Re: RC1 and Wallstreet Powerbook MIssing File In Install
Hi Bob, The gui installer starts fine. I select language and alternate mouse keys. It asks which usb module to try. I've tried kbddev, usbcore, and usb-hci. The installer kicks back an error and says the file /lib/modules/kernelstuff2.4**.cz is missing (sorry, I'm away from the machine and can't remember the exact error message) Well obviously there is a file [probably a whole module] missing, but Wallstreet's do not have USB... unless in a pcmcia card. On 8.2 you needed the usb HID for keyboard etc, I am not sure whether that is still true. Q.
Re: Compiler speed (fwd)
Hi,
On Thursday 16 January 2003 23:26, Quentin Mason wrote:
Hi,
I was wondering whether it was common knowledge or whether anyone else has
noticed how much faster it is to compile things on MandrakePPC over an
Intel machine ?
For example:
500 Mhz 1Mb L2 Cache, 384Mb RAM Wallstreet Powerlogix G3 upgrade, g++ 2.95
~3mins Dual 933 MHz Pentium 3, 1Gb RAM ,
g++ 3.2.1 2.25 Hrs
You should try this again using the SAME compiler version on both machines.
gcc 2.95 is known to run circles around 3.x on the same machine, as 3.x had a
full API change and is now waaay slower than gcc 2.95, though it produces
faster and smaller binaries.
That has not been my experience -- I have found that g++3.2.1 is somewhat
faster (and smaller memory+stackspace footprint whilst compiling) than
2.95.3. I found an older compiler on the office machines, and it did
indeed take longer than 3.2.1:
Dual 933Mhz P3g++2.95.3 3.5 hrs
Dual 933Mhz P3g++3.2.12.25 hrs
500Mhz Wallstreet g++2.95.3 ~3mins
I am compiling over NFS -- surely that cannot be the problem for such a
large file and long compile time ?
One of the advantages of the 3.x series is the -Q flag on the compiler
-- it tells you exactly what template instantiation or subroutine that it
is working on [watch how compiler-expensive all those #includes are --
what pre-compiled headers could do...], and spits out a time spent
analysis on all the steps (including optimisation). However I do not have
a g++3 for PPC [anyone have a {ppc,src}.rpm to coexist with system
compiler ?] to compare where the time is spent.
Q.
Compiler speed
Hi, I was wondering whether it was common knowledge or whether anyone else has noticed how much faster it is to compile things on MandrakePPC over an Intel machine ? For example: 500 Mhz 1Mb L2 Cache, 384Mb RAM Wallstreet Powerlogix G3 upgrade, g++ 2.95~3mins Dual 933 MHz Pentium 3, 1Gb RAM , g++ 3.2.1 2.25 Hrs This is on a ~ 1Mb file that I was trying to compile for my research without ANY optimisation options at all. The file was generated by some other code and consists entirely of statements like -1/3*sin(a+b+c)+2*sin(a-b+c). I discovered this by accident trying to do some work on my laptop at home over Xmas; I thought it would be fun to try several of these huge object files out on my laptop when they take so long on the machines at work. I have some other files to compile that take 24 hrs w/ 900Mb RAM + 10Mb stacksize in the office. I could imagine a factor of 2-3 difference, but to have around 100 including the processor speeds was _very_ surprising. Does anyone have any experience to compare to ? Thanks, Q. I hope you still have a job, Stew!
Re: G4 Upgrade for Wallstreet?
Hi, Has anyone tried any of the Wallstreet G4 upgrades with cooker-ppc? I have the BlueChip (Powerlogix) G3/500 in a Wallstreet 2/233. It runs _extremely_ hot but it does linux (linuxppc/2000 + Mandrake beta,8.0,8.2) fine. Powerlogix warn about sleep support and reduced max ram on their stuff for the G4 version which is significantly more expensive, and not IMHO worth the extra. AFAIK the graphics chip in the machine is going to hold back any multimedia app that actually has hand coded Altivec instructions (this is in linux). The wallstreet has no fan and the case is often very very hot to the touch. There was a batch of processors which burnt out and had to be replaced (I got free upgrade from G3/466 to G3/500 after 3 months waiting for new chip). I guess they had too many problems cos I do not see the G3/500 available on their website now. Try phoning them to ask about linux support -- in the past they have been very explicit about their support and their monitoring tool has a check box for BootX support. So far, I have: One solid report that the BlueChip will not work with Yellow Dog, An unproven claim that the Newer Nu Pwr surgical upgrade will work with *Linux, No info on the Sonnet Crescendo/WS G4. Good luck, Q. PS: expat from UK in Cornell, NY, USA.
