Problem installing xserver
Well, I thought I'd seen this problem in the archives, but I can't for the life of me find it... so apologies if this is a FAQ by now: I cannot seem to get xserver-xfree86 configured on a Powerbook G3 (Lombard). I've tried multiple kernels to no avail, including yesterday's snapshot of Cort's pmac 2.4.4 tree, debian 2.2.19, etc. The symptoms are that the xserver package hangs when configuring (both 4.0.2-13 and 4.0.3-1), forcing me to manually break (^C) to get back to that prompt. I did realize that I had an old XF86Config, and no XF86Config-4. I dropped in the XF86Config-4 from Sergio Brandano's site (thanks!), but that hasn't had any effect on the problem. Any clues for the clueless? 8-) Thanks, John
Re: Problem installing xserver
I wrote: > I cannot seem to get xserver-xfree86 configured on a Powerbook G3 > (Lombard). [...] > > The symptoms are that the xserver package hangs when configuring (both > 4.0.2-13 and 4.0.3-1), forcing me to manually break (^C) to get back > to that prompt. For the record, the problem turns out to have been with the "mdetect" mouse autodetection package. xserver-xfree86 was attempting to use it during configuration, and hanging. Once the mdetect package was removed, installation proceeded smoothly. -- John
PPC Package issues and possible dselect bug
A few "bugs" in package dependencies:
logrotate requires popt but popt "doesn't appear to be available"
hfsutils-tcltk requires "tk8.0" instead of "wish", and thus won't
accept the tkstep8.0 version of wish.
Bug in perl dependencies (or dselect itself, depending on how you look
at it), as evidenced by the following:
> perl-5.00N recommends perl-5.00N-doc
> perl-5.00N-doc recommends perl-5.00N
("N" can be either 4 or 5, both versions have analogous dependencies.)
Even with both packages selected for installation, dselect insists on
bringing up the dependency/conflict page until one finally hits 'Q' to
dismiss its confusion.
-- John "Sooner or later, this Lombard WILL run Debian" Whitley
kbd package install problem and Lombard problems
In installing the kbd package (0.99-2, I think... whatever is currently in http://http.us.debian.org/ unstable main), I've gotten into an infinite loop in the script asking the "Do you want the kernel's default keymap to be reloaded?" If I answer 'n RET', and I just get the question again. If I just hit RET, I get the message "Cannot find defkeymap.map" There is no way to break out of the script; neither ^C nor ^Z work. (^ this was annoying. I'm new to Debian, and unfamiliar with exactly how dselect and its child processes/scripts work together, but SIGINT should either work outright or be trapped to produce a confirmation query.) During the original config, I also specified 'n RET' as per the instructions at http://www.debian.org/~wmono/powerpc/ , then hit return again to abort the keymap selection process. -- John Whitley
Lombard problems
Hi all, First, despite my problems below, thanks to those who have posted their experiences getting Debian up on a Lombard... you've helped me get this far. 8-) I'm having a heck of a time getting a reasonable Debian install on my Lombard G3/400. I have no trouble with the install (via NFS) up to the point of the first time dselect runs. I made sure to apt-get the current libc6 and bash during the install, and both installs ran fine. However, once I start grabbing packages things break. Specifically, there seems to be a specific package(s) and/or libraries in the default dselect install that causes the system to freeze, requiring a hard reset. There seems to be little noticable pattern to the lockups, except that it is frequent, occurs while an app is running (see below) and affects most apps. dselect, dpkg, apt-get (natch), and gcc all seem particularly susceptible to locking up (but are also the apps I'm trying to run, so take this with a grain of salt.) I've yet to get even _near_ a working X installation, much less messing around with server configuration, despite success stories to the contrary. I've tried different kernels to no avail: current "LinuxPPC Standard", vmlinux.lombard from the LinuxPPC site vs. the Debian 2.2.10 kernel (installed then copied the kernel over to the BootX folder). vmlinux.lombard is the only pre-built kernel I've found with working ethernet support. I've tried to get Paul Mackerra's stable kernel compiled with as few necessary packages installed as possible, as mentioned in a recent thread, but even when I get gcc installed, it locks up very early into make dep. I've yet to be able to build any kernel from source. I've been going around with this for a couple days now, and am at wits' end. Any ideas? Am I missing something obvious, or a critical bit of info about current bugs/install peculiarities? Thanks much, John Whitley
Broken http.us.debian.org site
It appears that the http.us.debian.org site ftp1.us.debian.org, aka IP 206.187.92.15, isn't correctly mirroring the unstable/main directories for powerpc. Specifically: http://http.us.debian.org/debian/distsunstable/main/binary-powerpc is an empty directory. I took the liberty of emailing the webmaster of that site regarding the mirror failure. This is the source of those annoying 404 Not Found errors that crop up when using the Debian package tools. -- John
Lombard finally working...
