Re: powerpc woody bf installation status
On Thu, May 17, 2001 at 02:32:44PM -0600, Matt Kraai wrote: On Thu, May 17, 2001 at 02:55:39PM -0500, Stephen R Marenka wrote: Well actually I meant chroot /target or chroot target gives me the error: chroot: USER=root: No such file or directory I tried using the chroot that was installed with base and it worked fine. :-( It looks to me like chroot is invoking the builtin shell, sees input of USER=root ... and tries to execute the file 'USER=root', which fails and causes the message you see. The reason it displays chroot: is that the applet_name is not properly set when calling shell_main. I don't know who is feeding it the USER=root command. It apparently got fixed in cvs (or whatever was weird is gone). Thanks, Stephen -- Stephen R. Marenka If life's not fun, you're not doing it right! [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Wed, May 16, 2001 at 07:00:35PM -0700, Matt Kraai wrote: On Wed, May 16, 2001 at 08:41:09PM -0500, Stephen R Marenka wrote: On Wed, May 16, 2001 at 05:31:03PM -0700, Matt Kraai wrote: It appears that sed.c was in an indeterminate state when the sed fix backport occurred. The appended patch backports some more fixes and takes care of this problem. Time for 0.51-5? Matt Would you check chroot first? I'm not sure if it was something weird I was doing or if it really was broken (I was using 0.52pre out of cvs). I assume that you are refering to it segfaulting when passed no command is specified? It looks like it calls shell_main with argc=0, argv={NULL}, which the shell doesn't like. I don't think this should be a problem for the boot-floppies since we don't include the BusyBox shell so it shouldn't call this code. Well actually I meant chroot /target or chroot target gives me the error: chroot: USER=root: No such file or directory I tried using the chroot that was installed with base and it worked fine. :-( Thanks, Stephen -- Stephen R. Marenka If life's not fun, you're not doing it right! [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Thu, May 17, 2001 at 02:55:39PM -0500, Stephen R Marenka wrote: Well actually I meant chroot /target or chroot target gives me the error: chroot: USER=root: No such file or directory I tried using the chroot that was installed with base and it worked fine. :-( It looks to me like chroot is invoking the builtin shell, sees input of USER=root ... and tries to execute the file 'USER=root', which fails and causes the message you see. The reason it displays chroot: is that the applet_name is not properly set when calling shell_main. I don't know who is feeding it the USER=root command. Matt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Mon, May 14, 2001 at 01:51:45PM -0600, Matt Kraai wrote: On Mon, May 14, 2001 at 01:35:20PM -0600, Erik Andersen wrote: On Mon May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: Currently using a 0.50 build with the md5sum fix in it, but in that sed doesn't work, sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' $reldest only spits out the md5sum: line of the release file and none of the following lines. Again, havn't checked cvs yet. Could you send me what $reldest expands to so I can test this and fix it if needed? I think that this was bug 91758, which was fixed in 0.51-1. And I think $reldest expands to one of the Release files, such as http://http.us.debian.org/debian/dists/woody/Release . Now using 0.51-4 source, on hppa, and sed still appears to be broken. Now it isn't appending '\n' to its output lines, so all output appears as one continuous line: Test file: === cut === Origin: Debian Label: Debian Suite: testing Codename: woody Date: Fri, 11 Apr 2001 17:00:00 UTC Architectures: hppa Components: main Description: Debian HPPA Testint distribution - Not Released md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz === cut === richard@slab:~/bin$ cat Release| sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz This would appear to be because a call to chomp() has been added to sed.c:process_file(). Richard -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Thu, May 17, 2001 at 12:47:06AM +0100, Richard Hirst wrote: On Mon, May 14, 2001 at 01:51:45PM -0600, Matt Kraai wrote: On