Re: OT: (but I do not know who ask for): buildin an RPM from scratch
I will write a mini setup program as Microsft does ;) Up to now I just echo out the instructions and the license (with a line that sounds like 'using the software presumes you have accepted the license terms at all Intel's free compiler, Mathematica, and maple take the following approach: write a simple script that reviews the license, asks/determines which parts /things to install, unpacks the appropriate rpm's and installs them -- relocating to users chosen path if that is an option. These can be run from a script by catting in the required answers: install_ace_package my_options. Generally such packages are add-ons and go in /opt and are individually installed. On the other hand if you have used rpm to mark your license type and have included it in the distro then arguable an installer is/should be aware of it. VMware + staroffice(?) take an alternative approach -- rpm's to install but a first-time script to be run by users to setup the network daemons etc: vmware-config.pl and soffice. Alternatively you can have the program auto-detect different features like SSE1/2 MMX support for multimedia libraries. Hopefully you should be inspired by one of these examples. Good luck, Q.
Re: Mount ext2 partition in OSX or OS9
Does anybody know how to mount a linux partition on the Mac desktop. And even better, to do drag and drop in there. Cannot be done in OS9. It used to work with IDE drives and MountX but that stopped working ~18 months ago when ext2fs was updated and Ben stopped supporting it. The next best thing is to just use MOL and export linux with netatalk -- then you are running MacOS and can access all your linux files. If you really desperately need to get some data from the linux side then you should boot linux-rescue from the CD and ftp somewhere, or use an HFS drive (NB not hfs+) as a transfer partition. OSX ?? Q.
Re: mol in 8.2?
Hi, Doesn't sound like the kernel is compiled with the mol module in 8.2, am i right? I have to run mol with the -a option, it launches up, I have the mac window opening, i can see the extensions loading, then the mac desktop appears and it stays glued here showing the desktop with no menu and no icons. Any idea? Is there an 2.4.18 kernel with mol option available somewhere. Should I recompile it?, Which version of MacOS do you have ? There is a known problem with MOL0.63 and MacOS 9.2.1. Also if you install MacOSX on the same partition then that is much more difficult -- you must bless the regular MacOS system folder after everytime that you use OSX, until Samuel completes OSX support (almost there, apparently). The other question is what extensions are you trying to load ?? There are some things which MOL still cannot handle, so try booting with extensions off [press shift during boot]. If you have managed to get as far as you did then the kernel module probably should not matter -- but either way I heartily recommend getting the most recent release as there some significant speedups to window mode and a new keymapping tool for example. More help to be found on the mol-user list, which recently switched over to Terra-soft: http://www.maconlinux.net/lists/index.html Generally the MOL-in-a-window is much slower than in a console -- try editting the /etc/molrc for that. Q.
Re: bug in talk-server + /dev/pts/?
Hi, # Uncomment the following if you want to set the group to tty for the # pseudo-tty devices. This is necessary so that mesg(1) can later be used # to enable/disable talk requests and wall(1) messages. REGISTER ^pty/s. PERMISSIONS -1.tty 0600 REGISTER ^pts/.* PERMISSIONS -1.tty 0600 which suggests that by default devfsd is trying to do the right thing. Well then I'm not sure why the permissions are wrong on your machine. devfsd is properly setting the permissions for me. Have you tried deleting the contents of your /lib/dev-state directory and rebooting? It shouldn't be trying to do any thing with the tty's but *shrug* it's all I can think of. It turned out that devfs was not running; on oldworld BootX login's chkconfig --on devfsd is useless as the partitions are not mounted with devfs [no /dev/.devfsd file]. One must explicitly add devfs=mount to the kernel arguments. This has the added benefit of making the tun module do something useful for MOL [though it gets the link wrong /dev/net/tun - misc/net/tun should go to /dev/misc/net/tun], and the evdev module allows /dev/input/keyboard for use with pbbuttonsd, which is nice for us laptop users. Upshot: use devfs to get some advanced functionality, you will need kernel argument devfs=mount to do this. Q.
Re: bug in talk-server + /dev/pts/?
