Re: Potato->Woody
Hi, Anyone tried SGI's xfs filesystem on sparc? Thanks in advance, Peter Firmstone. On Mon, 12 Mar 2001, Ragga Muffin wrote: > > Ben Collins <[EMAIL PROTECTED]> wrote: > > > On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > > - Are there any issues with 2.4.x on sun4m ? > > A few, but you don't need kernel 2.4.x to run woody. > > 2.4.2-ac18 would not compile on sparc/woody . Do other > 2.4.x kernels compile fine ? > > > > - Any caveats with reiserfs ? > > Yeah, it does not work on sparc (only works on i386, and maybe alpha by > > now). > > Well, reiserfs, iptables and a better vm/mm were my hopes with 2.4.x. > Guess I'll wait for now. > > > The potato->woody dist-upgrade went mostly fine. Apart from setting up XF4 > it was a breeze. (kudos to all!) > > However, at boot I now get these messages (2.2.19pre16): > > Adding Swap: 65396k swap-space (priority -1) > cp[171]: Unimplemented SPARC system call 69 > chown[206]: Unimplemented SPARC system call 35 > mv[209]: Unimplemented SPARC system call 69 > portmap[231]: Unimplemented SPARC system call 87 > automount[272]: Unimplemented SPARC system call 69 > ps[274]: Unimplemented SPARC system call 69 > > They don't seem to be fatal though, any idea ? > > TIA, > > Ragga > > > -- > To UNSUBSCRIBE, email to [EMAIL PROTECTED] > with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED] > >
Re: Potato->Woody
On Wed, Mar 14, Ragga Muffin wrote: > > Ben Collins <[EMAIL PROTECTED]> wrote: > > > On Mon, Mar 12, 2001 at 04:38:16PM +0900, Ragga Muffin wrote: > > > > > > Ben Collins <[EMAIL PROTECTED]> wrote: > > > > > > > On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > > > > - Are there any issues with 2.4.x on sun4m ? > > > > A few, but you don't need kernel 2.4.x to run woody. > > > > > > 2.4.2-ac18 would not compile on sparc/woody . Do other > > > 2.4.x kernels compile fine ? > > > > You probably want the CVS source from vger.samba.org > > Just cvs'ed the latest, and > > ... > gcc -D__KERNEL__ -I/usr/local/src/linux-2.4.3pre/include -Wall > -Wstrict-prototypes -O2 -fomit-frame-pointer -fno-strict-aliasing -m32 -pipe > -mno-fpu -fcall-used-g5 -fcall-used-g7-c -o sun4c.o sun4c.c > sun4c.c: In function `sun4c_paging_init': > sun4c.c:2505: `highend_pfn' undeclared (first use in this function) > sun4c.c:2505: (Each undeclared identifier is reported only once > sun4c.c:2505: for each function it appears in.) > sun4c.c:2507: warning: implicit declaration of function `calc_highpages' I posted a fix on the sparclinux@vger.kernel.org list for one/two days. > > > On Mon, Mar 12, 2001 at 08:36:31PM +0100, Thorsten Kukuk wrote: > > > swafnir:~ # uname -a > > > Linux swafnir 2.4.2 #10 Mon Mar 12 20:16:22 CET 2001 sparc unknown > ^ > Is the loopfs broken on sparc too ? Yes, it looks like so, too. Thorsten -- Thorsten Kukuk http://www.suse.de/~kukuk/ [EMAIL PROTECTED] SuSE GmbHSchanzaeckerstr. 1090443 Nuernberg Linux is like a Vorlon. It is incredibly powerful, gives terse, cryptic answers and has a lot of things going on in the background.
