Re: Floppy from sparc on x86
On Wed, 20 Dec 2000, Rafael E. Herrera wrote: I have tried before to use the one from an IPC without success. The last responses I got were: Matthew Haas wrote: On Sat, 25 Nov 2000, Rafael E. Herrera wrote: Perhaps I didn't phrase the question right. How do I get the floppy to work on a PC? I know of the 'eject' and 'floppycontrol'. Any comments? I heard from somewhere that Sun machines use the alternate ID (ie B: instead of A:--- they don't use the twist in the cable) and that you must plug the drive in before the twist in order to get it recognized as A: (I've never personally tried it, but I have heard that people have had success). Alain Knaff [EMAIL PROTECTED] wrote: I pulled a floppy drive from an old SUN IPC box and I'm trying to use it in my PC. It has a motor that ejects the floppy and I'd like to use that feature. I've tried to get info from the web about it, but have had no luck. I also installed the fdutils package to use the floppycontrol utility. I've tried several setting in the BIOS and in the floppy to get it to work, no dice. If you have any pointers I'd be grateful. Rafael For the moment, this feature is only supported for Linux on Sparc (Sparcstation usually come with floppy drives with an autoeject mechanism). Do you have any documentation with your drive, the name or number of the wire used to trigger the eject might be useful I think the last one was from the author of floppycontrol. If you get it to work let me know and Alain, too. I got the IPC for $12 at the GoodWill. The funny thing is that the floppy drive (Sony MFD-17W-P1) sale for $85 refurbished at www.rapid-tech.com, isn't it funny? HEH! I got my IPX probably from the exact same Goodwill last october. Was over there visiting a friend at CMU. (noting your deleted SIG) Bought it for $10 (there was a guy who was going to buy it just for the hard drive - until I told him I could put Linux on it), took it to CMU with my friend, and loaded it up. I'm using it as a testbed DHCP server right now. My floppy drive just died a few weeks ago; OpenPROM floppy diagnostics fail, won't boot, etc. I've been unable to get it working on a peeci either. I'm also considering sticking in a SCSI-interface LS-120. You can get them from Winstation (www.winstation.com I hear), and they're using some kind of low-profile SCSI-ATAPI protocol converter retrofitted onto the existing ATAPI drives. Also, they're supposed to be coming out with a 240MB model, so the 120's are at End of Life. I'm wondering if the IPX would be able to boot off a 1.44MB floppy in it as a SCSI device.
partial success with external modem on IPX
I'm using it right now with serial ports jumpered to RS232, 38400N81, Mac modem cable, AND modem handshaking disabled. It's a bit slow, but -test11 eats filesystems on my normal gateway under high ethernet load. :/ Does anyone have the kernel patch to drive the Zilog 8530 at 64k? I have been historically unable to locate it.
cross-compiler mini-HOWTO 0.0.1
Finally I have my mental notes into a somewhat coherant form. I welcome any suggestions, pointers, etc. This is my first time writing an HOWTO. I had planned to write a little script to perform the steps described here, but I ran into a little policy-esque technical snag that I needed some expert advice on. However, it's been about a week with not even a reply to the questions I asked on debian-devel. Perhaps I could find the advice I seek here? HOWTO is attached. [document 0.0.1] [copyright 2000, Beiad Ian Q. Dalton. Permission to distribute and maintain this document in the same manner as the HOWTO collections is hereby given.] This HOWTO describes a quick and dirty way to produce a minimal cross-compiling system. A cross-compiler is used to compile software on a fast machine of one archetecture for a slower machine of a different archetecture. A good example of this is compiling for a 68k Mac or a Sparc IPX or SS10 on a recent Pentium II system. This HOWTO currently assumes that you will be performing all the steps as root. You need: A fairly fast computer with Woody installed on it. Potato should work as well, but I don't have a system I can test for it. Access to Woody's source package archive. It's probably a good idea to do all this under /usr/local/tmp or /var/tmp, or a scratch partition, if you don't already have your own set-up to compile local packages. Download and unpack the following source archives: (apt-get source is best if you have an Internet connection) binutils 2.10 gcc-m68k-linux 2.8.1 If you're going to build for a Sparc target, you may also want to get the sparc-utils package for the elftoaout converter and the tftpboot stuff. Now download 'lsdo', 'libc6', and 'libc6-dev' binary debs for your target archetecture and distribution. Now, install dpkg-cross, if you don't have it already. You will must to edit /etc/dpkg/cross-compile and change 'crossbase' to be '/usr' instead of '/usr/local'. gcc 2.8.1 will not build successfully if you do not do this. Use dpkg-cross to install the ldso, libc6, and libc6-dev packages that you download, one package per invocation. Now, cd into the binutils source directory and type 'debian/rules TARGETS=dest arch binary' When the build has completed successfully, you should have 'binutils', 'binutils-dev', 'binutils-doc', 'binutils-multiarch', and 'binutils-dest arch'. At a minimum you should install the 'binutils', 'binutils-dev', 'binutils-multiarch' and 'binutils-dest arch' packages. Now, you should be ready to build gcc. IFF your target arch is not m68k: cp -a gcc-m68k-linux-2.8.1 gcc-target arch-linux-2.8.1 With a text editor (sed should work fine here) change all instances of 'm68k' to 'target arch' in all files in the gcc-target arch-linux-2.8.1/debian/ directory. Now, cd into gcc-target arch-linux-2.8.1 and run 'debian/rules binary'. Assuming this compile succeeds, install the gcc-target arch-linux deb. Known problems: You need gcc 2.91 in order to compile the 2.4.0 test kernels. If the gcc source package could be extended in the same way that binutils was, we could obsolesce the gcc 2.8.1 package and Be Happy. Currently, if you don't have 'crossbase = /usr' it will Fail To Build. This is because the Makefile in gcc 2.8.1 isn't smart enough to consider that the header files and the binutils might not be in the same superdir, and I'm just not up to trying to Make It Work. Besides, making 2.91 work would be a better use of effort. ; It also doesn't appear possible to make a script that could be packaged ; and added to Debian, currently. The sticking point is the issue with ; dpkg-cross. In order to work, the script basically has to make the admin ; change the value of crossbase, or be allowed to override it on the ; dpkg-cross commandline. I asked about this on debian-policy, also ; asked the binutils maintainer in private email, but have heard ; nothing. ; Personally, I think overriding the admin's decision isn't the best ; idea, but it is the best thing to make the script do what it's ; supposed to, given the current state of the packages it needs.
Re: Question about a Sony MFD-17W-P1 floppy disk drive.
I don't think you can, actually, but I'm not an authorative source. On Sat, 25 Nov 2000, Rafael E. Herrera wrote: Perhaps I didn't phrase the question right. How do I get the floppy to work on a PC? I know of the 'eject' and 'floppycontrol'. Any comments? -- Rafael -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: sunlance.c - network driver for sun4m
Yah. Say, I'm working on a little HOWTO and hopefully an automated build script for cross-compiling kernels. I go Intel-Sparc. I'll make another post when I get the basic instructions complete. On Tue, 21 Nov 2000, Ian Chilton wrote: Hello, and an other faq tells that i have to convert an elf kernel to aout with the elf2aout tool. but the tool tells me that it is not an elf file. Yes, I had this toodo this: mv vmlinuz vmlinux-elf.gz gzip -d vmlinux-elf.gz elf2out -o vmlinux vmlinux-elf Then use vmlinux to boot from.. Bye for Now, Ian \|||/ (o o) /---ooO-(_)-Ooo---\ | Ian Chilton(IRC Nick - GadgetMan) ICQ #: 16007717 | |-| | E-Mail: [EMAIL PROTECTED] Web: http://www.ichilton.co.uk | |-| | I used up all my sick days, so I'm calling in dead. | \-/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
cross-compiler update
(can't sleep. Clowns will eat me) Need to compile and install binutils-sparc (is not built from the binutils package for some reason unknown to me right now). Then install dpkg-cross, and import ldso, libc6, and libc6-dev debs from Sparc. Then symlink /usr/local/sparc-linux/[include|lib] into /usr/sparc-linux/. Then tweak gcc-m68k-linux into building for sparc and depending on binutils-sparc instead of binutils-sparc-linux. Assuming this works, we would somehow need a libc6-dev crosscompile package built for each supported arch and a few other minor package fixes. I think I'm starting to drift off-topic.. Sometime tomorrow I'll post a compressed log of kernel-source-2.2.17 compiling for people to look at. I'm on a dialup connection for the time being though. Then I have a question of testing the kernel. How should I write a boot floppy with the kernel, assuming it builds?
Sparc cross-compiler update
Looks like my FDD is bad. I test it in prom (test floppy) and it reports it's unable to read the first sector. So, I need to learn how to build or edit a tftpboot image.
Still trying to build an i386-sparc cross-compiler
I have been having an extremely bad time with this. I can't even get the i386-m68k cross-compiler (gcc-m68k-linux) to compile. If it did compile, I could just edit the control and rules files to make it build to sparc. I believe I've had two open bugs against the package so far. And I'm very interested in how the heck it compiles in the Potato or Woody autobuilder, or if it even has been. MD5sums for the Slink and the Woody version match, so I'm betting the Woody and Potato versions were built on Slink. Meanwhile, would someone who has built and uses a cross-compiler be willing to help me build one from scratch?