Hi, On Mon, Apr 15, 2002 at 02:08:21AM -0400, Quentin Mason wrote: I usually have mesg y in the interactive section of my login scripts so that people may talk or ytalk to me when I am interactively using that login [cf mesg n for remote batch jobs]. On my new vanilla Mandrake 8.2\beta2 install mesg y fails in bash, tcsh etc at all times. The error is: mesg: error: tty device is not owned by group `tty' and sure enough /dev/pts/? is owned by user.user not user.tty; mesg n is fine however. I'm pretty sure this is a symptom of msec being run at a high security level. Try reading man msec and it's accompanying documents. There is a way to override the permissions that is causing your issue. Or you can just lower your security level. I have mine on level 3 and mesg y and mesg n works just fine. You can find out your current level via: grep SECURITY /etc/sysconfig/system I am on level 2 [internet access is through a firewall], so it is lower than yours. I could not find any info in /usr/share/doc/msec-0.19, /var/lib/msec/security.conf or any of the perms files in /usr/share/msec, man msec or man mseclib Incidently my vanilla /etc/devfsd.conf file contains: # Uncomment the following if you want to set the group to tty for the # pseudo-tty devices. This is necessary so that mesg(1) can later be used # to enable/disable talk requests and wall(1) messages. REGISTER ^pty/s. PERMISSIONS -1.tty 0600 REGISTER ^pts/.* PERMISSIONS -1.tty 0600 which suggests that by default devfsd is trying to do the right thing. Incidentally the talk issue is a known issue of running at a higher security level and it was determined that it was an acceptible problem because very few people that use that feature also run at high security levels. Fair enough. There is still the problem with 2 copies of the the config file though. Q.
bug in talk-server + /dev/pts/?
Hi,
I usually have mesg y in the interactive section of my login scripts so
that people may talk or ytalk to me when I am interactively using that
login [cf mesg n for remote batch jobs].
On my new vanilla Mandrake 8.2\beta2 install mesg y fails in bash, tcsh
etc at all times. The error is:
mesg: error: tty device is not owned by group `tty' and sure enough
/dev/pts/? is owned by user.user not user.tty; mesg n is fine
however.
Furthermore, I see in /etc/xinetd.d/{,n}talk that the group is set to
tty [these are both provided by talk-server-0.17-6mdk, this package
seems to have talk file links for every ntalk file + an additional
/etc/xinetd.d/talk config file]
I had thought that changing disable=yes to no and SIGHUP'ing xinetd would
enable talk support but it does not seem to work [connection refused]. I
tried restarting xinetd, and sure enough talk is not removed in the
logs and should therefore work. It was not until I did the same for ntalk
that I got the refusing messages line. When I further chown q.tty
/dev/pts/? talk started to work. I am not familiar with whichever package
is setting up the ownership of /dev/pts sockets and whether such a change
is a security risk.
talk-0.17-6mdk.src.rpm explicitly makes two xinetd.d conf files in the
server rpm, but the changelog was last editted 1 year ago. This is true
on x86 too. The redhat HPC nodes in the office only have one copy.
My inclination is a bug in talk-server spec is duplicating the files so
that both /etc/xinetd.d/{,n}talk have to be editted to get the same daemon
/usr/sbin/in.talkd-/usr/sbin/in.ntalkd to run, and that something is
getting the wrong ownership of /dev/pts/?.
Anyone have ideas what is going on/supposed to be going on?
Q.
I would like to thank Stew for all the work that he has put into the
distro. My Wallstreet is back in business after some horrible keymapping
problems in 8.0 [I lost the arrow keys, and had utterly different keycodes
in X and console], pcmcia is looking good, XFree4.2 is wonderful,
emacs-nox-21 is ace for remote console editting. Thanks again.
Re: Begging for help!! Can't start my x86 Mandrake 8.2 box
Hi, My x86 machine, running Mandrake 8.2, is stuck in a loop... everytime it boots itsays that /dev/hda1 was not unmounted cleanly, and asks if I want to test the drive... I can'tsay NO! It can't check because the drive is mounted, So it then drops me into a rescue login. Irun the e2fsck, and exit from the prompt and the machine reboots... but I get the same thingevery boot!! What is on /dev/hda1 ?? If you have the installation CD's then boot in them in rescue mode, and try a careful manual fsck of that partition. The rescue mode can fsck better because it does not have to mount that partition first. I assume that you are still booting with the same kernel, have not changed grub or lilo or messed with fdisk at all... What kind of errors is e2fsck reporting ? It is in principle possible that the hard-drive has started to go... I lost 9 months work from my laptop once which was not fun. If your essential data is not on hda1, you can atleast boot the CD's in rescue mode, mount the other partitions and copy the data somewhere else. I have had to do this before -- if you can mount /usr or something to get to scp/ftp or whatever. If you are really desperate for a single file, then even grub can list the files contents for you [command-line cat /...]!!! Good Luck, Q.