Re: Abstraction (Was: Re: Potato->Woody)
Onno Benschop wrote: > > Ok, ok, I'll bite. Don't you mean `bark'? ~:^) >> So, then someone asks, how come reiserfs doesn't work on sparc. > > I wonder what am I missing? A lot. First, get a book on Unix kernel development and read it completely. Then, design and write a journaling filesystem for any Unix and see how long it takes for you to get a version that runs for longer than ten seconds without panicing the kernel. After you get it to run for several minutes without crashing, then you will already know the answers to this and many other questions, Grasshopper. ~:^) Oh yeah, and don't forget that you have to port every utility that has knowledge of the file system structure, including, but not limited to: mkfs, fsck, dump, restore. Second, someone posted a message showing that it does run on SPARC. Third, a file system is responsible for the meaning and content of it's on-disk data structures; the disk driver layers have nothing to do with that. If the filesystem code inadvertently assumes a certain endianness with respect to how the bytes come off the disk and are placed into words, then things can get broken on machines where that assumption isn't correct. It's a lot easier than you might think for this kind of error to occur. Getting any piece of kernel code to work starts by getting it to work on a specific platform first, then porting/testing it on other platforms as time/demand permits. You might try digging into the list of current bugs for reiserfs to become acquainted with the level of difficulty of the topic you are thinking about while sticking pencils in the ceiling tiles. ~:^) a
Re: Abstraction (Was: Re: Potato->Woody)
On Wed, Mar 14, 2001 at 10:00:20AM +0800, Onno Benschop wrote: > > So, then someone asks, how come reiserfs doesn't work on sparc. > > I wonder what am I missing? This is what's known as bad, non-portable coding. You see, on different architectures, things are, well, different. We have big and little endians, and we have 32bit and 64bit. When you code something, and don't have the foresight to work with these nuances, you get problems down the road. You see, someone thought it would be easier to "just get it working, sort out the portability later" than to "think now, do it right, and save mind later". Of course, Linus once said that Linux would never work on anything other than i386 too :) Ben -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Abstraction (Was: Re: Potato->Woody)
Ok, ok, I'll bite. In order to further my understanding of my place in the world (that is, ask silly questions, see what happens), I feel the need to ask this (from blissful ignorance, so be gentle). Let us imagine an operating system that is built from a whole lot of different modules that all talk to each other. Sounds a little like Linux - no? So someone comes along and invents a new file system. The developer community says "Cool" and life moves on. The question is coming, be patient :-) So in the scheme of things, some bright spark wrote the bit that speaks to the disk drive - known as a driver. This person's goal in life is to make their little driver be the fastest, bestest, smartest, tiniest piece of code you can imagine. This is what they do in life, this is why they get out of bed - you get the picture. Their driver takes all the hassle out of writing bits to disk. Question just over the horizon... Then the inventor of the new file system just writes their little module, so it talks to the disk-driver. So when someone else comes along and writes a disk-driver for their blue-spangled-magneto-laser-disk thingy, the new file system still talks to a disk-driver, and it all still works. So, then someone asks, how come reiserfs doesn't work on sparc. I wonder what am I missing? -- ()/)/)()..ASCII for Onno.. |>>? ..EBCDIC for Onno.. --- -. -. --- ..Morse for Onno.. ITmaze - ABN: 56 178 057 063 - ph: 04 1219 - [EMAIL PROTECTED]
Re: Potato->Woody
Ben Collins <[EMAIL PROTECTED]> wrote: > On Mon, Mar 12, 2001 at 04:38:16PM +0900, Ragga Muffin wrote: > > > > Ben Collins <[EMAIL PROTECTED]> wrote: > > > > > On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > > > - Are there any issues with 2.4.x on sun4m ? > > > A few, but you don't need kernel 2.4.x to run woody. > > > > 2.4.2-ac18 would not compile on sparc/woody . Do other > > 2.4.x kernels compile fine ? > > You probably want the CVS source from vger.samba.org Just cvs'ed the latest, and ... gcc -D__KERNEL__ -I/usr/local/src/linux-2.4.3pre/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -fno-strict-aliasing -m32 -pipe -mno-fpu -fcall-used-g5 -fcall-used-g7-c -o sun4c.o sun4c.c sun4c.c: In function `sun4c_paging_init': sun4c.c:2505: `highend_pfn' undeclared (first use in this function) sun4c.c:2505: (Each undeclared identifier is reported only once sun4c.c:2505: for each function it appears in.) sun4c.c:2507: warning: implicit declaration of function `calc_highpages' make[2]: *** [sun4c.o] エラー 1 make[2]: Leaving directory `/usr/local/src/linux-2.4.3pre/arch/sparc/mm' make[1]: *** [first_rule] エラー 2 make[1]: Leaving directory `/usr/local/src/linux-2.4.3pre/arch/sparc/mm' make: *** [_dir_arch/sparc/mm] エラー 2 > > On Mon, Mar 12, 2001 at 08:36:31PM +0100, Thorsten Kukuk wrote: > > swafnir:~ # uname -a > > Linux swafnir 2.4.2 #10 Mon Mar 12 20:16:22 CET 2001 sparc unknown ^ Is the loopfs broken on sparc too ? It's quite essential that we be able to export iso images... Tnx guys, -- Ragga
Re: Potato->Woody
* on the Tue, Mar 13, 2001 at 08:20:19AM +0100, Andreas Tille was blubbering: > Just for the sake of interest: How far depends filesystem code from the > processor architecture? - Endianness - 32bit/64bit ... (and counting ;) These would be the most prominent, I guess. Peter -- "Any good Unix security engineer can clean up any Unix box. But I'm not sure there are people even within Microsoft who know how to clean up an NT box." -- Michael Zbouray
Re: Potato->Woody
On Mon, 12 Mar 2001, Ben Collins wrote: > Ok, let me rephrase, I doubt it will work on ultrasparc :) Nice to see > it has made some progress since the last time I checked into it where I > was told it wasn't going to happen in the forseeable future. Just for the sake of interest: How far depends filesystem code from the processor architecture? I just planed to use Reiserfs in the next month on an Ultra-Sparc but this discussion let me rethink my plans. Do you think ext3fs is an option and where can I get kernel-patches working with Ultra-Sparc safely. Kind regards Andreas.