My sincere thanks to those who responded to my queries for install assistance on a Lombard powerbook. I have finally determined the source of the problem: a bad DIMM. !!! On the up side, I am now very familiar with current linux library organization and the Debian package management software. 8-) With the offending DIMM removed, I've got Debian running with Paul Mackerras' latest. I even got X running, altho the performance is slow at present. Moving forward now, John
Re: Finally launched
Phillip R. Jaenke writes: > For most of us, I don't think that's the question.. question > is.. when can *WE* get our hands on it? :) Ah, but that's a mere matter of shipping schedules and personal finance... easily determined. 8-) My question is: when will gcc have support for AltiVec instructions?? -- John
Re: Minor Problem Potpourri
T.E.Dickey (thru Deja) writes: > your arrow keys are in normal mode, where the termcap says they're in > application mode (a problem that happens when someone has removed the > ti/te settings) Hmm. On this basis, as a hack, I tried adding these codes (actually smcap and rmcap for terminfo) from the xterm entry to the linux terminfo definition from the xterm terminfo description (actually, I tried both the smcap/rmcap pairs from IRIX and from Debian, as they differed), to add to the linux term defs. Didn't work, of course. 8-) The entries probably aren't recognized or correct. But then I note that none of the vt1xx/2xx/3xx entries have the smcap/rmcap either. Next gross ugly hack: change the terminfo arrow key defs for linux to \EOA thru \EOD. Works like a charm. But none of this answers the question: why/how did my arrow keys change modes in the first place? why/how is this not accounted for in the 'linux' terminfo entry? Are the ti/te settings actually broken, or is something else amiss? Yet all of that is moot compared to: "How do I fix it--the right way?" -- John
Re: Minor Problem Potpourri
Robert Ramiega writes: > Check Your mirror manually or change to some other Debian mirror. Here is > what dpkg -l gives me: Well, it's good to know that the package exists somewhere. ;-> I've checked numerous mirrors, also gone to saens.debian.org, and the package just isn't there. There is netscape-base-408 and netscape-base-46, both of which also pre-depend on netscape-base-4. In fact, (at least on saens) i386 doesn't have netscape-base-4 either. (Nor is it in binary-all..) -- John
Got xmcd running
If anyone's interested, I managed to get xmcd working, but only from a non-debian source install. Seems that the debian source and binary xmcd packages (at least for powerpc) are broken; xmcdconfigure (sp?) doesn't work and xmcd's configuration files aren't set up correctly. Also, the behavior of various mixer programs, at least on my Lombard, seems to be somewhat strange, at the least. Anyone have similar experiences to swap, before I bugreport? Thanks, John
Re: Got xmcd running
Jeramy B Smith writes: > Consider funtional sound on Apple equipment a black art for now. I finally figured out what was going on with the mixer packages (at least with wmmixer) -- the master volume control seems to be for the external audio jack -- the speaker control is for the internal speaker -- there is no intelligent auto-mute of the internal speakers when the external audio jack is in use, one must manually turn off the speaker volume and adjust the external volume. (I'd incorrectly assumed that speaker muting was effected at the hardware level.) There are still some quirks, such as a volume control that seems to work in inverse to the regular volume control.. but sound otherwise works fine with the above considerations. Later, John
Screenblank powerdown LCD display?
More wishful thinking: Does there yet exist a way to turn off a Powerbook's LCD (ala the MacOS option) when screenblanking takes effect? -- John
Re: Screenblank powerdown LCD display?
