the mouse around the X axis. mouse_wheel_steps controls
 # how many steps make up a full 360-degree turn and thus how much rotation
 # occurs with each step. The default is 18 steps (20 degrees per step), the
 # real-world value for the IntelliMouse. Adjust it to match your mouse or to
 # suit your preferences; a negative number reverses the direction and zero
 # disables rotation.
 
 mouse_wheel_steps 18  # For MS IntelliMouse (default)
 # mouse_wheel_steps 24  # For Logitech FirstMouse+
 # mouse_wheel_steps -18  # Reverses direction
 # mouse_wheel_steps 0  # Disables rotation
 
 # mouse_fake_kbd_event sends a fake keyboard event to the program when the
 # wheel on a Microsoft IntelliMouse, Logitech MouseMan+, or similar wheel
 # mouse is turned. This can be useful for programs that do not recognize the
 # Z axis, but only works with some programs that use raw keyboard.
 # The format is:
 #
 #   mouse_fake_kbd_event upscancode downscancode
 #
 # The up and down scancodes are the scancodes of the keys to simulate when
 # the wheel is turned up and down, respectively.
 #
 # Scancodes can be specified numerically or symbolically; the symbolic names
 # are determined by the keymap (see below), if no keymap is loaded the default
 # is the standard US QWERTY keyboard with the following names available:
 # letters (a-z), numbers (zero-nine), function keys (F1-F12), the keypad
 # numbers (KP_0-KP_9) and other keys (KP_Multiply, KP_Subtract, KP_Add,
 # KP_Period, KP_Enter, and KP_Divide), and the following - minus, equal,
 # Delete, Tab, bracketleft, bracketright, Return, Control, semicolon,
 # apostrophe, grave, Shift, backslash, comma, period, slash, Shift, Alt, space,
 # Caps_Lock, Num_Lock, Scroll_Lock, Last_Console, less, Control_backslash,
 # AltGr, Break, Find, Up, Prior, Left, Right, Select, Down, Next, Insert,
 # and Remove.
 #
 # Note that this option has no effect unless the IntelliMouse or IMPS2 mouse
 # type is used (see above). Also note that the simulated keypresses are
 # instantaneous, so they cannot be used for functions that require a key to
 # be held down for a certain length of time.
 
 #  This example simulates a press of the left bracket ([) when the wheel is
 #  turned up and a press of the right bracket (]) when the wheel is turned
 #  down (good for selecting items in Quake II):
 # mouse_fake_kbd_event bracketleft bracketright
 
 # Keyboard config:
 
 # kbd_keymap allows you to use an alternate keyboard layout with programs that
 # use raw keyboard support by translating scancodes from the desired layout to
 # their equivalents in the layout expected by the program. This option has no
 # affect on programs that do not use raw keyboard.
 #
 # Keymap files to convert between any two arbitrary keyboard layouts can be
 # generated with the svgakeymap utility, but there are limitations to the
 # translations that can be performed. Read the file README.keymap in the
 # svgalib documentation directory for more in-depth information.
 #
 # You must specify the full path to the keymap file; it is recommended that
 # keymaps be kept in the same directory as libvga.config, normally /etc/vga.
 # The keymap specified in the configuration file can be overriden by setting
 # the environment variable SVGALIB_KEYMAP to point to another keymap file;
 # this can be useful for setting keymaps on a per-program basis.
 #
 #  This example will use the provided US-Dvorak to US-QWERTY map to allow a
 #  Dvorak keyboard layout to be used with a program that expects a standard US
 #  QWERTY keyboard, for instance Quake:
 # kbd_keymap /etc/vga/dvorak-us.keymap
 
 # There is a potential security risk in allowing users to remap keyboard
 # scancodes at will; with this option enabled only keymap files owned by
 # root can be used. Normally you should leave this on, but if you have a
 # single-user box or you really trust your users you may find it convenient
 # to run without it and allow users to load arbitrary keymaps.
 
 kbd_only_root_keymaps
 
 # kbd_fake_mouse_event, as it says, sends a fake mouse event to the program.
 # The format is: kbd_fake_mouse_event scancode [flag(s)] command [argument]
 #   Scancode is a raw scancode or a descriptive name, the same as with fake
 #   keyboard events (see above). If you use keymap conversion, specify
 #   scancodes for the keyboard layout the program will receive.
 #   Flags:   down   - trigger event when the key is pressed (default)
 #      up     - the opposite
 #      both   - trigger in both case, if pressed/released
 #      repeat - repeat events if the key is kept pressed (off by default)
 #   commands: delta[xyz]  - send a fake delta event as if you have moved your
 #        mouse. If the parameter is 'off' / 'on' it will turn
 #       off/on the respective mouse axis (requires a
 #       parameter, of course)
 #       button[123] - send a fake event that the mouse button is pressed
 #       or released that's given by the parameter.
 #        ('pressed' or 'released')
 # Here are some examples:
 
