I need the following tested on your arch. If an ebuild doesn't work, please let me know, so hopefully we can fix it before the next release. app-misc/livecd-tools-1.0.3 - These are tools used for auto detection of hardware for the LiveCD's. Please report if any of these parts do not work on your arch, or if your arch doesn't have need for them. A good example would be the X stuff on mips. sys-apps/ddcxinfo-knoppix-0.6 - This is used for detecting monitor frequencies for X. sys-apps/hwdata-knoppix-0.107 - Definitely test this one, as I would like to remove the older versions from portage if they are not needed. sys-apps/kudzu-knoppix-1.1.36 - This should be a drop-in replacement for kudzu. sys-apps/hwsetup-1.0-r1 - This is just the latest release of hwsetup. x11-misc/mkxf86config-0.8-r1 - Test this to see if we can use this on your arch. Currently, it only works with <=xfree-4.3.0-r5 as it is not updated with the new font paths using in -r6 and higher and xorg-x11. Also, if any of the above do not work on your arch, but you have a package which fills the same role and does work, please let me know.
Guys? If you know that these won't work on your arch, please let me know.
I would say that most of these tools are completely useless on mips, however I can only speak for certain on those pertaining to configuration of X. Anything having to do with detection of monitor frequencies or setup of an X config file is definitely pointless. Currently, the only box that really works with X is an indy, and no matter what you set the monitor modeline to in the config file, it will automatically run at 1280x1024 at 60hz. Another thing, if mkxf86config doesn't work with xorg-x11, then in my opinion it is broken and shouldn't be used. Mips is moving towards using xorg-x11 as default. Iluxa could probably tell you a lot more about mips livecd stuff and the relevancy of the other tools in this bug, as he has been working on making one for a while now.
Thanks for the response, I figured the X stuff was moot on mips. As for the mkxf86config, well, I'm about to put out a new version that works with xorg-x11, since x86 is now going to be running xorg-x11 for the 2004.2 release. Iluxa: Any information on the other tools?
For sparc, the X tools are useless since we use xorg-x11, and the kudzu tools are useless, since they only work with PCI.
I'd wait for pvdabeel to comment more, anyway mkxf86config is unnecessary on ppc (we have other tools like xautoconf)
i'd like to note that xfree is officially deprecated on amd64 due to -fPIC issues and the like. so if it doesnt do xorg-x11, it's pretty useless here too. ^^;
Lv: I need to know if it *works* at all... it *will* work with xorg-x11, just all the fonts will be wrong... that is nothing more than a sed script in the ebuild to fix... which I'll be doing in the next couple of days... What I'm really looking for is if you see any reason why they wouldn't run. I'm pretty sure they'll work fine on amd64, though.
Does this need to remain open, since we made it past the 2004.2 release? They worked for amd64.
These have little to do with 2004.2, so yeah... Though I will say that I've updated them to work with xorg-x11 *or* xfree, so they should work on any platform that uses them (and PCI).
removing ppc from cc
Actually, I need to know the state for ppc on these items, since we're looking at adding them into the way catalyst builds X-capable LiveCD's.
The versions on a couple packages have changed due to bug fixes. app-misc/livecd-tools-1.0.5 sys-apps/hwsetup-1.0-r2 x11-misc/mkxf86config-0.8-r3 (worls with xorg-x11) Would all arches still on the CC please test these and report back on their functionality? Thanks.
The current 2004.2 X-enabled ppc livecd only uses the net-select provided by livecd-tools. We use Xautoconfig instead of xf86config. Xautoconfig works fine with both xorg and xfree, except that it defaults to a font server (/me wonders whether that still is the default on redhat?). The hwdata, kudzu and hwsetup seem relatively useless for ppc: If I'm not mistaken these tools do some fancy hardware detection and modprobe modules for the stuff they detect. I feel it would be overkill to enable this since the amount hardware we have to support is really small, and it's often easier to just enable the things we need it in the kernel. I wouldn't be opposed to adding it if it is needed. dccxinfo, haven't tried but... Apple displays have really, really strange resolutions and interfaces, which is also part of the reason why we don't have bootsplash enabled out of the box, even though it works on screens with 'normal' resolutions. We also don't have vesa or vga. That makes things rather tricky. Right now Xautoconfig detects resulutions by querying the framebuffer (radeon, nvidia, or openfb). That seems to work just fine. Openfirmware boots your machine automatically in the biggest possible resolution for your framebuffer. (If you used 1600x1200 in OS X, it will try to use that resolution to boot linux. We instruct the user to pass the required resolution to the kernel.) Since we don't have vga=xxx as on x86, this part is rather different too. dccxinfo might be usefull, but we haven't really tried. I only have regular pc screens attached to my macs. Has anyone in ppc got an apple cinema display or something similar he/she can try things on?