Benjamin Herrenschmidt writes: > John Whitley <[EMAIL PROTECTED]> wrote: > >Does there yet exist a way to turn off a Powerbook's LCD (ala the > >MacOS option) when screenblanking takes effect? > > It should be automatic. The various video drivers used by PowerBooks > contains calls to the PMU driver to do that when VESA blanking occurs. If > your console/X does VESA blanking, the LCD should turn off. Hmm.. no go. I'm using frame buffer acceleration in combination with the ATI Rage LT Pro in my PB G3/400, and XF68_FBDev on top of that. Relevant output of 'xset q': Screen Saver: prefer blanking: yesallow exposures: yes [.. deletia ...] DPMS (Energy Star): Display is not capable of DPMS I have tried using the "setvesablank" program, with no effect. All the screen does is black out, but not power down. -- John
Re: Anacron and xcdroast problem report
Frank Petzold writes: > > problem. Here's a strace on "anacron -s": FYI, the anacron -s problem has been identified, a patch submitted, and should cease obnoxiously hanging your system once the next version is uploaded. The current broken one is 2.1-1; wait for its successor. -- John
gcc 2.95.1 safe to use?
Someone whose attribution got lost in the shuffle wrote: > PowerPC is now completely broken, as the old binutils won't accept the > target requested by gcc 2.95.1. On a vaguely related topic, is gcc 2.95.1 safe to use for compiling kernels yet? I've put my gcc related packages on hold until I can find reliable confirmation that kernel building will work with the new gcc. -- John
Lombard status update
It seemed a good time to again compare notes with others running Debian on Lombard, to share solutions and find common problems. Notable packages/services that work: X68_FBDev in unaccelerated mode (via "No video driver" in BootX) anacron -- I submitted a bugfix that corrects the hang on startup; watch servers near you for the updated version. WindowMaker -- patched 0.60.0 deb srcs by hand to fix crash on desktop change. This fix was mentioned on the wmaker list, but is post 0.60.0, thus hasn't propagated to the world at large. (This is an endianness-related bug, and affects all big-endian systems, not just PPC.) xmcd -- built from most recent upstream source, .deb is hopelessly outdated. xmms -- and mpg123, naturally) cdparanoia -- works fine with the G3/400's DVD (audio apps are notable, as I do audio compression research) hdparm -- put script with "hdparm -p /dev/hda" and "hdparm -S 36 /dev/hda" in /etc/rc.boot. Works like a charm. What mostly works: XF68_FBDev in accelerated mode -- With just video=atyfb:vmode:14,cmode:32, the accelerated fb driver is loaded according to kernel log messages. Yet problems abound in accelerated mode. It's the most hosed excuse for an X server I've ever seen. * Various line/rectangle/polygon primitive errors (unsure precisely which of these are affected) e.g. rectangles with a side connecting diagonally, as if the algorithm tried to close from 1st to to the N-1st point instead of 1st to Nth point. * Odd problem exemplified by WindowMaker's fading desktop name animations: the text underneath isn't properly redrawn, only background for the obscured window remains. This happens in other contexts, such as with Netscape IMG tag bubble windows and the like. * Greyed out options (presumably rendered using a GC stipple) in various applicatons are unreadable, appearing as a few scattered lines and dots. sound -- wmmixer can be used to control sound levels on the Lombard, but as my earlier messages indicate, things are a little wierd right now. See the following "snooze" entry, also. snooze -- sleeps the laptop just fine, but various things are not reset correctly or are outright hosed on wakeup: * sound vanishes after sleep/wakeup, and doesn't come back. * trackpad settings (e.g. "trackpad notap") are not preserved XEmacs 21 -- Built from upstream source. Options menu has virtually all fonts greyed out, for no immediately obvious reason. Even the Courier font used by default is greyed out. Does accept direct font settings via set-font-face. What Masticates the Mustelidae: console video mode -- garbled