 #  This is one I use in *uake: it turns around, looks down a bit and when the
 #  key is released it does the opposite, so it gets back to the starting state.
 #  With this one and the help of a rocket you can fly though the whole map :)
 #  (Scancode 28 is enter)
 # kbd_fake_mouse_event 28 both deltax 8182 down deltay -1500 up deltay 1500
 
 #  This one will switch off the y axis of the mouse while the key (right ctrl)
 #  is kept pressed.
 # kbd_fake_mouse_event 97 down deltay off  up deltay on
 
 #  This one is the same as if you were pressing the left mouse button. (But
 #  if you move your mouse then the button state will reset even if you keep
 #  right ctrl down...)
 # kbd_fake_mouse_event 97 down button1 pressed up button1 released
 
 # Monitor type:
 
 # Only one range can be specified for the moment.  Format:
 # HorizSync min_kHz max_kHz
 # VertRefresh min_Hz max_Hz
 
 # Typical Horizontal sync ranges
 # (Consult your monitor manual for Vertical sync ranges)
 #
 # 31.5 - 31.5 kHz (Standard VGA monitor, 640x480 @ 60 Hz)
 # 31.5 - 35.1 kHz (Old SVGA monitor, 800x600 @ 56 Hz)
 # 31.5 - 35.5 kHz (Low-end SVGA, 8514, 1024x768 @ 43 Hz interlaced)
 # 31.5 - 37.9 kHz (SVGA monitor, 800x600 @ 60 Hz, 640x480 @ 72 Hz)
 # 31.5 - 48.3 kHz (SVGA non-interlaced, 800x600 @ 72 Hz, 1024x768 @ 60 Hz)
 # 31.5 - 56.0 kHz (high frequency, 1024x768 @ 70 Hz)
 # 31.5 - ???? kHz (1024x768 @ 72 Hz)
 # 31.5 - 64.3 kHz (1280x1024 @ 60 Hz)
 
 HorizSync 31.5 35.5
 VertRefresh 50 90
 
 # If you have a NeoMagic card on a Toshiba Libretto 100, 110 use that instead
 
 # HorizSync 31.5 70
 # VertRefresh 50 100
 # Modeline "800x480"   50  800  856  976 1024   480  483  490  504 +hsync +vsync
 # newmode  800 480 256       800 1
 # newmode  800 480 32768    1600 2
 # newmode  800 480 65536    1600 2
 # newmode  800 480 16777216 2400 3
 
 # Monitor timings
 #
 # These are prefered over the default timings (if monitor and chipset
 # can handle them). Not all drivers use them at the moment, and Mach32
 # has its own syntax (see below).
 # The format is identical to the one used by XFree86, but the label
 # following the modeline keyword is ignored by svgalib.
 #
 # Here some examples:
 
 # modeline "640x480@100"  43  640  664  780  848   480  483  490  504
 # modeline "800x600@73"   50  800  856  976 1024   600  637  643  666
 # modeline "1024x768@75"  85 1024 1048 1376 1400   768  771  780  806
 
 # It seems there is a need for a 512x384 mode, this timing was donated
 # by Simon Hosie <gumboot@clear.net.nz>: (it is 39kHz horz by 79Hz vert)
 
 # Modeline "512x384@79"     25.175 512  522  598  646   384  428  436  494
 
 # Here's a 400x300 Modeline (created by svidtune). Note that for
 # doublescan modes, the Vertical values are half the real one (so XFree86
 # modelines can be used). 
 
 # Modeline "400x300@72" 25.000 400 440 504 520 300 319 322 333 doublescan
 
 # Here is a mode for a ZX Spectrum emulator:
 # Modeline "256x192@73" 12.588 256 269 312 360 192 208 212 240 doublescan
 # newmode 256 192 256 256 1
 
 # the width must be divisible by 8. Some cards require even divisiblity by
 # 16, so that's preferable, since there are no standard modes where the
 # width is not divisible by 16.
 