Re: modem weirdness was: minor error in ppp chatscript
Hrmmm. I must have been more asleep than I thought.. :/ I am dialing out to my secondary ISP. When modem is connected to the i386 box, I connect to the ISP, press 'enter' and get a login prompt. When modem is connected to the IPX, I connect to the ISP, press 'enter' and fail to get a login prompt. In both cases I am dialing out with the modem command ATDTphone # On the i386 I am using a Belkin serial modem cable. On the IPX I am using a (I THINK Interex) Mac modem cable. On Mon, 7 Aug 2000, Leonardo Rochael Almeida wrote: Hi, It is not clear what your problem is. Where are you connecting? (i.e.: what is on the other side of the phone line?) Are you trying to dial into your IPX or out of it? Whose minicom seems locked? the one on the IPX dialing out or the one on the machine you are trying to dial into your IPX? In order to dial into your IPX you have to have a getty of sorts listening on the serial port. Debian's /etc/inittab usually has an example of how to do that, but it is usually commented out. As for the alarm problem, I have no clue. Regards, Leo On Sat, 5 Aug 2000 [EMAIL PROTECTED] wrote: I'm having a different but certainly related problem now. I can talk to the modem fine from within minicom, but my terminal session appears to lock after the modem negotiates connection to the remote and returns the CONNECT string. After connecting I press enter and nothing at all happens until I'm disconnected a minute or so later. However, I connect the modem to my intel box and dial in, I press enter and am greeted with a login prompt. I don't understand what's happening here. Could the cable be bad in some subtle way, or do I need some special init string? I'm just lost right now. On Fri, 4 Aug 2000 [EMAIL PROTECTED] wrote: I'm curious why I would be getting a timeout and alarm at this point of chat. The modem is an external USR Sportster X2 flashed to v.90 code, and the cable is a mac serial crossover cable. I can talk to the modem through minicom and do not see any problems there. The port settings is 38400N81 with hardware flow control. The computer is a Sparc IPX running Potato. Aug 4 22:02:16 runescribe chat[27627]: abort on (BUSY) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO CARRIER) Aug 4 22:02:16 runescribe chat[27627]: abort on (VOICE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIALTONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIAL TONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO ANSWER) Aug 4 22:02:16 runescribe chat[27627]: send (ATZ^M) Aug 4 22:02:16 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: alarm Aug 4 22:03:01 runescribe chat[27627]: send (AT^M) Aug 4 22:03:01 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: AT^M^M Aug 4 22:03:01 runescribe chat[27627]: OK Aug 4 22:03:01 runescribe chat[27627]: -- got it Aug 4 22:03:01 runescribe chat[27627]: send (ATX4DT4286535^M) Aug 4 22:03:01 runescribe chat[27627]: expect (CONNECT) Aug 4 22:03:01 runescribe chat[27627]: ^M Aug 4 22:03:03 runescribe chat[27627]: ATX4DT4286535^M^M Aug 4 22:03:03 runescribe chat[27627]: NO DIAL TONE Aug 4 22:03:03 runescribe chat[27627]: -- failed Aug 4 22:03:03 runescribe chat[27627]: Failed (NO DIAL TONE) Aug 4 22:03:03 runescribe pppd[27620]: Connect script failed
minor error in ppp chatscript
I'm curious why I would be getting a timeout and alarm at this point of chat. The modem is an external USR Sportster X2 flashed to v.90 code, and the cable is a mac serial crossover cable. I can talk to the modem through minicom and do not see any problems there. The port settings is 38400N81 with hardware flow control. The computer is a Sparc IPX running Potato. Aug 4 22:02:16 runescribe chat[27627]: abort on (BUSY) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO CARRIER) Aug 4 22:02:16 runescribe chat[27627]: abort on (VOICE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIALTONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIAL TONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO ANSWER) Aug 4 22:02:16 runescribe chat[27627]: send (ATZ^M) Aug 4 22:02:16 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: alarm Aug 4 22:03:01 runescribe chat[27627]: send (AT^M) Aug 4 22:03:01 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: AT^M^M Aug 4 22:03:01 runescribe chat[27627]: OK Aug 4 22:03:01 runescribe chat[27627]: -- got it Aug 4 22:03:01 runescribe chat[27627]: send (ATX4DT4286535^M) Aug 4 22:03:01 runescribe chat[27627]: expect (CONNECT) Aug 4 22:03:01 runescribe chat[27627]: ^M Aug 4 22:03:03 runescribe chat[27627]: ATX4DT4286535^M^M Aug 4 22:03:03 runescribe chat[27627]: NO DIAL TONE Aug 4 22:03:03 runescribe chat[27627]: -- failed Aug 4 22:03:03 runescribe chat[27627]: Failed (NO DIAL TONE) Aug 4 22:03:03 runescribe pppd[27620]: Connect script failed
modem weirdness was: minor error in ppp chatscript
I'm having a different but certainly related problem now. I can talk to the modem fine from within minicom, but my terminal session appears to lock after the modem negotiates connection to the remote and returns the CONNECT string. After connecting I press enter and nothing at all happens until I'm disconnected a minute or so later. However, I connect the modem to my intel box and dial in, I press enter and am greeted with a login prompt. I don't understand what's happening here. Could the cable be bad in some subtle way, or do I need some special init string? I'm just lost right now. On Fri, 4 Aug 2000 [EMAIL PROTECTED] wrote: I'm curious why I would be getting a timeout and alarm at this point of chat. The modem is an external USR Sportster X2 flashed to v.90 code, and the cable is a mac serial crossover cable. I can talk to the modem through minicom and do not see any problems there. The port settings is 38400N81 with hardware flow control. The computer is a Sparc IPX running Potato. Aug 4 22:02:16 runescribe chat[27627]: abort on (BUSY) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO CARRIER) Aug 4 22:02:16 runescribe chat[27627]: abort on (VOICE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIALTONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO DIAL TONE) Aug 4 22:02:16 runescribe chat[27627]: abort on (NO ANSWER) Aug 4 22:02:16 runescribe chat[27627]: send (ATZ^M) Aug 4 22:02:16 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: alarm Aug 4 22:03:01 runescribe chat[27627]: send (AT^M) Aug 4 22:03:01 runescribe chat[27627]: expect (OK) Aug 4 22:03:01 runescribe chat[27627]: AT^M^M Aug 4 22:03:01 runescribe chat[27627]: OK Aug 4 22:03:01 runescribe chat[27627]: -- got it Aug 4 22:03:01 runescribe chat[27627]: send (ATX4DT4286535^M) Aug 4 22:03:01 runescribe chat[27627]: expect (CONNECT) Aug 4 22:03:01 runescribe chat[27627]: ^M Aug 4 22:03:03 runescribe chat[27627]: ATX4DT4286535^M^M Aug 4 22:03:03 runescribe chat[27627]: NO DIAL TONE Aug 4 22:03:03 runescribe chat[27627]: -- failed Aug 4 22:03:03 runescribe chat[27627]: Failed (NO DIAL TONE) Aug 4 22:03:03 runescribe pppd[27620]: Connect script failed -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: serial console problem
I have the same problem when I boot my serial console. I just have to remember to umplug the cable until after I start minicom. BTW, 'minicom -o' will skip the modem init for you, if you someday switch to Linux on your x86 box. On Sun, 18 Jun 2000, Ben Collins wrote: On Sat, Jun 17, 2000 at 08:59:39PM -0700, Moe wrote: Hello all I have a debian 2.2 installed on a sparc 1+. I connect via the serial port from my windows machine. When I shutdown my windows box, then restart it I go to the windows serial console and I see the '' meaning that the sparc has crashed or windows console(hyperterm) sends a break signal to my sparc. When the PC boots (powercycle) it inits the serial port, and sends a break, IIRC. Expect it. You can disable this on the sparc with: echo 0 /proc/sys/kernel/stop-a But that means you have to revert that with a 1 when you want it to work :) -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---' -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Unable to compile; sys/io.h not found (fwd)
I'm going to be recompiling libc6 as a verification pass. The powertweak people are saying we should have a sys/io.h, but it looks like it's something specific to arches that use ISA. -- Forwarded message -- Date: Tue, 06 Jun 2000 17:28:18 -0700 From: Sudhakar Chandra [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Re: Unable to compile; sys/io.h not found Dave Jones proclaimed: [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: My system doesn't seem to have a sys/io.h under /usr/include. Then it would appear to be borken. I'm running Debian potato prerelease. Kernel 2.2.15, libc6 2.1.3, with standard libc headers. Looks like a debian-specific problem. Thaths?, Anyone ? If you've got the devel stuff installed, then it maybe that the packager screwed up, and missed a file ? See if you sys/io.h is listed in /var/lib/dpkg/info/libc6-dev.list It is is not, then the package maintainer messed up. If it is, your system has somehoe landed in an inconsistent state. Download the libc6-dev deb again and reinstall it. Before reinstalling it do a 'dpkg --contents libc6-dev*.deb | more' to see if sys/io.h is part of the package that you downloaded. S. -- Now, let's all turn around and pay attention to me again. -- Homer J. Simpson Sudhakar C13nhttp://www.aunet.org/thaths/Lead Indentured Slave
Re: Boot partitions
Yes, Potato install will be a lot less of an headache to get up and running. And from memory (I have an IPX also, set up a /boot partition since I replaced the OEM drive with a 9GB -- no idea if the later IPX prom has the 1GB problem) you'll have to boot manually from silo the first time, then edit /etc/silo.conf (And I probably should move mine to /boot also) On Wed, 31 May 2000, Ben Collins wrote: I am a Primary School teacher whose school has inherited a SPARC IPX running SunOS 4.1.3U. I am new to SPARC's but not to Linux. I've installed the base package of SLINK 2.1 (kernel 2.0.35) in a '/boot' partition '/sda1' (28MB). I previously used to fire up the Sun with the following from the new prompt: This is where the problem lies. Silo looks for the image on the boot partition (very unlike a PC, in which lilo references an disk block as opposed to being able to read the filesystem like silo can). You can either boot with init=/bin/bash and edit /etc/silo.conf manually, or do an install with just one partition (no need to seperate partitions if you don't have to). I also suggest going with a potato install as opposed to slink.
IPX and install to large hard drive
I remember hearing that some of the older Sun machines could only boot from the first GB of the disk. Is there any need to have a separate partition at the beginning of the disk for /boot with the kernels and silo in it? Installation was smooth. The only problem came when I was asked to choose the hostname. ;) I did make a 16MB partition starting at offset 0 for /boot, and silo didn't want to find the kernel image on reboot.
Re: IPX and install to large hard drive
Question is, do I even NEED /boot to be seperate? ROM version is 2.6, disk is a 9.1GB. On Sun, 14 May 2000, Ben Collins wrote: On Sun, May 14, 2000 at 01:44:05AM -0700, [EMAIL PROTECTED] wrote: I remember hearing that some of the older Sun machines could only boot from the first GB of the disk. Is there any need to have a separate partition at the beginning of the disk for /boot with the kernels and silo in it? Installation was smooth. The only problem came when I was asked to choose the hostname. ;) I did make a 16MB partition starting at offset 0 for /boot, and silo didn't want to find the kernel image on reboot. That's because silo has to know the full path of the kernel. If you put silo on part1 (/) and /boot on part2, then silo needs to look for: image=2/vmlinuz-2.2.15 Where 2 is the partition number of the current drive and /vmlinuz-2.2.15 is the path to the image relative to the partition it is on (in this case, it is not in /boot, since that is the partition we are looking on). I'm not sure there is a sane way to handle this in dbootstrap, but I'll take a look. H. Need to change the 'partition' option in silo.conf, but then one would need to mangle the 'vmlinuz' symlink into the root of whatever partition that is. Ouch, definately not trivial. Ben -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---' -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Success story: potato on Sparc Classic
H. I'm running two 36-bit and two 32-bit 16MB simms on my Sparc IPX, aparantly without any problems. The 36-bit needs to be in the first bank, though. On Tue, 11 Apr 2000, Phil Brutsche wrote: A long time ago, in a galaxy far, far way, someone said... [snip] Oh, something different: what kind of memory does a Classic need? It currently has 24MB installed, and I have a couple of 8MB 60 and 70 ns 72 pins PC SIMMs, but I'm not sure if it needs fast page mode or EDO SIMMs. I'm also not sure if a Classic really needs parity memory. It's my understanding that late Sun4c's (IPX and maybe sparctation 2) and early Sun4m's (like your Classic) will take parity fast page mode 72 pin 60 ns PC sims, but that PC simms can't be mixed with Sun simms (apparently Sun simms are 33 bit and PC simms are 36 bit) The people on the sunhelp mailing list ([EMAIL PROTECTED]) will be able to help you more than the ramblings of some college student :) -- -- Phil Brutsche [EMAIL PROTECTED] There are two things that are infinite; Human stupidity and the universe. And I'm not sure about the universe. - Albert Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: TFTPbooting potato?
On Thu, 23 Mar 2000, Jean-Paul Blaquiere wrote: On Mar 22, [EMAIL PROTECTED] scrawled : Would anyone have gotten a working i386-sparc cross-compiler that could fairly easily be packaged? If not, I'll look at modifying the m68k cross-compiler package in Potato, but last try on that (~4 weeks ago) it would not build. no. I have tried to get gcc2.95-2 to compile as a 386i -- sparc x-compiler but it would not build. If you get the m68k one working, I would love to hear from you. gcc-m68k-linux still fails in the same place it was before, compiling objc/runtime.h I'm personally wondering how the gcc-m68k-linux binary package gets built. Am I missing something I should have installed? compiling on an ELC compared to a dual celery 450 is marked. cya, -- Jean-Paul Blaquiere || Do not tempt the daemons [EMAIL PROTECTED] || -- Why? Can they hurt me? || No. But it's ... tacky. --Adam, atsr -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: TFTPbooting potato?