Re: 8.2 beta report - several problems
What about Lyx/Latex? Aren't they supposed to be installed? What package do I have to chose in order to get them? I chose the scientific one, office etc. but Lyx/Latex don't seem to have been installed. Get tetex-* for LaTeX. I am not sure where they are in an install, but those are the ones that provide xdvi, latex, pdflatex, dvips, etc. You may also find a2ps, mpage and psutils useful. Others are gv, acroread, ggv [gnome-gv] and kdegraphics [provides kghostview]. Auctex is a great emacs LaTeX mode addition, and you should check out preview-latex [at sourceforge] for something similar to lyx. As for power management, i can't get it working properly yet.(i can only get suspend) I am testing the latest pmud, but it needs some work. Don't worry about any referance to APM, thats for i386. http://linuxppc.jvc.nl/RPMS/pmud-0.10-1.ppc.rpm It also has a sourceforge page now. It may not work perfectly with a brand new ibook2, but is great for Wallstreet/Lombard/ibook. Best combined with gkrellm-pmu IMHO. Have you tried typing snooze into a terminal ? The older pmud's had a problem waking up after having a CDROM mounted but that should have been fixed with wakebay. Pmud tends to work best with BenH kernels -- there is an option in .config to enable support for it. It used to need the loopback net device which is why it sometimes fails on boot if you have not ifup-ed the network; I am not sure whethre this is still the case. What does it say when you try and start it with debug, no detach pmud -d on the terminal and in /var/log/messages ? Q.
warning: KDE control centre
This is just a warning to all of you about the KDE control centre,
specifically the keyboard selection part.
I have been enjoying KDE 2.2 for a while, but decided the time had come
for some customisation. I have an English Wallstreet powerbook, and the
option key was not working. I saw that I could change the keyboard
mapping in the control centre. This was exciting and I duly selected the
English keyboard. This was a disaster -- none of the keys returned the
correct letter. Oh well. Imagine, however my chagrin when I switched
back to the default, only to find that the default was a US keymap and
again a PC keymap so all of the keys returned the wrong letters. The
problem is that their default is not the default of the distro. Ahh!
Fortunately I had lots of text on the screen at the time, and was able to
use the mouse to copy individual letters to do the following:
su -
password:
echo 1 /proc/sys/dev/mac_hid/keyboard_sends_linux_keycodes
which immediately fixed everything. This allows you to use PC keymaps.
Indeed now the option key works, although the arrow keys are defunct
[ctrl-{f,b,p,n} work though]
So this is rather a gotcha -- do not mess with the keymaps in the KDE
control centre!!
Note that to make the change permanent it must be added to
/etc/sysctl.conf
The part I am not sure about is whether this change would have affected
all users -- I did not feel like trying it out and being stuck entering a
ROT13-like password. Loggin in remotely and su-ing would probably have
worked too.
Q.
Re: CDROM Question
Hi Terrence, I've burned a CD which contains a bunch of mp3s. When I put the disc into the CD/DVD drive I'm able to cd /mnt/cdrom and browse the disc. When I put an audio CD that I have purchased (Led Zeppelin II) into the CD/DVD drive and try to do the same thing I get this in Konsole Mandrake is automounting the CD for you, but you cannot mount an audio CD. Data CD's contain file-system information much like hard-disks. There are HFS CD's, ISO, Joliet etc. There are also audio cd's, but you cannot mount them like file-systems. I think the data [and error correction in particular] on them is done differently. If you want to listen to it use any number of cdplaying linux applications, if you want to copy the audio use a cdburning/ripping application or go commandline with cdparanoia. If you use KDE then kscd is the kde cd-player. xplaycd, xmms etc are alternatives. For ripping of audio tracks into wav/mp3/ogg files then look at something like grip (X-based), lame, oggtools etc. HTH, Q.