junk, all the time, unless an explicit vmode is run on the console command line. *NO* kernel vmode settings that I have tried (this includes everything on this list, the web, and DejaNews) under any of Mackerras' or vger kernels give readable output. * The visual effects are random, from "swirling noise" (aka "penguin in a blender") to black. This seems to indicate a problem with fb and/or console driver initialization. * I'd _like_ to use the SUN12x22 console font, but for some reason, it never does work right anymore. Yes, it is compiled into my kernels. In two early attempts, it was working just fine, but I've never been able to recreate those conditions since. Explicit "vmode 14 32" brings up the standard PeeCee VGA font, never the SUN font. Oddly, however, if the SUN12x22 was specified in the kernel video options, 80x30 is the terminal window size after the manual vmode. * Once X starts, it returns the framebuffer to sanity. Lack of Meta policy -- There is no sensible policy whatsoever regarding option versus Command (aka Apple key). Eterm wants "option". WindowMaker defaults to option, but is readily configurable. XEmacs wants command, but not option (irritatingly, it recognizes option as the bound-to-nothing "Alt".) * Solution 1: determine which major apps display inconsistency in selection of the "right" Modifier key when several are present. Lobby developers for change. * Solution 2: lose the potential flexibility of separate mod keys and remap option and command to identical keysyms at the X level. SUMMARY Overall, I find my system quite useful. I hope that near-future kernel revisions and/or XFree 3.3.5 fix the remaining video/sound problems. -- John
Re: Lombard status update
I wrote: > problems abound in accelerated mode. It's the most hosed > excuse for an X server I've ever seen. Hmm, this was much harsher than I meant to be. Shouldn't post past 2am. I should say that accelerated mode is quite usable.. just quirky at times. For the record, I just noticed that the redraw problem (e.g. with wmaker animations or other transient windows) seems to be sporadic. I'm getting some intermittent line drawing errors, but at this moment, redraw isn't a problem. No clue as to why. -- John
Re: where is xfree?
Sergio Brandano writes: > My X still does not work. Then I checked for an update. > I checked all the tree in potato but I could not find > any file, exept the single file xfree86-common_3.3.5-1. > Where is XFree? I noticed that the old xfree stuff vanished a coupled days ago when the Debian XFree 3.3.5 packages were announced. I note that the source debs have hit the mirrors... -- John
Problems since upgrading..
I've encountered two new problems w/ my Lombard PB since upgrades (including X 3.3.5) last night. 1) After X was upgraded to 3.3.5, the delete key is broken under X. xev shows keycode 59, keysym 0x "Delete". This is "delete forward", instead of the"delete backward" (BackSpace) keysym. While I can fix this manually with xmodmap, the default setting is definitely broken. 2) xterm is broken. An attempt to start an xterm ends abruptly with: xterm: fatal pty error 23 (errno=22) on tty /dev/pts/1 OTOH, Eterms work just fine. -- John
Re: Anacron still hangs
>[strace log elided...]
The strace is not terribly useful. I have already reported this
problem, including a working patch, and been ack'ed by the maintainer.
The fix was supposed to appear in 2.1-2, but hasn't shown up at all.
If you are still having this problem, submit it to the BTS, to get things
rolling on this fix.
The problem is that anacron expects the return value of getopt() to be
EOF instead of the proper libc6 compliant value of -1. I.e. there is
a while loop in main.c of the form:
while ((opt = getopt(...stuff..)) != EOF)
{
...
}
that hangs because getopt _never_ returns EOF under libc6.
I'm having to build from source debs, reapplying the patch each time
to get anacron working...
Alas, I can't post the patch at this time, as I'm in MacOS trying to
determine if a hard lockup problem is hardware or software by working
back and forth 'tween the two OSes. Right now, looks like Linux has a
serious problem...