Would anyone have gotten a working i386-sparc cross-compiler that could fairly easily be packaged? If not, I'll look at modifying the m68k cross-compiler package in Potato, but last try on that (~4 weeks ago) it would not build. -- Beiad Ian Q. Dalton On Wed, 22 Mar 2000, Ari Heitner wrote: On Wed, 22 Mar 2000, Ben Collins wrote: Machine type would be helpful. Also, trying each of the boot types (tftp, floppy) and reporting success and failures is also helpful. You can also try building your own kernel and testing that with the floppy images (just mount them and replace linux with your kernel image). Could you post the .config you're using to build the c and cdm kernels? That way we could play around and see what works/doesn't. I'll be home on spring break this coming week, and my IPX will be up and idle, so I can do some toying around (well, I can reboot and hope it come up :) Ari -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: sun4c floppy boot images status
On Wed, 22 Mar 2000, Ben Collins wrote: On Wed, Mar 22, 2000 at 11:42:29AM -0500, Ari Heitner wrote: On Tue, 21 Mar 2000, Ben Collins wrote: Well, I finally got ahold of a sparc 1+ sun4c to test the floppy boot images, and to my dismay, they worked perfectly (both the sun4c and Yeah, I think i said this (meant to if i didn't). I took the bootdisk I had been using for my 1+, tried it on my IPX, and it died with the funky read error. BTW i did use an i386 box to write the images Ok, then it sounds like we are having problems with only IPX and IPC. Does anyone know anything about the floppy on this architecture? Are there known issues? Note, that the sun4c images are using a non-compressed kernel, so it isn't a matter of CPU/Mem. Plus it never reads the entire image anyway. Ben I know I was never able to do anything with the floppy on my IPX (read 'cat /dev/fd0 | less', write 'cat /foo/bar /dev/fd0', or format 'superformat') past performing the initial install (from Slink) and could never figure out any answers or suggestions. I usually would get about a screenful (~40 lines) of what I remember were mostly timing and sense key errors? -- Beiad Ian Q. Dalton
2.2.8 no go on Sparc IPX
Latest tests, Sparc IPX 64MB memory: Booting from floppy: read error on block 73 Booting from net: TILO, then Booting Linux... then hangs. Disk set 2.2.8-2000-03-09, rescue.bin from sun4cdm directory and tftpboot.img. One of the older builds from two or three months ago boots, but of course that one couldn't NFSmount anything. -- Beiad Ian Q. Dalton
HELP: FTP through squid proxy wants to get ENTIRE area under /
It didn't do this to me with the previous set of images, but this go around I'm getting stuff outside the tree I specify: # wget -r ftp://marcus.debian.net/pub/debian/disks-sparc/2000-02-04/ # du 912 ./marcus.debian.net/pub/dce-patches 3 ./marcus.debian.net/pub/sxid/deb 4 ./marcus.debian.net/pub/sxid/old 18 ./marcus.debian.net/pub/sxid 48580 ./marcus.debian.net/pub 2 ./marcus.debian.net/bin 2 ./marcus.debian.net/etc 2 ./marcus.debian.net/lib I did move my proxy and modem to a different machine, but I copied /etc/squid.conf right over and initialised it, so that shouldn't be it. Suggestions? -- Ferret
Re: sparc installation
On Tue, 1 Feb 2000, Anton Ivanov wrote: -BEGIN PGP SIGNED MESSAGE- [snip] Looks like it. OH. Different problem, but it might have some bearing. I just switched my network from /etc/hosts to DNS zone, and now none of my clients have permission to mount my NFS exports. Have you removed the old /etc/hosts entries? If they stay and have shorthand aliases like X.X.X.X machine.network.somewhere machine mount will fail with a forward/reverse mismatch Huh. I'm getting no permission to mount UNLESS I have the above X.X.X.X machine.internal.net machine line for the NFS client on the server. It doesn't seem to matter how the client is configured in terms of name service. And the IP address appears in the system logs too. Either remove the shorthand aliases (i.e. machine) and make sure hosts and dns match or remove all irrelevant /etc/hosts entires and rely on DNS only. [snip] - -- Anton R. Ivanov IP Engineer Level3 Communications RIPE: ARI2-RIPE E-Mail: Anton Ivanov [EMAIL PROTECTED] @*** Drucker's Dicta: (No 8 of 15) *** Start with what is right rather than what is acceptable. - -- -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.0 (GNU/Linux) Comment: For info see http://www.gnupg.org iQEVAwUBOJcFMSlWAw/bM84zAQEniQf9EIYaPmbx1DPTVK4UwB+rAnJs5ohb+14g x8DndM6adPObfrZXO9A+ib+um6OF50ow7LtPAMfVXsPFceAliYAEv8pH0FanEBPN J3ZmpgNsxLCEFoLPIPR19Rtvdwqnb+E6HRi6/N78WbueegK9zelxME9kVPeL1KV5 xP1z4FmF6YFLKWI0s7V+14uM81ferQzhnUywqvxTh1nM1yMNz+qlJlpscG5Cls+9 DpBekjelX7/Fo3Fixqu5M09IHq/SLPXI2VzIf6V1lGAY4LS71d/mB1HpCNcmul1w nLZJ1ZbR+WpRTo9Vs5FbrASroSuOzBfojGEa0ww7LtKFBsp9N+SWHQ== =suns -END PGP SIGNATURE-
Re: sparc installation
Ooookay. I get home from work today, poke at NFS, and it lets me mount my exports. I changed nothing from this morning, when it wasn't working at all. /etc/hosts in the server only lists localhost and the server's IP and hostname. The client lists localhost, the client, and the server. I guess it's just one of those things, like when I had to reboot my gateway after installing ipmasq before it would masq and not return `sendto' errors. ARGH! Thanks for the help anyway, guys. *takes a baseball bat to the cluster*
Re: sparc installation
On Tue, 1 Feb 2000, Eric Delaunay wrote: Ben Collins wrote: On Mon, Jan 31, 2000 at 10:01:06AM -0800, [EMAIL PROTECTED] wrote: Still no go.. ok boot net nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ip=192.168.1.50 even specifying ALL the options to ip= and adding root=/dev/nfs to the boot paramaters. What else can I try? That's pretty much it. It works perfectly for me. I suspect there is an underlying thing that either you or I am missing :) just a thought: is the partition exported rw, no_root_squash ? /mount/nfs/nfsroot/goober *.mentasm.org(rw,no_root_squash) Looks like it. OH. Different problem, but it might have some bearing. I just switched my network from /etc/hosts to DNS zone, and now none of my clients have permission to mount my NFS exports. I've asked about it on debian-user, but would any of y'all happen to have any suggestions? I can mount when the client is in the server's /etc/hosts, and according to the DNS howto and nslookup, everything's kosher. are you sure you are using the right kernel ? I didn't read the start of this thread, but on which subarch are you installing? (sun4c, sun4m, sun4u). Not sure, actually. The machine's a sun4c (IPX), and I'm booting both Slink's 2.2.1 and the 2.2.14 from the last set of Potato floppies. The system I installed to NFS is the year-old Slink set. I'm keeping the hard drive clean for the next Potato install. Hope this helps. -- Eric Delaunay | La guerre justifie l'existence des militaires. [EMAIL PROTECTED] | En les supprimant. Henri Jeanson (1900-1970)
Re: sparc installation
Maybe I got the boot args wrong. I was going by the diskless HOWTO, and I had it working at one point in time. RARP on boot server points to the uncompressed a.out kernel image (compiled elf2aout on i386) and the kernels do boot to the point where they can't mount root FS on 00.00 ok boot net nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober . :VFS: Cannot open root device 00:00 :Kernel panic: VFS: Unable to mount root fs on 00:00 As far as I can tell here, everything's set up as it should be. I'm booting the 2.2.14 kernel from the last set of floppy images. On Sun, 30 Jan 2000, Ben Collins wrote: On Sat, Jan 29, 2000 at 06:42:36PM -0800, [EMAIL PROTECTED] wrote: Ben, Do the 2.2.14 Sparc kernel images have root on NFS/IP PnP support? Yes, but the PnP RARP is disabled on all of them. It's just as simple to type it in once on the command line for install, and recompile for those few who RARP or BOOTP all the time for diskless clients. Even then, you can still put the command line stuff in the PROM boot args, if needed. -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Re: sparc installation
Still no go.. ok boot net nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ip=192.168.1.50 even specifying ALL the options to ip= and adding root=/dev/nfs to the boot paramaters. What else can I try? On Mon, 31 Jan 2000, Ben Collins wrote: On Sun, Jan 30, 2000 at 09:33:58PM -0800, [EMAIL PROTECTED] wrote: Maybe I got the boot args wrong. I was going by the diskless HOWTO, and I had it working at one point in time. RARP on boot server points to the uncompressed a.out kernel image (compiled elf2aout on i386) and the kernels do boot to the point where they can't mount root FS on 00.00 ok boot net nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ok boot net nfsroot=... ip=this systems ip: Read the nfsroot.txt in linux/Documentation from the kernel. That one extra ip addres is what the kernel needs (the original RARP that the PROM gets isn't used, maybe it should be, but that is a kernel bug). -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Re: sparc installation
Ben, Do the 2.2.14 Sparc kernel images have root on NFS/IP PnP support? On Sat, 29 Jan 2000, Ben Collins wrote: On Sat, Jan 29, 2000 at 06:58:53PM +0900, Ragga Muffin wrote: [snip of installing Slink first] A) You can always upgrade the kernel first. I just uploaded 2.2.14 images yesterday. B) This only affects sun4m systems (sparc 5, 10, 20, etc...). Sun4c and Sun4u do not exhibit the problem and only require a = 2.2.0 kernel. Still unable to fully install Slink or Potato yet. I'm keeping my Sparc's hard drive clean until I can boot the Potato install images on it and install there. And Slink won't let me boot my NFS-installed system. : Do you have an ETA for a new set of images to test? -- Ferret
more problems netbooting
Basically I can't get a kernel to mount root NFS. I converted the 2.2.1 kernel in the slink bootdisks to a.out and booted that, then I tried the kernels in the 2000-01-15 potato floppy test, and no go. AFAIK (and what I've been told previously on this list) my PROM boot arguments are correct. ok boot net boot=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ip=192.168.1.50:192.168.1.42 And I did install the slink base system into /mount/nfs/nfsroot/goober Am I doing something wrong here, or do none of these kernel images have nfsroot support? Could someone compile me a 2.2.1 or 2.2.14 kernel with nfsroot enabled and send it to me in a.out format? It's a sun4c (IPX) with serial console. grumpy
Re: Partitioning SparcStation
(I'm pointing you to debian-sparc here..) On my IPX, there are some options in the PROM environment that have to do with the boot process. Key in `printenv' at the PROM console prompt (ok) and go from there. I'm afraid I can't step through my machine right now. Hope this helps. On Mon, 17 Jan 2000, Fraser Campbell wrote: I have a sun4m machine and Debian installed great on it with one exception. It will not boot correctly (or at least the way I'd like it). The first hard drive has partition 3 labelled as type Whole Disk. The hard drive is then partitioned with 500 MB for / and 500 MB for /tmp. When I reboot the computer waits at what seems to be a Sun boot prompt. I need to manually type in boot disk1:1 to tell the machine to boot from the first partition of the first disk. After doing that I proceed to the SILO prompt where the Linux boot procedure that I am familiar with commences. I really need this machine to be remotely rebootable ... can it be done? Thanks, Fraser
Re: New set of sparc boot disks
On Sun, 16 Jan 2000, Ben Collins wrote: On Sat, Jan 15, 2000 at 08:33:42PM -0800, John Chapman wrote: On Sat, 15 Jan 2000, Ben Collins wrote: The new set of disks are at: ftp://marcus.debian.net/pub/debian/disks-sparc/current sun4c failures tftpboot seems to hang on my IPX.. |ok boot net | |SPARCstation IPX, No Keyboard |ROM Rev. 2.6, 64 MB memory installed, Serial #2165154. |Ethernet address 8:0:20:c:dc:80, Host ID: 572109a2. | | |Rebooting with command: net |Boot device: /sbus/[EMAIL PROTECTED],c0 File and args: |2f7800 |TILO |Booting Linux... And stays there. (I powered down after 8 minutes) Booting from floppy, I get the same `read error on block 73', sun4cdm boot image. And the sun4c boot image does the same thing. This is truly odd since they are the exact same kernels as before. Well, what else changed?