Re: Potato->Woody
On Mon, Mar 12, 2001 at 08:36:31PM +0100, Thorsten Kukuk wrote: > On Mon, Mar 12, Ben Collins wrote: > > > > > > - Any caveats with reiserfs ? > > > > Yeah, it does not work on sparc (only works on i386, and maybe alpha by > > > > now). > > > > > > Well, reiserfs, iptables and a better vm/mm were my hopes with 2.4.x. > > > Guess I'll wait for now. > > > > I doubt we'll ever see reiser on sparc anytime soon. > > swafnir:~ # uname -a > Linux swafnir 2.4.2 #10 Mon Mar 12 20:16:22 CET 2001 sparc unknown > Ok, let me rephrase, I doubt it will work on ultrasparc :) Nice to see it has made some progress since the last time I checked into it where I was told it wasn't going to happen in the forseeable future. Ben -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Re: Potato->Woody
On Mon, Mar 12, Ben Collins wrote: > > > > - Any caveats with reiserfs ? > > > Yeah, it does not work on sparc (only works on i386, and maybe alpha by > > > now). > > > > Well, reiserfs, iptables and a better vm/mm were my hopes with 2.4.x. > > Guess I'll wait for now. > > I doubt we'll ever see reiser on sparc anytime soon. swafnir:~ # uname -a Linux swafnir 2.4.2 #10 Mon Mar 12 20:16:22 CET 2001 sparc unknown swafnir:~ # cat /proc/filesystems nodev sockfs nodev shm nodev pipefs nodev proc ext2 minix iso9660 reiserfs nodev nfs nodev devpts swafnir:~ # mount /dev/sda4 on / type ext2 (rw) proc on /proc type proc (rw) devpts on /dev/pts type devpts (rw,mode=0620,gid=5) /dev/sda1 on /boot type ext2 (rw) /dev/sdc8 on /abuild type ext2 (rw) shmfs on /dev/shm type shm (rw) /dev/sdb4 on /mnt type reiserfs (rw) swafnir:~ # ls -alF /mnt/ total 24709 drwxr-xr-x3 root root 90 Mar 12 20:29 ./ drwxr-xr-x 27 root root 4096 Mar 6 18:10 ../ drwxr-xr-x 12 root root 381 Mar 12 20:31 linux/ -rw-r--r--1 root root 25295341 Feb 22 01:00 linux-2.4.2.tar.gz I don't know how stable it is, but at least I can untar a kernel and compile it ;-) This is a 2.4.2 kernel with the PowerPC reiserfs patch. I haven't tried it on a UltraSPARC yet, will follow tomorrow. Thorsten -- Thorsten Kukuk http://www.suse.de/~kukuk/ [EMAIL PROTECTED] SuSE GmbHSchanzaeckerstr. 1090443 Nuernberg Linux is like a Vorlon. It is incredibly powerful, gives terse, cryptic answers and has a lot of things going on in the background.