 # The following modes are defined in svgalib, but have no timings in
 # timing.c, so you'll have to add a modeline in order to use them:
 # 1280x720, 1360x768, 1800x1012, 1920x1080, 1920x1440, 2048x1152
 # and 2048x1536   
 
 # Mach32 timings:
 
 # e.g. Setup a 320x200 mode for the mach32:
 
 #define 320x200x32K 320x200x64K 320x200x16M 320x200x16M32
 #                     16 320 392 464 552 200 245 265 310
 
 # These are REQUIRED for above mode, please edit to suit your monitor.
 # (No, I won't pay for a new one)
 # HorizSync 29 65
 # VertRefresh 42 93.5
 
 # Chipset type:
 #
 # Use one of the following force chipset type.
 # Autodetects if no chipset is specified.
 #
 # If you have a PCI or AGP card, don't use chipset type forcing.
 # If the card is not autodetected, its a bug, and it will probably
 # not work even with forcing. Try running vgatest (with no chipset 
 # line), and send to me (matan@svgalib.org) the output, a copy of
 # /proc/pci (or lspci -n -vv) and whatever info you have on the card. 
 #
 # If a chipset driver gives trouble, try forcing VGA.
 
 # chipset VGA  # Standard VGA
 # chipset EGA  # EGA
 # chipset ET3000 # Tseng ET3000
 # chipset ET4000 # Tseng ET4000
 # chipset Cirrus # Cirrus Logic GD542x
 # chipset TVGA  # Trident TVGA8900/9000
 # chipset Oak  # Oak Technologies 037/067/077
 # chipset S3  # S3 chipsets
 # chipset GVGA6400 # Genoa 6400
 # chipset ARK  # ARK Logic
 # chipset ATI  # old ATI VGA
 # chipset Mach32 # ATI Mach32
 # chipset ALI  # ALI2301
 # chipset Mach64 # ATI Mach64 - deprecated
 # chipset ET6000        # Tseng ET6000
 # chipset APM   # Alliance Technology AT 24/25/3D
 # chipset NV3  # nVidia Riva 128 / TNT / GeForce
  chipset VESA          # nicely behaved Vesa Bioses
 # chipset MX  # MX86251 (some Voodoo Rush boards)
 # chipset PARADISE  # WD90C31
 # chipset RAGE  # RagePro (and might work with some older mach64)
 # chipset BANSHEE # Banshee/V3.
 # chipset SIS  # SiS 5597/6326/620/530 cards / integrated vga.
 # chipset I740  # Intel i740 based cards.
 # chipset NEOMAGIC
 # chipset LAGUNA # Cirrus Logic Laguna series (546X)
 # chipset FBDEV  # Use kernel fbdev, instead of direct hardware.
 # chipset G400  # Matrox Mystique/G100/G200/G400/G450
 # chipset R128  # Ati Rage128
 # chipset SAVAGE # S3 chipsets Savage
 # chipset C&T  # Chips and Technologies
 
 # EGA Color/mono mode:
 # Required if chipset is EGA.
 #
 # Use one of the following digits to force color/mono:
 
 # monotext  # Card is in monochrome emulation mode
 # colortext # Card is in color emulation mode
 colortext
 
 # RAMDAC support:
 # Some chipsets (e.g. S3 and ARK) allows specifying a RAMDAC type.
 # If your RAMDAC is not autodetected, you can try specifying it.
 # Do NOT specify a RAMDAC if you card uses the S3 Trio chipset
 # (the RAMDAC is built in).
 
 # Ramdac Sierra32K
 # Ramdac SC15025
 # Ramdac SDAC         # S3 SDAC
 # Ramdac GenDAC       # S3 GenDAC
 # Ramdac ATT20C490    # AT&T 20C490, 491, 492 (and compatibles)
 # Ramdac ATT20C498    # AT&T 20C498
 # Ramdac IBMRGB52x    # IBM RGB524, 526, 528 (and compatibles)
 
 # Dotclocks:
 # Some chipsets needs a list of dot clocks for optimum operation.  Some
 # includes or supports a programmable clock chip.  You'll need to specify
 # them here.
 
 # Fixed clocks example:
 # (The following is just an example, get the values for your card from
 #  you XF86Config)
 
 # Clocks 25.175 28.3 40 70 50 75 36 44.9 0 118 77 31.5 110 65 72 93.5
 
 # Programmable clockchip example:
 
 # Clockchip ICD2061A  # The only one supported right now
 
 
 
 # Here are miscellaneous options to help with specific problems.
 
 # VesaText       # Helps the VESA driver with text mode restoration
         # problems.
 
 # VesaSave 14       # changing value might help text mode restoring
         # problems with VESA driver. Legal values: 0-15
 
 # NoVCControl       # Disables svgalib's finding a new VC if run
         # from X. Good fo using dumpreg under X, but
         # probably bad for standard usage.
 