Re: ANN: initial sparc potato boot floppies ready for testing
On Wed, 12 Jan 2000, Eric Delaunay wrote: [EMAIL PROTECTED] wrote: Can you make the images available by FTP as well, or tweak the http to You can connect ftp to xia01 with your developper's account. Look under /home/buildd/... or something like that. It was what I used the last time to download BF. Er, IANADDY? : I don't have an account for FTP access.
Re: boot floppies and NFSroot?
On Wed, 12 Jan 2000, Ben Collins wrote: On Wed, Jan 12, 2000 at 08:48:38AM -0800, [EMAIL PROTECTED] wrote: I'm tftpbooting the sun4cdm/linux-a.out image, boot args at PROM: The disks on the site right now do not have this. The kernel images on that same URL do have this support, so you can extract one of those and run elftoaout if you want. Two questions: Can I do this to a Sparc binary on an i386 system, and where do I find this `elftoaout' utility? I currently do not have a bootable Sparc system.
Re: ANN: new boot disks to test
New go with the boot disks, mostly the same problems. I'm setting up an NFSroot with the old Slink disks, but I've forgotten how to make it boot over NFS again. I copied the vmlinuz file from the install into /tftpboot/HEX IP address.SUN4C and passed some boot paramaters. |ok boot net root=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ip=192.168.1.50:192.168.1.42 |Boot device: /sbus/[EMAIL PROTECTED],c0 File and args: root=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober ip=192.168.1.50:192c4c00 |The file just loaded does not appear to be executable. |ok Could someone get me quickly fixed here, then I'll print out the instructions and try not to lose them this time? Machine is a Sparc IPX with generic 4x16MB 36-bit SIMMS and OEM hard drive on serial console. 2000-01-06 images. Booted by TFTP, serial console supplied by minicom. Rebooted and supplied `TERM=vt102' to netboot to get rid of the slower colour/hi-graphics stuff. Not too bad. Cursor is one line above and one charcell behind the hilighted selection. But it is hilighted. Buttons have the cursor immediately after the button hilighted. Wishlist: Change the wording about the new ext2 `features'. Something like `do you wish to use these new features (not reccomended if you're going to use a 2.0 kernel), then the `yes' option will enable the features. The current wording smacks me like a double-negative. Now I go on to install the kernel and modules from `network'. I configure the network up. I leave the `network' install site as the default, and after a few minutes (with modem activity) get |nf_http_fetchfile :: couldn't open file |/target/tmp/sun4c/disks-1.44/rescue.bin for writing |Download of file disks-1.44/rescue.bin failed I'm still getting the `invalid argument' error when mounting by NFS. Though my NFS server tells me the mount was successful. I have a working floppy drive on my lintel again, so I'll try making bootfloppies in a little while. Long download on a 19.2k connection. Then I get this at the shell (with a typo): |# ae /etc/resolv/conf |S-Lang Error: Unknown Error Code |SLcurses_initscr: init failed And I'm locked except I can break to PROM. AND I forgot how to nfsrootboot linux on this thing. :/
Re: ANN: initial sparc potato boot floppies ready for testing
Can you make the images available by FTP as well, or tweak the http to make downloading more wget-friendly? Currently I'm using |wget -r -c --follow-ftp \ http://xia01.kachinatech.com/~buildd/sparc-2.2.14-potato-boot/2000-01-06/ But I find I have to manually get each directory under that because it only downloads the index.html file. Or perhaps I'm using the wrong options to wget (but those look right, according to the manual), and if so, what options should I be using?
boot floppies and NFSroot?
I'm tftpbooting the sun4cdm/linux-a.out image, boot args at PROM: |ok boot net root=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober,rsize=8192,wsize=8192 ip=192.168.1.50:192.168.1.42 |SPARCstation IPX, No Keyboard |ROM Rev. 2.6, 64 MB memory installed, Serial #2165154. |Ethernet address 8:0:20:c:dc:80, Host ID: 572109a2. |Rebooting with command: net root=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober,rsize=8192,wsize=8192 ip=192.168.1.50:2Boot device: /sbus/[EMAIL PROTECTED],c0 File and args: root=/dev/nfs nfsroot=192.168.1.42:/mount/nfs/nfsroot/goober,rsize=8192,wsize=82180c00 |Booting Linux... |PROMLIB: Sun Boot Prom Version 2 Revision 2 |Linux version 2.2.14 ([EMAIL PROTECTED]) (gcc version 2.95.2 19991109 (Debian |GNU/Linux)) #1 Thu Jan 6 15:02:08 EST 2000 |ARCH: SUN4C |TYPE: Sun4c SparcStation IPX |Ethernet address: 8:0:20:c:dc:80 |Loading sun4c MMU routines . |VFS: Cannot open root device 00:00 |Kernel panic: VFS: Unable to mount root fs on 00:00 |Press L1-A to return to the boot prom This is according to linux/Documentation/nfsroot.txt sun4cdm/linux-a.out does have NFSroot support?
question about floppy drive support
I've been having bad troubles with the FDD on my Sparc IPX. Well, I booted the 2.2.14 a.out kernel from the new testing run of install disks, and after a fresh install with the old 2.2.1-based Slink disks I can access the FDD. I slipped a floppy in and could cat it without any trouble. Then I installed `fdutils' from potato. Accessing the same disk yields |floppy0: sector not found: track 0, head 0, sector 1, size 2 |floppy0: sector not found: track 0, head 0, sector 1, size 2 |end_request: I/O error, dev 02:00 (floppy), sector 0 |cat: /dev/fd0: Input/output error and of course superformat isn't working. Still, it's working better than it was. is there something besides superformat I can use on the Sparc, or do I just have a funky machine?
Re: ANN: initial sparc potato boot floppies ready for testing
On Sun, 9 Jan 2000, Eric Delaunay wrote: [EMAIL PROTECTED] wrote: Machine is a Sparc IPX with generic 4x16MB 36-bit SIMMS and OEM hard drive on serial console. Booted by TFTP Console is `minicom' running on Slink i386. Noticed that dbootstrap is not drawing with the IBM `box' characters like 2.1 dbootstrap did. Pass TERM=vt100 or TERM=vt102 at boot time. Current init process does not enforce TERM settings anymore ;(( I don't know wether I will enable it again or just let the user pass TERM=xxx at boot time. Maybe a serial target for silo could also be helpful for floppy boot. On this topic, do you know about a way of identifying the type of console connected to the serial line ? Is there an escape sequence that returns vt100 or the like ? There is such a beastie as far as I've heard, but I don't know what it might be. You might check the `ctlseqs.ms' file in the xterm source to see what xterm uses, then check it against minicom. The file is 14k compressed, in case anyone needs a copy. I've read the thing, but I'm not even a novice at reading escape codes. ;
Re: ANN: initial sparc potato boot floppies ready for testing
On Sun, 9 Jan 2000, Michael Stone wrote: On Sun, Jan 09, 2000 at 04:29:40PM +0100, Eric Delaunay wrote: Pass TERM=vt100 or TERM=vt102 at boot time. Current init process does not enforce TERM settings anymore ;(( I don't know wether I will enable it again or just let the user pass TERM=xxx at boot time. Maybe a serial target for silo could also be helpful for floppy boot. On this topic, do you know about a way of identifying the type of console connected to the serial line ? Is there an escape sequence that returns vt100 or the like ? IMHO, it's safe enough to assume vt100 as long as there's a way to override that assumption if necessary. How many people actually use terminals that don't speak vt100? I know of quite a few Wyse (50|60) terminals still alive and well being used as serial terminals. Don't know of any being used on Sparcs right now, but a colleague is using one on his i386. ;
Re: ANN: initial sparc potato boot floppies ready for testing
Summary: Can't do a network install with the bootdisks. (And don't have a way of writing floppy disks for the moment) Transcript of my install attempts attached. Machine is a Sparc IPX with generic 4x16MB 36-bit SIMMS and OEM hard drive on serial console. Booted by TFTP Console is `minicom' running on Slink i386. Noticed that dbootstrap is not drawing with the IBM `box' characters like 2.1 dbootstrap did. (I'm replacing a system that got a corrupted libc6, BTW) Initialised and activated swap. Initialised ext2 partition with the `new ext2 filesystem features' (Just what are these new features?) Did a `network' install and configured the network (very helpful to be sent to the configuration option here) It would be helpful to be able to specify a proxy server as part of the network configuration, though, and have it default to the gateway just as DNS does. Then I get a problem: nf_http_fetchfile :: couldn't open file /target/tmp/sun4c/disks-1.44/rescue.bin for writing Download of file disks-1.44/rescue.bin failed I'm trying the network install a second time to see if it works. I enter my HTTP proxy information AGAIN (not helpful) Same problem again. I try to do an install from NFS, and it refuses to mount my server. Error mounting NFS filesystem 192.168.1.42:/ , please try again. Then I try mounting NFS from the shell: # mount -t nfs 192.168.1.42:/ /mnt Mounting 192.168.1.42:/ on /mnt failed: Invalid argument But my server says it has been successfully mounted(!) Jan 7 13:31:22 playdough mountd[302]: NFS mount of / attempted from # 192.168.1.44 Jan 7 13:31:22 playdough mountd[302]: / has been mounted by # 192.168.1.44 # ifconfig eth0 Link encap:Ethernet HWaddr 08:00:20:0C:DC:80 inet addr:192.168.1.44 Bcast:192.168.1.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:103 errors:0 dropped:0 overruns:0 frame:0 TX packets:87 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 Interrupt:5 Base address:0xb400 loLink encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:3924 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 # route Kernel IP routing table Destination Gateway Genmask Flags Metric RefUse Iface 192.168.1.0 * 255.255.255.0 U 0 00 eth0 default 192.168.1.420.0.0.0 UG1 00 eth0 # After this I restarted dbootstrap, redid the network setup, then something had indigestion when I went to a shell. Finally threw me back into openprom. Nxt boot through, get to installing from NFS: Error mounting NFS filesystem 192.168.1.42:/mount/nfs/nfs/nasal/local , please try again. Seems like NFS mounting is broken. Next set perhaps?
Kernel stopped compiling after update to newest Potato..