Did you mean net-setup? Also, could you check if mkxf86config works for ppc? It should work for XFree86 3.x, XFree86 4.x, and X.Org, without using a font server. In fact, it uses the font locations that Gentoo uses by default (via a patch I made). As for the hwdata/kudzu/hwsetup, could you test them on PPC to ensure they function? While I understand that you may not use them for your release media, one of our users may decide to use it, and I plan on using them for the creation of all of the GameCD additions to catalyst, so it would be beneficial to know what archetectures the applications actually compile and work on. As for ddcxinfo, I use it in the x-setup script, which uses the information gotten from hwsetup/kudzu and mkxf86config to produce a working X configuration. I've tested it on both amd64 and x86, and if we could end up using a common interface for this across as many arches as possible, it would be very beneficial, as it could be added into catalyst.
Yes, I meant net-setup. Sorry about that. I didn't have much success with the tools you asked me to test: - ddcxinfo-knoppix fails to emerge. dual-g5 ddcxinfo-knoppix # emerge mkxf86config Calculating dependencies ...done! >>> emerge (1 of 6) sys-apps/ddcxinfo-knoppix-0.6 to / >>> md5 src_uri ;-) ddcxinfo-knoppix_0.6-5.tar.gz >>> Unpacking source... >>> Unpacking ddcxinfo-knoppix_0.6-5.tar.gz to /var/tmp/portage/ddcxinfo-knoppix-0.6/work >>> Source unpacked. gcc -E -M -Wall -O2 -g *.c > .depend gcc -Wall -O2 -g -c -o ddcprobe.o ddcprobe.c gcc -Wall -O2 -g -c -o lrmi.o lrmi.c gcc -Wall -O2 -g -c -o vesamode.o vesamode.c gcc -Wall -O2 -g -c -o vbe.o vbe.c ddcprobe.c:3:20: sys/io.h: No such file or directory vbe.c:2:20: sys/io.h: No such file or directory ddcprobe.c:5:22: sys/vm86.h: No such file or directory bioscall.c:2:20: sys/io.h: No such file or directory bioscall.c:4:22: sys/vm86.h: No such file or directory lrmi.c:16:20: sys/io.h: No such file or directory lrmi.c:17:22: asm/vm86.h: No such file or directory lrmi.c:183: error: field `vm' has incomplete type lrmi.c: In function `pushw': lrmi.c:214: error: dereferencing pointer to incomplete type lrmi.c:215: error: dereferencing pointer to incomplete type lrmi.c:215: error: dereferencing pointer to incomplete type lrmi.c: In function `set_regs': lrmi.c:307: error: `IF_MASK' undeclared (first use in this function) lrmi.c:307: error: (Each undeclared identifier is reported only once lrmi.c:307: error: for each function it appears in.) lrmi.c:307: error: `IOPL_MASK' undeclared (first use in this function) lrmi.c: In function `em_inbl': lrmi.c:517: warning: implicit declaration of function `inb' lrmi.c: In function `em_outbl': lrmi.c:547: warning: implicit declaration of function `outb' lrmi.c: In function `run_vm86': lrmi.c:805: warning: implicit declaration of function `VM86_TYPE' lrmi.c:805: error: `VM86_INTx' undeclared (first use in this function) lrmi.c:807: warning: implicit declaration of function `VM86_ARG' lrmi.c:818: error: `VIF_MASK' undeclared (first use in this function) lrmi.c:818: error: `TF_MASK' undeclared (first use in this function) lrmi.c:823: error: `VM86_UNKNOWN' undeclared (first use in this function) lrmi.c: In function `LRMI_int': lrmi.c:910: error: `IF_MASK' undeclared (first use in this function) lrmi.c:910: error: `IOPL_MASK' undeclared (first use in this function) make: *** [lrmi.o] Error 1 make: *** Waiting for unfinished jobs.... ddcprobe.c:3:20: sys/io.h: No such file or directory ddcprobe.c:5:22: sys/vm86.h: No such file or directory vbe.c: In function `vbe_get_mode_info': vbe.c:42: warning: implicit declaration of function `iopl' vbe.c:43: warning: implicit declaration of function `ioperm' lrmi.c:16:20: sys/io.h: No such file or directory lrmi.c:17:22: asm/vm86.h: No such file or directory make: *** [ddcprobe.o] Error 1 !!! ERROR: sys-apps/ddcxinfo-knoppix-0.6 failed. !!! Function src_compile, Line 20, Exitcode 2 !!! (no error message) !!! If you need support, post the topmost build error, NOT this status message. dual-g5 ddcxinfo-knoppix # svgamodes.c:2:20: sys/io.h: No such file or directory svgamodes.c:4:22: sys/vm86.h: No such file or directory vbe.c:2:20: sys/io.h: No such file or directory This probably has something todo with ppc not having svga, nor vesa. - I started hacking and emerged mkxf86config without ddcxinfo-knoppix dual-g5 mkxf86config # mkxf86config.sh Video is (yet) unknown card, using X(vesa) Server