 
 # RageDoubleClock     # If your card is based on ATI's rage card, and
         # the pixel clock is double what it should be 
         # (main symptom is some modes are out of sync),
         # try enabling this. If it helps, please report to
         # me (matan@svgalib.org) 
 
 # NeoMagicLibretto100
         # Enable if you have a NeoMagic card on a Toshiba
         # Libretto 100, 110, etc

  chipset VESA          # nicely behaved Vesa Bioses

uncomment L344, 使用 VESA driver。基本上以目前的顯示卡來說, vesa driver 應該都可以正確執行。都不改也可以, svgalib 會自動偵測可用的顯示模式。 再來看看測試目前可以設定的解析度有哪些。 svgalib-1.9.25/demos/vgatest (這是 svgalib 自帶 source code 裡頭的程式) 執行畫面:

vga driver

寄件者 svgalib test

使用的是 vga driver, 最近研究了 vga, 看這上面列出的解析度, 我已經了解為什麼是這些解析度, 在我剛開始使用 svgalib 時, 我完全不知道 vga 相關技術。

vesa driver

在我的卡上, vga driver 只有 10 種解析度可使用, vesa driver 就有比較多的解析度可用。vesa 這部份不知道為什麼可以有這麼多的解析度可以選擇, 列入下次的功課。

640X480X24 bit color

我選擇了幾個模式, 似乎 640x480X24 bit color 是比較好的選擇, 有些模式跑不起來或是怪怪的。選擇了要使用的解析度之後, 就可以來練習寫 console 繪圖程式了。理論上 vesa driver 應該可以跑得起來, 若不行, 那只能使用 vga driver, 不過只能有 320x200x256 色這樣的環境。如果連 vgatest 都跑不了, 那還是用 linux framebuffer 吧。

以下影片是我在 eeepc901/linux 上的測試



選擇 640x480X32bit color, 程式會正常的秀出所有測試效果, 只是有點慢就是。



simple os 目前只能使用 320x200x256 色, 我希望可以編寫 vesa 的版本, 使其使用 640x480X24 bit color, 不過資料真的不好找。

1.4.3 的版本需要使用 root 權限來執行 compile 出來的程式, 1.9.25 則需要 kernel module, 其實不好用, 因為已經跟不上 linux kernel 的版本了。練習的話建議使用 1.4.3 即可。

現在我有了 simple os, 比較少碰 svgalib 了, 紀錄下來做個回憶。

在我的 mac mini/eeepc 901 intel i915 card 上, 若是載入 i915 gpu module, vgatest 沒有任何一個顯示模式可正常使用, 大概是相衝了。我將整個 i915 module 移除則可以正常顯示其中幾個模式。

fbset -s 可以用來測 linux framebuffer 是不是起動了。以下是有啟動 linux framebuffer 的結果。

mode "640x480-60"
    # D: 25.176 MHz, H: 31.469 kHz, V: 59.942 Hz
    geometry 640 480 640 480 4
    timings 39721 48 16 33 10 96 2
    rgba 6/0,6/0,6/0,0/0
endmode

為了學習 svgalib 我還特地買了這本書:

linux graphics programming with svgalib

寄件者 computer books

也可參考這篇 - SVGAlib Tutorials svgalib 有兩個系列的繪圖函式: vga_xxx gl_xxx gl_xxx 系列在繪圖上比較快, 用法有點複雜, 建議從 vga_xxx 開始學習比較簡單。

svgalib 很低階, 有些用法需要知道 vga 相關技術, 否則看 man page 的說明, 無法知道為什麼會有這樣的限制。

為什麼要學習繪圖程式, 因為我想輸出中文, 使用中文的人如果不會寫中文系統, 好像說不過去, 無論如何我想要知道如何秀出中文, 若不依靠中文系統或是中文終端機要如何秀出中文。

早期 dos 時代有很多這樣的資料, 現在的書似乎比較少著墨。

我有一本書專門在講這個:

文字中文系統徹底研究 - 輸入法與秀字

有點過時, 不過基本原理是不變的。

寫這篇時, 我竟然手癢, 順手修改了 jmcce, 讓他可以在某種程度上繼續執行。



ref:
http://www.svgalib.org/ http://mail.arava.co.il/~matan/svgalib/ https://github.com/ryanmcgrath/svgalib-1

git 版本讓我驚訝, 原來新版的 svgalib source code 早已經搬到 github 上了。