Kernel compile that WAS working as of a week ago. I had to do a full reinstall of the locak disk, but I have /usr/src/NFS-mounted, and the kernel tree didn't change AFAICT. What would be the most likely culprit? BTW, I'm still trying to compile libc, and I found that I had to run with 64MB in order to do that. But bad memory would be much more likely to produce SIGSEGV's, wouldn't it? -- Ferret no baka --- gcc -D__KERNEL__ -I/usr/src/linux-2.2.13/include -c -o btfix.o btfix.s make[2]: Leaving directory `/usr/src/linux-2.2.13/arch/sparc/boot' ld -m elf32_sparc -T arch/sparc/vmlinux.lds arch/sparc/kernel/head.o arch/sparc/kernel/init_task.o init/main.o init/version.o \ --start-group \ arch/sparc/kernel/kernel.o arch/sparc/mm/mm.o kernel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o arch/sparc/math-emu/math-emu.o arch/sparc/boot/btfix.o \ fs/filesystems.a \ net/network.a \ drivers/block/block.a drivers/char/char.a drivers/misc/misc.a drivers/net/net.a drivers/scsi/scsi.a drivers/cdrom/cdrom.a drivers/sbus/sbus.a drivers/video/video.a \ /usr/src/linux-2.2.13/lib/lib.a /usr/src/linux-2.2.13/lib/lib.a /usr/src/linux-2.2.13/arch/sparc/prom/promlib.a /usr/src/linux-2.2.13/arch/sparc/lib/lib.a \ --end-group \ -o vmlinux arch/sparc/kernel/kernel.o(__ex_table+0x46c): undefined reference to `kernel_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x474): undefined reference to `kernel_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x47c): undefined reference to `kernel_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x484): undefined reference to `kernel_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x48c): undefined reference to `kernel_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x494): more undefined references to `kernel_unaligned_trap_fault' follow arch/sparc/kernel/kernel.o(__ex_table+0x544): undefined reference to `user_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x54c): undefined reference to `user_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x554): undefined reference to `user_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x55c): undefined reference to `user_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x564): undefined reference to `user_unaligned_trap_fault' arch/sparc/kernel/kernel.o(__ex_table+0x56c): more undefined references to `user_unaligned_trap_fault' follow make[1]: *** [vmlinux] Error 1 make[1]: Leaving directory `/usr/src/linux-2.2.13' make: *** [stamp-build] Error 2 ---
weeeeeeird libc/ntpdate behaviour?
`ntpdate' is actually working, but only when I give it a DQ IP address. Any time I give it an hostname it seems to get stuck in :0xe0144c40 in __pthread_compare_and_swap () from /lib/libpthread.so.0 according to gdb. ntpdate with debugging symbols: --- Program received signal SIGINT, Interrupt. 0xe0144c40 in __pthread_compare_and_swap () from /lib/libpthread.so.0 (gdb) backtrace #0 0xe0144c40 in __pthread_compare_and_swap () from /lib/libpthread.so.0 #1 0xe0144afc in __pthread_lock () from /lib/libpthread.so.0 #2 0xe0142aa0 in pthread_mutex_lock () from /lib/libpthread.so.0 #3 0xe00f7c1c in gethostbyname () from /lib/libc.so.6 #4 0x133ac in getnetnum (host=0xefffeea1 playdough, num=0xefffebe4) at ntpdate.c:1873 #5 0x12b9c in addserver (serv=0xefffeea1 playdough) at ntpdate.c:1294 #6 0x116ac in ntpdatemain (argc=3, argv=0xefffedb4) at ntpdate.c:484 #7 0x11254 in main (argc=3, argv=0xefffedb4) at ntpdate.c:295 --- I also tried compiling libc on my box. First problem: didn't have enough swap, so I made and activated a 64MB swap file. Second problem: --- gcc nsswitch.c -c -O2 -Wall -Winline -Wstrict-prototypes -Wwrite-strings -fstrict-aliasing -g -fPIC-I../include -I. -I/usr/local/packages/glibc-2.1.2/sparc-linux/obj/nss -I.. -I../libio -I/usr/local/packages/glibc-2.1.2/sparc-linux/obj -I../sysdeps/sparc/sparc32/elf -I../crypt/sysdeps/unix -I../linuxthreads/sysdeps/unix/sysv/linux -I../linuxthreads/sysdeps/pthread -I../linuxthreads/sysdeps/unix/sysv -I../linuxthreads/sysdeps/unix -I../linuxthreads/sysdeps/sparc/sparc32 -I../linuxthreads/sysdeps/sparc -I../sysdeps/unix/sysv/linux/sparc/sparc32 -I../sysdeps/unix/sysv/linux/sparc -I../sysdeps/unix/sysv/linux -I../sysdeps/gnu -I../sysdeps/unix/common -I../sysdeps/unix/mman -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix/sparc -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/sparc/sparc32 -I../sysdeps/wordsize-32 -I../sysdeps/sparc/fpu -I../sysdeps/sparc -I../sysdeps/ieee754 -I../sysdeps/libm-ieee754 -I../sysdeps/generic/elf -I../sysdeps/generic -nostdinc -isystem /usr/lib/gcc-lib/sparc-linux/2.95.2/include -isystem /usr/src/linux-2.2.13//include -D_LIBC_REENTRANT -include ../include/libc-symbols.h -DPIC -o /usr/local/packages/glibc-2.1.2/sparc-linux/obj/nss/nsswitch.os gcc: Internal compiler error: program cc1 got fatal signal 10 make[3]: *** [/usr/local/packages/glibc-2.1.2/sparc-linux/obj/nss/nsswitch.os] Error 1 make[3]: Leaving directory `/usr/local/packages/glibc-2.1.2/glibc-2.1.2/nss' make[2]: *** [nss/subdir_lib] Error 2 make[2]: Leaving directory `/usr/local/packages/glibc-2.1.2/glibc-2.1.2' make[1]: *** [all] Error 2 make[1]: Leaving directory `/usr/local/packages/glibc-2.1.2/sparc-linux/obj' --- signal 10==SIGBUS What would have caused that? Oh, same version of ntpdate compiled on i386 runs without any problem. -- Ferret no baka
Re: 'console-tools' cannot find console
On Sun, 12 Dec 1999 [EMAIL PROTECTED] wrote: On Sun, 12 Dec 1999, Ben Collins wrote: On Sun, Dec 12, 1999 at 09:31:05AM -0800, [EMAIL PROTECTED] wrote: On Sun, 12 Dec 1999, Ben Collins wrote: On Sun, Dec 12, 1999 at 08:19:27AM -0800, [EMAIL PROTECTED] wrote: `Couldnt get a file descriptor referring to the console' I get this message every time I boot my Sparc. Now 'console-tools' is erroring when dpkg tries to install it. What sould I look for? crw-r--r--1 root root 5, 1 Dec 11 18:36 /dev/console I have my /dev/console set up for 2.2.x kernels, and it seems to work perfectly fine. IMO, it is a bug in console-tools, but I can't pinpoint the problem. Should file a bug on that package marked important for severity. Would it matter that my console is on ttya? No, because mine isn't, yet I have the same problem. Actually, I think I had the same problem before installing console-tools (which the UTF-8 support needs) I did an xtrace on console-tools' postint, and it looks like the offender is actually debconf in this case. ::++ exec /usr/share/debconf/frontend ./console-tools.postinst configure ::Couldnt get a file descriptor referring to the console Although... I just switched my development lintel to serial console and I sometimes get the same error under other circumstances. I'm still working on that one. -- Ferret no baka
'console-tools' cannot find console
`Couldnt get a file descriptor referring to the console' I get this message every time I boot my Sparc. Now 'console-tools' is erroring when dpkg tries to install it. What sould I look for? crw-r--r--1 root root 5, 1 Dec 11 18:36 /dev/console I have my /dev/console set up for 2.2.x kernels, and it seems to work perfectly fine. -- Ferret no baka
Re: 'console-tools' cannot find console
On Sun, 12 Dec 1999, Ben Collins wrote: On Sun, Dec 12, 1999 at 08:19:27AM -0800, [EMAIL PROTECTED] wrote: `Couldnt get a file descriptor referring to the console' I get this message every time I boot my Sparc. Now 'console-tools' is erroring when dpkg tries to install it. What sould I look for? crw-r--r--1 root root 5, 1 Dec 11 18:36 /dev/console I have my /dev/console set up for 2.2.x kernels, and it seems to work perfectly fine. IMO, it is a bug in console-tools, but I can't pinpoint the problem. Should file a bug on that package marked important for severity. Would it matter that my console is on ttya?
Re: 'console-tools' cannot find console
On Sun, 12 Dec 1999, Ben Collins wrote: On Sun, Dec 12, 1999 at 09:31:05AM -0800, [EMAIL PROTECTED] wrote: On Sun, 12 Dec 1999, Ben Collins wrote: On Sun, Dec 12, 1999 at 08:19:27AM -0800, [EMAIL PROTECTED] wrote: `Couldnt get a file descriptor referring to the console' I get this message every time I boot my Sparc. Now 'console-tools' is erroring when dpkg tries to install it. What sould I look for? crw-r--r--1 root root 5, 1 Dec 11 18:36 /dev/console I have my /dev/console set up for 2.2.x kernels, and it seems to work perfectly fine. IMO, it is a bug in console-tools, but I can't pinpoint the problem. Should file a bug on that package marked important for severity. Would it matter that my console is on ttya? No, because mine isn't, yet I have the same problem. Actually, I think I had the same problem before installing console-tools (which the UTF-8 support needs)
Re: Still probelms to boot from SD
On Wed, 8 Dec 1999 [EMAIL PROTECTED] wrote: In a message dated 12/7/99 11:40:34 PM US Mountain Standard Time, [EMAIL PROTECTED] writes: /dev/sda1 0 1100396000 83 Linux native /dev/sda2 u 1100 1151 18360 82 Linux swap /dev/sda3 0 11514143600 Empt So why is the swap partition marked as read only? This is what the u flag designates, I believe. Swap seems to work if it is set as read only or without the flag. I have the swap flagless on a SPARC10. Just for cleanness and that Debian uses it anyway and ignores the read only flag? Just curious I really don't know. That's how the Slink installer's fdisk set it up when I did the partitioning. All I can say is that it works. -- Ferret no baka (6/7th-clued Sparc newbie)
Re: Still probelms to boot from SD
On Thu, 9 Dec 1999, Sanjeev Gupta wrote: On Wed, 8 Dec 1999 [EMAIL PROTECTED] wrote: In a message dated 12/7/99 11:40:34 PM US Mountain Standard Time, [EMAIL PROTECTED] writes: /dev/sda1 0 1100396000 83 Linux native /dev/sda2 u 1100 1151 18360 82 Linux swap /dev/sda3 0 11514143600 Empt The u is unmountable. No idea why, but it exists, as a flag, on SunOS as well. To the original poster, are you sure that you are formatting the disk on a Sparc Linux box? Yep. That Sparc was purchased second-hand with an unbootable Solaris/SunOS system on it. After mounting the original partition and poking around I removed all the partitions except sda3 and repartitioned it with fdisk from the slink floppy install set. -- Ferret no baka
Re: Still probelms to boot from SD
On Wed, 8 Dec 1999, Manfred Plagmann wrote: Hi, I still can't get my Sparc IPX booting Linux from the HD. I have received some answers to this problem which gave me a better insight. As I understand the HD has to have a partition which spans the whole drive. I intent to have one Linux and one swap partition. I therefore set the partition table up to have one 83 (Linux native) and one 82 (Linux swap) the third partition should now span the whole disk and should have the type WHOLE DISK. But I don't see this type listed in fdisk. Someone wrote to use type 5 which is EXTENDED but this somehow didn't do it. [Nifty: line-justify in pine 3.96 auto-quotes a quoted line] Any help is appreciated. Manfred Here's the partitioning on my IPX's HD (The OEM hard drive AFAICT) nasal:~# fdisk -l /dev/sda Disk /dev/sda (Sun disk label): 9 heads, 80 sectors, 1151 cylinders Units = cylinders of 720 * 512 bytes Device FlagStart EndBlocks Id System /dev/sda1 0 1100396000 83 Linux native /dev/sda2 u 1100 1151 18360 82 Linux swap /dev/sda3 0 11514143600 Empty BTW, mine had a non-booting Solaris install on it when I got it, and I just wiped the Solaris partition and installed Linux instead. /dev/sda3 was there from Solaris AFAICT. -- Ferret no baka
Re: sd-partition
I'm also running an IPX, mine with a serial console. Running serial console on one of these is really the best way of getting it working, insofar as you can capture boot monitor output. I'd ask you to do 'printenv' from the openprom and post it. On Mon, 29 Nov 1999, Manfred Plagmann wrote: Hi, I inherited a Sun IPX which did not start up. I down loaded Debian-sparc on the weekend and found that the machine was booting :)) I installed the floppy disk base version and all seems OK the computer is however not booting by its own and needs a boot disk. What did I miss when I installed the system or is it not possible to have the IPX boot directly from the HD? I guess that I forgot something when I set the partitions of the primary disk. The installation manual on the debian server is unfortunately incomplete (at least on the Australian and the US server) as no text follows the subsection heading Partitioning prior to installation. At startup the loader complains that the magic word on the boot disk is missing. How can I set this word. Any help is appreciated. Cheers, Manfred
Can't format floppies..