Monitor is Old Monitor (no DDC), H:28.0-96.0kHz, V:50.0-60.0Hz Using Modes "1024x768" "800x600" "640x480" dual-g5 mkxf86config # The G5 has a regular PC monitor (a Relisys 23inch) attached, so no special PPC resolutions on this one. The generated XF86Config didn't work. The vga16 driver is not a ppc driver, there is no accel module on ppc and fbdev works in the xorg.conf provided below, but not in the generated config. This is a working PPC xorg.conf for a Dual-G5 with a radeon 9600 (note Xautoconfig configured it for a font server, we have a bug open about that): Please also note that the keyboardmodel is different on ppc. dual-g5 X11 # cat xorg.conf # File generated by Xautoconfig. Section "ServerLayout" Identifier "XFree86 Configured" Screen 0 "Screen0" 0 0 InputDevice "Mouse0" "CorePointer" InputDevice "Keyboard0" "CoreKeyboard" EndSection Section "Files" # RgbPath is the location of the RGB database. Note, this is the name of the # file minus the extension (like ".txt" or ".db"). There is normally # no need to change the default. # Multiple FontPath entries are allowed (they are concatenated together) # By default, Red Hat 6.0 and later now use a font server independent of # the X server to render fonts. RgbPath "/usr/X11R6/lib/X11/rgb" FontPath "tcp/localhost:7100" EndSection Section "Module" Load "dbe" Load "extmod" Load "fbdevhw" Load "glx" Load "record" Load "freetype" Load "type1" #Load "dri" EndSection Section "InputDevice" Identifier "Keyboard0" Driver "keyboard" # Change "XkbModel" to "macintosh_old" if you are using # the deprecated adb keycodes. Option "XkbRules" "xfree86" Option "XkbModel" "macintosh" Option "XkbLayout" "us" EndSection Section "InputDevice" Identifier "Mouse0" Driver "mouse" Option "ZAxisMapping" "4 5" Option "Protocol" "IMPS/2" Option "Device" "/dev/input/mice" EndSection Section "Monitor" Identifier "Monitor0" UseModes "Modes0" ModelName "Monitor Model" Option "DPMS" HorizSync 30-95 VertRefresh 50-60 EndSection Section "Modes" Identifier "Modes0" # Generated # D: 108.003 MHz, H: 63.983 kHz, V: 60.021 Hz Modeline "1280x1024" 108.003 1280 1328 1440 1688 1024 1025 1028 1066 -HSync -VSync EndSection Section "Device" Identifier "Card0" ### Available Driver options are:- #Option "SWcursor" Option "HWcursor" #Option "PciRetry" #Option "NoAccel" #Option "ShadowFB" "true" Option "UseFBDev" "true" Driver "radeon" BusID "PCI:240:16:0" EndSection Section "Screen" Identifier "Screen0" Device "Card0" Monitor "Monitor0" DefaultDepth 16 SubSection "Display" Depth 8 Modes "1280x1024" EndSubSection SubSection "Display" Depth 16 Modes "1280x1024" EndSubSection SubSection "Display" Depth 24 Modes "1280x1024" EndSubSection EndSection Section "DRI" Group 0 Mode 0666 EndSection - dholm did some testing of hwsetup, ... The tools work (in ~ppc). For PPC, I'd like to patch Xautoconf to use font paths instead of a font server. It creates configs correctly. It's a non-interactive tool, so it can probably serve as the ppc alternative for gamecds. I'd prefer hotplug + udev instead of kudzu for automatic module loading on ppc, but kudzu would work just fine I guess.
Wow... that's just awesome. Thanks for all the information. Does Xautoconf work on x86? We actually don't use kudzu for module loading, but hwsetup + hotplug. The only thing is that hwsetup requires kudzu to be present for it to work properly.
ok.. I've tested those apps: - sys-apps/hwdata-knoppix-0.107 : compiles - sys-apps/hwsetup-1.0-r2 : compiles and 'works'. Well... I don't know what this programm is supposed to do, but it just does nothing. no output nor errors... - app-misc/livecd-tools-1.0.9 : compiles and net-setup and spind work fine, but x-setup doesn't. - sys-apps/ddcxinfo-knoppix : doesn't compile - sys-apps/kudzu-knoppix : segmention faults... I'll add some more information when I have some more comfortable browser than links2 is... ^_^ Markus
Markus: you never said on what architecture you were testing
nah... that was ppc64. sorry for the late answer. all those apps are now ~ppc64 and will be stable in a few weeks of testing. (besides of mkxf86config, which is not recessary on ppc64 as it also uses -like ppc- Xautoconfig to configure xorg) If you need further information or if something is wrong with this, please re-add ppc64 to CC. Markus P.S.: The segmention faults might be solved by the newer gcc version.. :-)
These have been tested for a while now... thanks everyone...