Re: Potato->Woody
On Mon, Mar 12, 2001 at 04:38:16PM +0900, Ragga Muffin wrote: > > Ben Collins <[EMAIL PROTECTED]> wrote: > > > On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > > - Are there any issues with 2.4.x on sun4m ? > > A few, but you don't need kernel 2.4.x to run woody. > > 2.4.2-ac18 would not compile on sparc/woody . Do other > 2.4.x kernels compile fine ? You probably want the CVS source from vger.samba.org > > > - Any caveats with reiserfs ? > > Yeah, it does not work on sparc (only works on i386, and maybe alpha by > > now). > > Well, reiserfs, iptables and a better vm/mm were my hopes with 2.4.x. > Guess I'll wait for now. I doubt we'll ever see reiser on sparc anytime soon. Depending on your needs you may want to check out jfs, xfs and ext3. I use ext3 mainly for the large RAID's so the fsck time is gone. You may want performance, which would take xfs or jfs (never used either one). > The potato->woody dist-upgrade went mostly fine. Apart from setting up XF4 > it was a breeze. (kudos to all!) Excellent! > However, at boot I now get these messages (2.2.19pre16): > > Adding Swap: 65396k swap-space (priority -1) > cp[171]: Unimplemented SPARC system call 69 > chown[206]: Unimplemented SPARC system call 35 > mv[209]: Unimplemented SPARC system call 69 > portmap[231]: Unimplemented SPARC system call 87 > automount[272]: Unimplemented SPARC system call 69 > ps[274]: Unimplemented SPARC system call 69 > > They don't seem to be fatal though, any idea ? Just ignore them. This is a LFS enabled system. Those syscalls are the libc trying to execute LFS syscalls, which the 2.2.x kernel's don't support. It falls back to the normal syscalls though. -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Re: Potato->Woody
Ben Collins <[EMAIL PROTECTED]> wrote: > On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > - Are there any issues with 2.4.x on sun4m ? > A few, but you don't need kernel 2.4.x to run woody. 2.4.2-ac18 would not compile on sparc/woody . Do other 2.4.x kernels compile fine ? > > - Any caveats with reiserfs ? > Yeah, it does not work on sparc (only works on i386, and maybe alpha by > now). Well, reiserfs, iptables and a better vm/mm were my hopes with 2.4.x. Guess I'll wait for now. The potato->woody dist-upgrade went mostly fine. Apart from setting up XF4 it was a breeze. (kudos to all!) However, at boot I now get these messages (2.2.19pre16): Adding Swap: 65396k swap-space (priority -1) cp[171]: Unimplemented SPARC system call 69 chown[206]: Unimplemented SPARC system call 35 mv[209]: Unimplemented SPARC system call 69 portmap[231]: Unimplemented SPARC system call 87 automount[272]: Unimplemented SPARC system call 69 ps[274]: Unimplemented SPARC system call 69 They don't seem to be fatal though, any idea ? TIA, Ragga
Re: Potato->Woody
On Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin wrote: > > Hi everyone, > > I'd like some input about upgrading an SS5 from potato to woody. > > Specifically: > - Are there any issues with 2.4.x on sun4m ? A few, but you don't need kernel 2.4.x to run woody. > - Any caveats with reiserfs ? Yeah, it does not work on sparc (only works on i386, and maybe alpha by now). > - How about XF4.0.2 with a cgsix ? That should work. > Also, what's the status of the installer package in woody ? > I didn't find anything looking like boot-floppies or tftp images > under testing or woody... There are no boot floppies for woody yet. If you want a clean install, use the potato boot floppies, and point apt towards woody. > If they exist somewhere, then instead of apt-upgrading, would testing > the new installer be of greater help ? I'll email here first, when I have some boot floppies ready for woody. -- ---===-=-==-=---==-=-- / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=--===-=-=-=-===-==---=--=---'
Re: Potato->Woody
* on the Fri, Mar 09, 2001 at 02:38:25PM +0900, Ragga Muffin was blubbering: > > Hi everyone, > > I'd like some input about upgrading an SS5 from potato to woody. > > Specifically: > - Are there any issues with 2.4.x on sun4m ? Nonworking on my SS10. Doesn't go over booting Linux > - Any caveats with reiserfs ? Apparently doesn't work on big-endian machines. So no reiserfs for Sparc/32. > - How about XF4.0.2 with a cgsix ? No idea, mine is headless. > Also, what's the status of the installer package in woody ? > I didn't find anything looking like boot-floppies or tftp images > under testing or woody... Use the potato ones, and then give woody as source for apt. Cheers Peter -- "Any good Unix security engineer can clean up any Unix box. But I'm not sure there are people even within Microsoft who know how to clean up an NT box." -- Michael Zbouray