I captured everything I could think of. homebrewed 2.2.13 from ftp.kernel.org source tarball, everything running potato. Known problem? Oh yeah, it's the IPX. -- Ferret no baka superformat /dev/fd0 Measuring drive 0's raw capacity floppy0: Getstatus times out (10) on fdc 0 floppy driver state --- now=15777840 last interrupt=15777840 diff=0 last called handler=f00c6f44 timeout_message=floppy start last output bytes: 8 80 15777834 8 80 15777834 12 80 15777837 0 90 15777837 13 80 15777837 0 90 15777837 1a 90 15777837 0 90 15777837 3 80 15777837 c1 90 15777837 11 90 15777837 7 80 15777837 0 90 15777837 8 81 15777837 4d 80 15777837 0 90 15777837 2 90 15777837 1 90 15777837 ff 90 15777837 3 90 15777837 last result at 15777837 last redo_fd_request at 15777834 20 0 status=d0 fdc_busy=1 fd_timer.function=f00c6e7c timer_table=f00c9134 expires=295 now=15777840 cont=f0136ecc CURRENT= command_status=-1 floppy0: get result error. Fdc=0 Last status= Read bytes=0 floppy driver state --- now=15777840 last interrupt=15777840 diff=0 last called handler=f00c6f44 timeout_message=floppy start last output bytes: 8 80 15777834 8 80 15777834 12 80 15777837 0 90 15777837 13 80 15777837 0 90 15777837 1a 90 15777837 0 90 15777837 3 80 15777837 c1 90 15777837 11 90 15777837 7 80 15777837 0 90 15777837 8 81 15777837 4d 80 15777837 0 90 15777837 2 90 15777837 1 90 15777837 ff 90 15777837 3 90 15777837 last result at 15777837 last redo_fd_request at 15777834 20 0 status=d0 fdc_busy=1 fd_timer.function=f00c6e7c timer_table=f00c9134 expires=295 now=15777840 cont=f0136ecc CURRENT= command_status=-1 floppy0: unexpected interrupt repl[0]=0 repl[1]=0 repl[2]=0 repl[3]=ff repl[4]=4 repl[5]=1 repl[6]=2 format: Input/output error nasal:~# uname -a Linux nasal 2.2.13 #1 Wed Nov 17 14:10:27 PST 1999 sparc unknown nasal:~# cat /proc/devices Character devices: 1 mem 2 pty 3 ttyp 4 ttyS 5 cua 7 vcs 10 misc 29 fb 128 ptm 136 pts Block devices: 2 fd 8 sd nasal:~# cat /proc/ioports fe00-fe000fff : auxiliaryIO fe001000-fe001fff : sun4c_interrupts fe002000-fe002fff : clock fe003000-fe003fff : timer fe004000-fe004fff : dma fe005000-fe005fff : floppy fe006000-fe006fff : ESP Registers fe007000-fe007fff : LANCE fff0-ffff : ESP DVMA Cmd Block fff01000-fff0d21b : LANCE DMA nasal:~# cat /proc/interrupts 3: 401937Sparc ESP SCSI 5:1539770LANCE 10: 15781426 + timer 11: 21 + floppy 12: 678008 + Zilog8530 nasal:~# c ls -l /dev/fd0* brw-rw1 root floppy 2, 0 Nov 25 15:44 /dev/fd0 brw-rw1 root floppy 2, 24 Nov 25 15:44 /dev/fd0u1760 brw-rw1 root floppy 2, 8 Nov 25 15:44 /dev/fd0u2880 brw-rw1 root floppy 2, 28 Nov 25 15:44 /dev/fd0u3840 brw-rw1 root floppy 2, 12 Nov 25 15:44 /dev/fd0u410 brw-rw1 root floppy 2, 16 Nov 25 15:44 /dev/fd0u420 brw-rw1 root floppy 2, 4 Nov 25 15:44 /dev/fd0u720 brw-rw1 root floppy 2, 20 Nov 25 15:44 /dev/fd0u880 nasal:~# superformat /dev/fd0u720 Measuring drive 0's raw capacity floppy0: Getstatus times out (10) on fdc 0 floppy driver state --- now=15784467 last interrupt=15784467 diff=0 last called handler=f00c6f44 timeout_message=floppy start last output bytes: 8 80 15784461 8 80 15784461 12 80 15784466 0 90 15784466 13 80 15784466 0 90 15784466 1a 90 15784466 0 90 15784466 3 80 15784466 e1 90 15784466 9 90 15784466 7 80 15784466 0 90 15784466 8 81 15784466 4d 80 15784466 0 90 15784466 2 90 15784466 1 90 15784466 ff 90 15784466 3 90 15784466 last result at 15784466 last redo_fd_request at 15784462 20 0 status=d0 fdc_busy=1 fd_timer.function=f00c6e7c timer_table=f00c9134 expires=295 now=15784467 cont=f0136ecc CURRENT= command_status=-1 floppy0: get result error. Fdc=0 Last status= Read bytes=0 floppy driver state --- now=15784467 last interrupt=15784467 diff=0 last called handler=f00c6f44 timeout_message=floppy start last output bytes: 8 80 15784461 8 80 15784461 12 80 15784466 0 90 15784466 13 80 15784466 0 90 15784466 1a 90 15784466 0 90 15784466 3 80 15784466 e1 90 15784466 9 90 15784466 7 80 15784466 0 90 15784466 8 81 15784466 4d 80 15784466 0 90 15784466 2 90 15784466 1 90 15784466 ff 90 15784466 3 90 15784466 last result at 15784466 last redo_fd_request at 15784462 20 0 status=d0 fdc_busy=1 fd_timer.function=f00c6e7c timer_table=f00c9134 expires=295 now=15784467 cont=f0136ecc CURRENT= command_status=-1 floppy0: unexpected interrupt repl[0]=0 repl[1]=0 repl[2]=0 repl[3]=ff repl[4]=4 repl[5]=1 repl[6]=2 format: Input/output error nasal:~#
dynamic linker bug?
Just updated potato yesterday, now I'm getting --- nasal:/usr/local/packages/modutils-2.3.6# debian/rules binary BUG IN DYNAMIC LINKER ld.so: dynamic-link.h: 57: elf_get_dynamic_info: Assertion `! bad dynamic tag' failed! Segmentation fault --- Trying to self-compile modutils. I downgraded ldso to 1.9.10-1 (slink) and get the same thing, so it doesn't look like a problem with ldso. Is this known? -- Ferret no baka
Re: dynamic linker bug?
On Thu, 25 Nov 1999, Ben Collins wrote: On Thu, Nov 25, 1999 at 09:46:08AM -0800, [EMAIL PROTECTED] wrote: Just updated potato yesterday, now I'm getting --- nasal:/usr/local/packages/modutils-2.3.6# debian/rules binary BUG IN DYNAMIC LINKER ld.so: dynamic-link.h: 57: elf_get_dynamic_info: Assertion `! bad dynamic tag' failed! Segmentation fault --- Trying to self-compile modutils. I downgraded ldso to 1.9.10-1 (slink) and get the same thing, so it doesn't look like a problem with ldso. Is this known? A reboot should cure this. Just a note, ld.so in potato is supplied by libc6 (except for libc5 programs, which use the ld.so in the ldso package). No one has been able to track this down, but I suspect it's caused by the glibc-pre 2.0.105 - 2.1 upgrade. Ah. No, unfortunately this is still happening after reboot. I had my Sparc down overnight (didn't have a working console for it due to smacking around a serial card with a clueby4 -- another story) Oh, I had some MAJOR breakage during the last upgrade. nscd was somehow getting respawned so I disabled it, then dselect crashed and dpkg was still configuring things in the background.. Maybe I should force-reinstall libc6? -- Ferret no baka
'function not implemented' when upgrading to potato
I get errors similar to and including the following during dist-upgrade. I'm running kernel 2.0.35 with Sparc patches and just hand-upgraded to glibc2.1 i mamually unpacked fileutils and used the ch* executables out of there, so I know it's not a problem with fileutils. I suspect it's a problem with libc6 :nasal:/usr/local/foo/baz/bin# apt-get install fileutils :Reading Package Lists... Done :Building Dependency Tree... Done :1 packages upgraded, 0 newly installed, 0 to remove and 132 not upgraded. :1 packages not fully installed or removed. :Need to get 0B/575kB of archives. After unpacking 897kB will be used. :(Scanning database ... 13129 files and directories currently installed.) :Preparing to replace fileutils 3.16-5.4 (using \ .../fileutils_4.0i-2_sparc.deb) ... :Unpacking replacement fileutils ... :dpkg: error processing \ /var/cache/apt/archives/fileutils_4.0i-2_sparc.deb (--unpack): : error setting ownership of `./usr/share/info': Function not implemented :dpkg-deb: subprocess paste killed by signal (Broken pipe) :Errors were encountered while processing: : /var/cache/apt/archives/fileutils_4.0i-2_sparc.deb :E: Sub-process returned an error code (1) -- Ferret no baka
More strange Sparc behaviour
Serial console goes offline at 'init 6', does not come back online. Runlevel change does not seem to work when 'debug=1' is set in /etc/init.d/rc It doesn't actually run the scripts. 'fixed' that behaviour though. Why would the serial console go offline on reboot/halt? -- Ferret no baka (I can post a rather large console log if anyone wants to see it)
Re: Cross-compiling on Debian systems?