Re: PMac 7300 - Install Issues
On Thu, 4 Oct 2001, James Meade wrote: For about a week now, I've been trying to get Mandrake to install on my G3-upgraded 7300 system without much luck. So far, I've been reduced to playing musical kernels to get it to recognize my SCSI hard drives, let alone the CD-ROM drive. Can someone provide a basic walk-thru on how to get this to install? IBM 2.1GB SCSI on SCSI bus 0, ID 0 MacOS 9.0.4 BootX 1.2.3 Mandrake 8/PPC distro CDs I have 2 ideas for you. Move the IBM away from SCSI id 0. There were some reports on the linuxppc-user list that linux does not like having harddrives at id=0 on those old machines. I thought that BootX did not work reliably with MacOS 9.0.x but that may only be on New World Machines. What specifically is the problem ? Perhaps some detail such as I boot with BootX from SCSI HD trying various kernels, the system gets to the such-and-such stage and then complains about ..., the last message on VT2,3 was ... Good luck, Q.
Re: cpu info
Hi, processor : 0 cpu : 750 temperature : 0 C clock : 195MHz revision : 2.2 (pvr 0008 0202) bogomips : 878.18 I am wondering if the CPU is running at 440Mhz or just 195MHz? Is it because the CPU information was burnt in ROM and cat /proc/cpu didn't really test the CPU speed? cat /proc/cpuinfo only gives what the machine thought it was during the startup. I do not know how that code works. All the PowerPC machines I have ever seen or heard off report BogoMIPS as 2x clock in MHz. So my guess is that it is running at 440 and lying to you, possibly the ROM thing you suggest. I have upgraded by replacing the processor and I get the correct clock and 2x for bogomips. Is the upgrade card the kind where you actually replace the processor ? The other thing to try is a simple benchmark program and see how many FLOPS you get -- if more than 200 MFLOPS (you will need a simple benchmark and -O3 to compile) then it must be the 440. [On an intel you can do better than the clock using SSE instructions] BTW machines often feel slow when all it is is an unaccelerated graphics card... So if you are suspicious you are not getting everything then try running something without X to see how fast it is, or work on accelerating the X server. HTH, Q.
Re: Silly Question about Ping
Hi, is there an RPM on the CD that includes ping? rpm -qf `which ping` on an x86 redhat machine at work gives iputils-20001010-1 rpmfind.net would probably also have helped you out here. Good luck, Q.
Re: M4753 Wallstreet
Does Mandrake 8.1 PPC CD boot on a M4753 Wallstreet upgraded with a Powerlogix Blue Chip 500 Mhz? Mandrake 8.1 is not out for PPC yet AFAIK. Linux works fine on my wallstreet II [512k L2 cache] w/ powerlogix blue chip 466 MHz [now 1mb L2]. I use MacOS 8.6 and BootX. MacOS can choose to view lcd or external monitor which linux picks up but I know of no way to switch once in linux. Mac-on-linux works well. I do not know whether the CD will auto-boot -- you may need to set up BootX instead but once you install linux will work just fine. I know another guy who has used linuxPPC on his wallstreet + powerlogix 500Mhz. Good luck. Q.
Re: ppc install freeze on beige g3
Hi Paul, video=atyfb:vmode:20,cmode:32 This style. Also, how do I get into text install mode? boot arg = install-text Will getting X up and running after a text install be possible? Yes. If you have made the appropriate sacrifices and are feeling lucky, log in and try startx. Unfortunately, I do not know what the settings are for an ATI Radeon. I use atyfb for the ATI Rage Mach 64 -- is it not atyfb128 or something similar for the radeon ? Q.
Re: Problems installing PPC 8.0 on PowerBook G3 (New World/400/128M/6GB/DVD)
On Mon, Sep 10, 2001 at 09:29:22AM -0700, Sean Fay wrote: So if I only wanted to boot linux I could remove all partitions using MacOS Drive Setup and then have mandrake create the partitions it wants?? Yes. Can you really delete ALL the partitions -- including the random 4 or 5 apple ones on the front ? I would leave those if I were you. I thought that open firmware needed something to tell if there is a mac partition map and which partition to boot from etc... AFAIK only MacOS utilities will initialise these driver partitions, so you will need to boot from a MacOS CD and initialise the whole drive as free space not HFS nor HFS+; this should put the driver and the other bizarre things on the front, whilst leaving the free space for the Mandrake installer. The best way to check is to use pdisk -- it is a program like fdisk, and runs in either linux or macos. Q.