-- John
Big Fat Bug
I have found an interesting bug. With a 256 MB DIMM installed in addition to the stock 64 MB module, 'free' incorrectly reports only 256 MB memory installed. 'cat /proc/cpuinfo' correctly indicates that 320 MB are installed... but apparently the info from /proc/meminfo is incorrect. This is Not Good, as the other possibly related symptom is that with the 256 MB module installed Linux becomes unusably unstable. The 100% kill test is to attempt to compile the kernel... but often the first command I run locks the system cold (e.g. I booted a few minutes ago, and my first 'ps aux' locked up.) I have currently removed the module, which permits me to write this message from Linux. Many of you will recall my problems getting Linux installed -- which I initially blamed on a bad DIMM. I have now gone through three such DIMMs from at least two different manufacturers. The most recent module is Simple Tech part no STA-MAIN/256. This is a PC100 rated module. (Which b.t.w. fits in the lower slot.) All modules have exhibited exactly the same symptoms: the Lombard becomes unusably unstable, subject to random hard lockups. Removing the module is the uniform method to fix the problem. The problem also persists when using only the 256 MB module. Nor does choice of slots for the 64/256 combination change matters. I had thought that MacOS was subject to the same problem, just less frequently but now I am not so sure. I have gotten a couple (two) random lockups, but I can't correlate these to any particular reproducable phenomena. Moreover, I can push the system very hard in MacOS and not get any crashes. MacOS correctly identifies the amount of RAM, and RAMometer has run its RAM test just fine for a total of several hours. (Good way to make sure the internal fan is working!) Other stress tests involving running realtime audio dsp, netscape, copying 500 MB of data over the LAN, then copying that locally, plus various other apps running in the cracks all at the same time for an extended period do nothing. Runs great. (wow.) This problem has also occured with a variety of kernels, compiled pre-gcc 2.95 and/or obtained in binary form from external sources. 1) Has anyone else noticed a problem similar to this, especially w.r.t. incorrectly reported memory sizes? It is possible that this could be a logic board problem with my Powerbook. 2) Where do bugreports for Vger kernels go to? Or else for Paul Mackerras' kernels (I don't have Mackerras' contact info presently..) -- John
Re: Anacron still hangs
Daniel Jacobowitz writes: > The most correct fix is to make opt an int. Indeed. My patch makes this necessary correction, but I admittedly hadn't delved into the headers once I got to the getopt() man page. The man page states nothing about EOF, just that -1 was the terminating return value. I corrected the type of opt en-passant, figuring that was also necessary, not realizing that EOF was effectively correct all along. Thanks for the heads-up. -- John
Re: Big Fat Bug
Brad Midgley writes: > Have you tried removing the 64M so you have exactly 256? Yes. Doesn't help one bit. I have tried all permutations of slots and DIMMs (64/256, 256/64, 256/0, etc.). If the 256MB DIMM is present anywhere, kablooie. -- John
Minor Problem Potpourri
A few glitches here and there: 1) the netscape packages require a missing netscape-base-4; netscape-base-4 doesn't exist (specifically, netscape-base-46 requires it). Is the dependency incorrect, or is a necessary package simply missing? 2) I'm getting a strange problem with console terminal emulation and the arrow keys. Sometimes the arrow keys emit ^[[A thru ^[[D. These codes are correctly recognized by dselect, make menuconfig, etc. (ncurses based apps, presumably.) But in recent muckings about with the X server and keyboard stuff, for some reason the arrow keys generate ^[OA thro ^[OD (that's big-Oh, not zero). Bash works fine with these, as does the vanilla d0 (erm, vt100) termcap entry when logging in remotely to systems not grokking the 'linux' terminal type. Any ideas on solutions for the above? -- John
Re: Big drives
Phillip R. Jaenke writes: > Dan Green wrote: > > We have a 37 gig drive in an Apple [... deletia ...] > > i'm more inclined to believe this is an error with ide before the > kernel. The problem is a non-ppc-specific kernel problem, caused by the failure to properly handle "ide growing pains." 8-) See the Large Disk HOWTO at: http://www.linuxdoc.org/HOWTO/Large-Disk-HOWTO.html The note on the Linux 65536 cylinder limit (aka the 33.8 GB limit) gives a specific explanation as to the nature of the problem: http://www.linuxdoc.org/HOWTO/Large-Disk-HOWTO-12.html#ss12.1 The above does indicate that the relevant patches were folded into the linux kernel as of 2.3.21. FWIW, I'm running a 2.2.13 SMP linux box with an IBM 34 GB drive by virtue of Andre Hedrick's IDE patches, available in incarnations for various stock kernel versions at: http://www.kernel.org/pub/linux/kernel/people/hedrick/ NOTE: I have no idea whether any of these patches will apply to any given version of a PPC linux kernel, much less work, nor do I know of anyone who has tried. That said.. good luck! 8-) -- John
Lombard console video initialization?
Has anyone running a Lombard Powerbook with a recent kernel (I'm running 2.2.14 now) managed to boot and get coherent console mode video with "No video driver" unchecked in BootX? (Or else via yaboot?) I've noted that atyfb acceleration doesn't get turned on if I check "No video driver"... but that's the only way console mode video is displayed properly. I've tried tinkering with a number of variations of the kernel video arguments over time, and none seem to matter. Thanks, John P.S. In what package does vmode reside? I believe that I inadvertently removed it at some point...