On Wed, 17 Nov 1999, Ben Collins wrote: On Wed, Nov 17, 1999 at 02:48:55PM -0800, [EMAIL PROTECTED] wrote: I'd like to set up a cross-compiler on my Lintel (AMD 266) to compile packages for my Sparc IPX. I'd preferably like to package the cross-compiler so I can esaily remove it if I no longer need it. The only cross-compiler in the slink packages list is for m68k. What do I need to do to set one up? The last cross-compiler I built (for kernel compiles) was using the m68k cross-compiler package. I just changed the m68k references to sparc, and ran dpkg-buildpackage -us -uc -rfakeroot (change all of the m68k references in debian/ to sparc). I just did a 'debian/rules binary' in 'gcc-m68k-linux-2.8.1' after changing 'm68k' to 'sparc', and I get a build error. Looks like the gcc buildstrap isn't working. ./xgcc -B./ -DCROSS_COMPILE -DIN_GCC-O2 -I./include -I. -I. \ -I./config -g0 \ -finhibit-size-directive -fno-inline-functions -fno-exceptions \ -c ./crtstuff.c -DCRT_BEGIN -o crtbegin.o as: option `-s' is ambiguous make[1]: *** [crtbegin.o] Error 1 make[1]: Leaving directory `gcc-m68k-linux-2.8.1' make: *** [build] Error 2 I get the exact same error if I use dpkg-buildpackage. I don't know enough about building gcc to know what's wrong. And this is the source package from ftp.ca.debian.org (us.debian.org not responding to me through lynx for some reason) I also found out I need the binutils-m68k-linux package, and that's being downloaded as I type.. -- Ferret no baka
Re: Cross-compiling on Debian systems?
On Thu, 18 Nov 1999, Erik Mouw wrote: On Wed, 17 Nov 1999 19:52:42 -0800 (PST), [EMAIL PROTECTED] wrote: On Wed, 17 Nov 1999, Ben Collins wrote: On Wed, Nov 17, 1999 at 02:48:55PM -0800, [EMAIL PROTECTED] wrote: I'd like to set up a cross-compiler on my Lintel (AMD 266) to compile packages for my Sparc IPX. I'd preferably like to package the cross-compiler so I can esaily remove it if I no longer need it. The only cross-compiler in the slink packages list is for m68k. What do I need to do to set one up? The last cross-compiler I built (for kernel compiles) was using the m68k cross-compiler package. I just changed the m68k references to sparc, and ran dpkg-buildpackage -us -uc -rfakeroot (change all of the m68k references in debian/ to sparc). I just did a 'debian/rules binary' in 'gcc-m68k-linux-2.8.1' after changing 'm68k' to 'sparc', and I get a build error. Looks like the gcc buildstrap isn't working. make bootstrap doesn't work for cross compilers, have a look at the cross gcc FAQ: http://www.objsw.com/CrossGCC/ . How was the package suppsed to build itself then, by hand? *g* ./xgcc -B./ -DCROSS_COMPILE -DIN_GCC-O2 -I./include -I. -I. \ -I./config -g0 \ -finhibit-size-directive -fno-inline-functions -fno-exceptions \ -c ./crtstuff.c -DCRT_BEGIN -o crtbegin.o as: option `-s' is ambiguous make[1]: *** [crtbegin.o] Error 1 make[1]: Leaving directory `gcc-m68k-linux-2.8.1' make: *** [build] Error 2 I get the exact same error if I use dpkg-buildpackage. I don't know enough about building gcc to know what's wrong. And this is the source package from ftp.ca.debian.org (us.debian.org not responding to me through lynx for some reason) Looks like you're using the native binutils instead of cross binutils. I also found out I need the binutils-m68k-linux package, and that's being downloaded as I type.. I can't get the cross' binutils-sparc (package hacked from binutils-m68k) to build. dpkg-gencontrol -Pdebian/tmp/binutils -pbinutils-sparc-linux dpkg-gencontrol: error: source package has two conflicting values - binutils-sparc-linux and binutils-m68k-linux make: *** [binary-arch] Error 29 I edited all the references in debian/* from m68k' to sparc', and I don't know enough about the package to even guess what else might need to be tweaked. You also need libc and kernel header files for the target system. The kernel has to be configured for the target system, but that can be done on the build host. Erik -- J.A.K. (Erik) Mouw, Information and Communication Theory Group, Department of Electrical Engineering, Faculty of Information Technology and Systems, Delft University of Technology, PO BOX 5031, 2600 GA Delft, The Netherlands Phone: +31-15-2785859 Fax: +31-15-2781843 Email [EMAIL PROTECTED] WWW: http://www-ict.its.tudelft.nl/~erik/
Re: Cross-compiling on Debian systems?
On Thu, 18 Nov 1999, Ben Collins wrote: On Thu, Nov 18, 1999 at 02:38:26PM +0100, Erik Mouw wrote: On Wed, 17 Nov 1999 19:52:42 -0800 (PST), [EMAIL PROTECTED] wrote: ./xgcc -B./ -DCROSS_COMPILE -DIN_GCC-O2 -I./include -I. -I. \ -I./config -g0 \ -finhibit-size-directive -fno-inline-functions -fno-exceptions \ -c ./crtstuff.c -DCRT_BEGIN -o crtbegin.o as: option `-s' is ambiguous make[1]: *** [crtbegin.o] Error 1 make[1]: Leaving directory `gcc-m68k-linux-2.8.1' make: *** [build] Error 2 I get the exact same error if I use dpkg-buildpackage. I don't know enough about building gcc to know what's wrong. And this is the source package from ftp.ca.debian.org (us.debian.org not responding to me through lynx for some reason) Looks like you're using the native binutils instead of cross binutils. Installing binutils-multiarch and compiling a native gas handles this. I also found out I need the binutils-m68k-linux package, and that's being downloaded as I type.. You also need libc and kernel header files for the target system. The kernel has to be configured for the target system, but that can be done on the build host. Install dpkg-cross and convert the native sparc libc6 and libc6-dev packages. This will install them into /usr/sparc-linux/{include,lib} for the cross-compile. Installed dpkg-cross (isn't listed in the FTP site but a package search brought it up. And the FTP archive doesn't have a source package) :# dpkg-cross --install sysvinit_2.76-3.deb libc6_2.0.105-2.deb \ dpkg_1.4.0.34.deb mount_2.9g-6.deb ldso_1.9.10-1.deb :Building sysvinit-sparc-cross_2.76-3_all.deb :Building libc6-sparc-cross_2.0.105-2_all.deb :Building dpkg-sparc-cross_1.4.0.34_all.deb :dpkg-cross: package mount doesn't provide any useful files. Skipping. This looks funny :Building ldso-sparc-cross_1.9.10-1_all.deb :Unpacking sysvinit-sparc-cross :Unpacking libc6-sparc-cross :Unpacking dpkg-sparc-cross :Unpacking ldso-sparc-cross :dpkg: dependency problems prevent configuration of sysvinit-sparc-cross: : sysvinit-sparc-cross depends on mount-sparc-cross (= 2.7l-1); however: : Package mount-sparc-cross is not installed. :dpkg: error processing sysvinit-sparc-cross (--install): : dependency problems - leaving unconfigured :dpkg: dependency problems prevent configuration of libc6-sparc-cross: : libc6-sparc-cross depends on sysvinit-sparc-cross (= 2.76-1); however: : Package sysvinit-sparc-cross is not configured yet. :dpkg: error processing libc6-sparc-cross (--install): : dependency problems - leaving unconfigured :dpkg: dependency problems prevent configuration of dpkg-sparc-cross: : dpkg-sparc-cross depends on libc6-sparc-cross (= 2.0.99); however: : Package libc6-sparc-cross is not configured yet. : dpkg-sparc-cross depends on libncurses4-sparc-cross; however: : Package libncurses4-sparc-cross is not installed. : dpkg-sparc-cross depends on libstdc++2.9-sparc-cross; however: : Package libstdc++2.9-sparc-cross is not installed. :dpkg: error processing dpkg-sparc-cross (--install): : dependency problems - leaving unconfigured :Setting up ldso-sparc-cross (1.9.10-1) ... :Errors were encountered while processing: : sysvinit-sparc-cross : libc6-sparc-cross : dpkg-sparc-cross :dpkg -i failed. All the Sparc packages were moved using dpkg-move running on the Sparc. I am doing this the right way? -- Ferret no baka
Re: Cross-compiling on Debian systems?
On Thu, 18 Nov 1999, Ben Collins wrote: On Thu, Nov 18, 1999 at 09:37:00AM -0800, [EMAIL PROTECTED] wrote: brought it up. And the FTP archive doesn't have a source package) :# dpkg-cross --install sysvinit_2.76-3.deb libc6_2.0.105-2.deb \ dpkg_1.4.0.34.deb mount_2.9g-6.deb ldso_1.9.10-1.deb :Building sysvinit-sparc-cross_2.76-3_all.deb :Building libc6-sparc-cross_2.0.105-2_all.deb :Building dpkg-sparc-cross_1.4.0.34_all.deb :dpkg-cross: package mount doesn't provide any useful files. Skipping. 1) Do this on the cross-compile system (ie. the i386) 2) You only need libraries and -dev packages. So init, and mount are somewhat irrelevant to compilation :(i386)# dpkg-cross --install libc6_2.0.105-2.deb :Building libc6-sparc-cross_2.0.105-2_all.deb :Unpacking libc6-sparc-cross :dpkg: dependency problems prevent configuration of libc6-sparc-cross: : libc6-sparc-cross depends on sysvinit-sparc-cross (= 2.76-1); however: : Package sysvinit-sparc-cross is not configured yet. :dpkg: error processing libc6-sparc-cross (--install): : dependency problems - leaving unconfigured :Errors were encountered while processing: : libc6-sparc-cross :dpkg -i failed. -- Ferret no baka
Major breakage in Slink..
I've just noticed a few things while installing Debian on my IPX. debhelper is missing several files (dh_list and a couple others) bug won't file bug reports setserial crashes the console (bad if you're running a serial console) setserial tries to initialise to Lintel IRQ settings on sparc (on kernel 2.2.13) something causes a kernel panic/OOPS having to do with the serial ports (in potato) 'login' can't change tty permissions, and man can't change its cat page permissions. And probably others that I haven't yet noticed. What needs to be done to fix stuff? -- Ferret no baka
Re: Major breakage in Slink..
On Wed, 17 Nov 1999, Ben Collins wrote: On Tue, Nov 16, 1999 at 08:02:39PM -0800, [EMAIL PROTECTED] wrote: I've just noticed a few things while installing Debian on my IPX. [snip] (upgrade stuff to potato's version) apt-move wanted too much stuff upgraded (actually broke some things), so I grabbed source and tweaked dependencies. Then I copied stuff from my Lintel slink box to make debhelper work. : setserial crashes the console (bad if you're running a serial console) setserial tries to initialise to Lintel IRQ settings on sparc dpkg --purge setserial Setserial makes little sense on stock sparc systems. Yeah. : (on kernel 2.2.13) something causes a kernel panic/OOPS having to do with the serial ports Haven't noticed this. Perhaps emailing sparclinux@vger.rutgers.edu would be of more help. Mmmm. I have a full console log of one of the 'nice sessions'. Basically go ahead and explain that I'm running a serial console, then send the .config (if I can recreate it) and the console log as attachments? (in potato) 'login' can't change tty permissions, and man can't change its cat page permissions. Login does not change tty permissions for a very good reason, mainly security related. I'm not sure what you mean about man and it's cat pages, I haven't noticed any problems with man at all. I'll check this if I ever to a full upgrade to potato again. Oh, now that I'm actually AWAKE (g), it was saying 'function not implimented'.
Cross-compiling on Debian systems?
I'd like to set up a cross-compiler on my Lintel (AMD 266) to compile packages for my Sparc IPX. I'd preferably like to package the cross-compiler so I can esaily remove it if I no longer need it. The only cross-compiler in the slink packages list is for m68k. What do I need to do to set one up? -- Ferret no baka
Re: 2.2.13 cannot load modules
On Mon, 15 Nov 1999 [EMAIL PROTECTED] wrote: I just compiled and booted 2.2.13 (sun4c, Sparc IPX) on a BARE Slink system (base dbootstrap install plus potato's apt, plus kernel-package+gcc) and saw a couple interesting problems. Firstly when I booted the machine to 2.2.13 for the first time I got several hundred lines of errors saying something like 'operation not permitted on /dev/cua0' and then a panic in the swapper process. So far this problem hasn't been reproducable. I am booting with console on serial port A, so it would already be in use. The main problem is that insmod says for my modules 'ELF file not for this archetecture' and fails to load anything. Update: I just installed and ran 'file' on my modules: ELF 32-bit MSB relocatable, SPARC, version1, not stripped Eh? I did compile on the Sparc, but had the kernel source tree NFS-mounted. Any ideas? -- Ferret no baka -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Q: External modem on Sparc
I'd wanted to use my USR 56K external modem on my Sparc IPX and use it for IP masquerading. Unfortunately the mac modem cable doesn't seem to work on it. When I'm talking to the modem from minicom, both RX and TX lights flash in unison and I don't get an OK or anything out of the modem. I'll try things as I can think of them, but does anyone have ideas? -- Ferret no baka
Re: Update: serial console on Sparc IPX
On Sat, 13 Nov 1999 [EMAIL PROTECTED] wrote: It turns out that the 'mac printer' cable does the crossover already. I connected a second null-modem adapter inline with my cable and did get serial output from the IPX. I'm posting the entire output below. Unfortunately at this point it doesn't seem to respond to input from the serial port (I'm running minicom on the Lintel end of things) Found that turning off hardware flow control fixed this particular problem. What keycode do I need to send to the IPX to get the openprom to break for me, being L1-S obviously can't be sent over? Well, I had thought it was break, but that was crashing minicom. Turning off hardware flow control fixed this too. Silly ferret. Am I reading the selftest messages correctly and thinking it wants parity memory and it's erroring because it has non-parity? -- Ferret no baka
Update: serial console on Sparc IPX
It turns out that the 'mac printer' cable does the crossover already. I connected a second null-modem adapter inline with my cable and did get serial output from the IPX. I'm posting the entire output below. Unfortunately at this point it doesn't seem to respond to input from the serial port (I'm running minicom on the Lintel end of things) What keycode do I need to send to the IPX to get the openprom to break for me, being L1-S obviously can't be sent over? Am I reading the selftest messages correctly and thinking it wants parity memory and it's erroring because it has non-parity? -- Ferret no baka Power-Up State Test Context Register Bit Test Context Register Addressing Test Segment Map RAM MATS Pattern Test, Context Segment Map RAM MATS Pattern Test, Context 0001 Segment Map RAM MATS Pattern Test, Context 0002 Segment Map RAM MATS Pattern Test, Context 0003 Segment Map RAM MATS Pattern Test, Context 0004 Segment Map RAM MATS Pattern Test, Context 0005 Segment Map RAM MATS Pattern Test, Context 0006 Segment Map RAM MATS Pattern Test, Context 0007 Page Map RAM MATS Pattern Test Limit 0 Register Test Counter Interrupt Level 10 Test Limit 1 Register Test Counter Interrupt Level 14 Test WARNING: Unable to determine keyboard type Synchronous Error Reg Test Synchronous Error Virtual Address Reg Test Asynchronous Error Reg Test Asynchronous Error Virtual Address Reg Test Asynchronous Error Data Reg1 Test Asynchronous Error Data Reg2 Test System Enable Register Bit Test Cache Data RAM MATS Pattern Test Cache Tag RAM MATS Pattern Test PTE Access Bit Test PTE Modify Bit Test PTE Write-Protect Bit Test PTE Write-Invalid Bit Test PTE Read-Invalid Bit Test PTE Type 2 Space Bit Test PTE Type 3 Space Bit Test Synchronous Timeout Test Asynchronous Timeout Test 16 MegaBytes Found in Address Range 0x to 0x00ff DRAM Word MATS Pattern Test (0x00fe - 0x0100) Parity/Memory Control Registers Bit Test 33-bit SIMM Parity Test FAILURE: No Trap Taken, Exp Trap Type = 0009 Setting Segment Map Setting RAM Parity Mode Mode set to 33-bit Sizing Memory Mapping ROM Mapping RAM Probing /[EMAIL PROTECTED],f800 at 0,0 dma esp sd st le Probing /[EMAIL PROTECTED],f800 at 1,0 Nothing there Probing /[EMAIL PROTECTED],f800 at 2,0 Nothing there Probing /[EMAIL PROTECTED],f800 at 3,0 cgsix Probing /[EMAIL PROTECTED],f800 at 0,0 dma esp sd st le Probing /[EMAIL PROTECTED],f800 at 1,0 Nothing there Probing /[EMAIL PROTECTED],f800 at 2,0 Nothing there Probing /[EMAIL PROTECTED],f800 at 3,0 cgsix SPARCstation IPX, No Keyboard ROM Rev. 2.6, 16 MB memory installed, Serial #2165154. Ethernet address 8:0:20:c:dc:80, Host ID: 572109a2. Power-On SelfTest FAILED ... Replace CPU Board Testing 16 megs of memory 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 SBus slot 0 le esp dma SBus slot 1 SBus slot 2 SBus slot 3 cgsix Initializing 16 megs of memory at addr 0 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 Type b (boot), c (continue), or n (new command mode)
Re: Serial console on IPX (Was: Sparc serial port pinout?)
Okay, following reccomendations and my own best guessing I went out and purchased a 'Mac to serial printer' cable and a null-modem adapter. I plug the cables in, fire up minicom on the Lintel (set to 9600N81 as I remember seeing in the IPX firmware), and turn on the IPX. Power comes on, the drive spins up, but I get nothing. Now, what comes to mind is: Wrong cabling. Wrong serial port settings on the Lintel. Bad serial ports on the IPX. IPX not using serial console. The null-modem adapter is a standard Laplink-comatible cable. The mac cable is a Fellowes part # 99519 (no idea about the pinout on it) I have no 'normal' console on the IPX so I can't change settings unless I get serial console working at this point. If I can rule out the serial cabling I'll probably go ahead and get the keyboard and video adapters instead. Oh yeah, I assume the serial cable should go in port 'A', but I tried it in 'B' anyway and didn't get anything there either. On Fri, 5 Nov 1999, Will Lowe wrote: One more question. If it doesn't have a keyboard and monitor plugged in, will it put the console out the serial port automagically? It should. I know Suns do that automagically, and I know solaris does, and I know Debian is supposed to, but I haven't tried it yet, so I can't say for certain under Debian. yes, this works. It's a sun hardware thing, rather than a software issue, methinks.
Re: Serial console on IPX (Was: Sparc serial port pinout?)
I don't know of anyone at all in the town I live in who would have the right equipment. I had access to a hard at the CMU campus at the time that I picked up the machine. I live in a fairly small town, and the closest city is about 75 miles away. So, unless I can actually scare up a temporary head locally (perhaps at one of the local ISPs) 'borrowing' a head will not be an option. On Sat, 6 Nov 1999, Walter Keeler wrote: Can a sysadmin friend take you into his jobsite some weekend? You don't necessarily need full-time access to monitor/keyboard/mouse, just long enough to be sure you're set up well enough to do further work headless. Also, having at least part-time access to a head makes recompiling kernels, etc., FAR less painful. (Believe me, I know!) On Sat, 6 Nov 1999 [EMAIL PROTECTED] wrote: On Sat, 6 Nov 1999, Walter Keeler wrote: Here's a basic question: What happens when you unplug the serial cable and then power up the IPX? Does it spin up and make disk access noises? Unplugged cable, powered on IPX. Disk spins up. If so, it sure sounds like a cabling issue of some sort. It would make the debugging job a whole lot easier to fire up the IPX with a head, if only to make sure it runs properly to begin with. Adding a head would put me back about $150 unless I could find a less expensive supplier for the keyboard and video converters, and then I wouldn't need serial console. Maybe that would be a better track? Sun sells a ps2 keyboard and mouse converter for $75, and Blackbox has a 13W3/VGA converter for about $45. I would love a head on this thing if it wouldn't be quite so expensive. Oh, and I have no desk/counter/shelf space for a second keyboard and mouse at this time. However, I did finally chase up a pinout for the serial port, and I might have ended up with either the wrong cable (might need a modem cable instead of a printer cable) or I might need a straight instead of a null modem cable. Unfortunately I wont be able to test any of those for the next two or three days.
Re: Serial console on IPX (Was: Sparc serial port pinout?)
On Sun, 7 Nov 1999 [EMAIL PROTECTED] wrote: Okay, following reccomendations and my own best guessing I went out and purchased a 'Mac to serial printer' cable and a null-modem adapter. Ooh, that sounds like a bad combo to me. A null modem should twist the send and receive lines so that the send from one machine goes to the receive of the other. If your Mac cable is designed to go to a printer with a standard RS232 port, then it probably already does that twist, and attaching another null modem adapter to it does a double twist, putting you back with send going to send, and receive going to receive. That's a combination that's guaranteed to fail. I'll try a straight-through as soon as I can get one. Either that or I'll move the IPX closer to the Lintel and just use a sex changer. The simplest way to make sure you have the right cabling is to make your own cables! I have a tendency to mess up my own cables, alas. :/
Re: [Nigel R. Armfield N.R.Armfield@lmu.ac.uk] Bug#49172: Boot problems
On 6 Nov 1999, Adam Di Carlo wrote: This sounds familiar -- anyone have a pointer here? I'm not sure. I've successfully installed an IPX using floppy using the 2.0.3x sparc disk images.. If need be I could read my disks and let you FTP them. I no longer have access to the machine I wrote them on, though, or I could send them to you from there. -- Ferret no baka
Re: Sparc serial port pinout?
On Thu, 4 Nov 1999, Kurt Mosiejczuk wrote: On Thu, 4 Nov 1999 [EMAIL PROTECTED] wrote: I want to build a null-modem cable to go between my SparcStation and my Lintel box, and my Sparc has the round 8-pin mini-DINs. These wouldn't be signal-compatible with Mac serial ports, would they? Alternately, are there any companies besides Sun who make such a thing? Yes, they are signal-compatible. Whether Mac cables will work for high speed modems is often debated, but I have been using a Mac cable for a while now and things seem fine. I do also have an acutal Sun cable too... and I haven't seen a difference, and I run it with a 33600 modem. Great! One more question. If it doesn't have a keyboard and monitor plugged in, will it put the console out the serial port automagically? -- Ferret no baka
Sparc serial port pinout?
I want to build a null-modem cable to go between my SparcStation and my Lintel box, and my Sparc has the round 8-pin mini-DINs. These wouldn't be signal-compatible with Mac serial ports, would they? Alternately, are there any companies besides Sun who make such a thing? -- Ferret no baka
Re: APT
On Sun, 24 Oct 1999, Jokker wrote: I remember awhile back there being problems with APT for sparc getting stuck in infinite loops...has this been resolved? or should I go for an older distrib? I am trying to do an ftp install I've had this exact same problem after updating my 'new' Sparc IPX to Slink, and had to unpack the apt executables and one library from the installation tarball to get apt working again.
Sparc Newbie looking for information...
Hello, I've recently picked up a used Sparc IPX machine and gotten Slink installed on it. I'm interested in any online information about the machine and what it's capible of, hardware-wise. It has an sbus card in it that I haven't been able to identify by FCC number, and I'd like to know how to find out what it is. The boot rom identifies it as a sunw, bpp sunw, lpvi Also, the board selftest fails, and asks that the CPU board be removed. Is there any diagnostic software available that could tell me where and what the failure is, rather than just there is a failure? And is there such a thing as a converter to use a PC keyboard and mouse on an IPX workstation?