Mon, May 14, 2001 at 01:35:20PM -0600, Erik Andersen wrote: On Mon May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: Currently using a 0.50 build with the md5sum fix in it, but in that sed doesn't work, sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' $reldest only spits out the md5sum: line of the release file and none of the following lines. Again, havn't checked cvs yet. Could you send me what $reldest expands to so I can test this and fix it if needed? I think that this was bug 91758, which was fixed in 0.51-1. And I think $reldest expands to one of the Release files, such as http://http.us.debian.org/debian/dists/woody/Release . Now using 0.51-4 source, on hppa, and sed still appears to be broken. Now it isn't appending '\n' to its output lines, so all output appears as one continuous line: Test file: === cut === Origin: Debian Label: Debian Suite: testing Codename: woody Date: Fri, 11 Apr 2001 17:00:00 UTC Architectures: hppa Components: main Description: Debian HPPA Testint distribution - Not Released md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz === cut === richard@slab:~/bin$ cat Release| sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz This would appear to be because a call to chomp() has been added to sed.c:process_file(). It appears that sed.c was in an indeterminate state when the sed fix backport occurred. The appended patch backports some more fixes and takes care of this problem. Time for 0.51-5? Matt diff -ur busybox-0.51/sed.c busybox-0.51-5/sed.c --- busybox-0.51/sed.c Mon May 14 13:12:48 2001 +++ busybox-0.51-5/sed.cWed May 16 17:26:24 2001 @@ -577,9 +577,7 @@ break; } - /* if there's anything left of the line, print it */ - if (*hackline) - fputs(hackline, stdout); + puts(hackline); /* cleanup */ free(regmatch); @@ -594,7 +592,7 @@ switch (sed_cmd-cmd) { case 'p': - fputs(line, stdout); + puts(line); break; case 'd': -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Thu May 17, 2001 at 12:47:06AM +0100, Richard Hirst wrote: Now using 0.51-4 source, on hppa, and sed still appears to be broken. Now it isn't appending '\n' to its output lines, so all output appears as one continuous line: Test file: === cut === Origin: Debian Label: Debian Suite: testing Codename: woody Date: Fri, 11 Apr 2001 17:00:00 UTC Architectures: hppa Components: main Description: Debian HPPA Testint distribution - Not Released md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz === cut === richard@slab:~/bin$ cat Release| sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' md5sum: e940164e4afe8cd2059629d898efd795 81 main/binary-hppa/Release 5942a65bc7c2e735b45a0ced9fa6434967079 main/binary-hppa/Packages f2932b0670399550fcb878050e007a1221317 main/binary-hppa/Packages.gz This would appear to be because a call to chomp() has been added to sed.c:process_file(). It turns out a fix went into CVS for this very thing about 4 hours after I put together 0.51-4. So I guess I need to make a 0.51-5 to fix this one. -Erik -- Erik B. Andersen email: [EMAIL PROTECTED], [EMAIL PROTECTED] --This message was written using 73% post-consumer electrons-- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Wed, May 16, 2001 at 05:31:03PM -0700, Matt Kraai wrote: On Thu, May 17, 2001 at 12:47:06AM +0100, Richard Hirst wrote: On Mon, May 14, 2001 at 01:51:45PM -0600, Matt Kraai wrote: On Mon, May 14, 2001 at 01:35:20PM -0600, Erik Andersen wrote: On Mon May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: It appears that sed.c was in an indeterminate state when the sed fix backport occurred. The appended patch backports some more fixes and takes care of this problem. Time for 0.51-5? Matt Would you check chroot first? I'm not sure if it was something weird I was doing or if it really was broken (I was using 0.52pre out of cvs). Thanks, Stephen -- Stephen R. Marenka If life's not fun, you're not doing it right! [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Wed, May 16, 2001 at 08:41:09PM -0500, Stephen R Marenka wrote: On Wed, May 16, 2001 at 05:31:03PM -0700, Matt Kraai wrote: It appears that sed.c was in an indeterminate state when the sed fix backport occurred. The appended patch backports some more fixes and takes care of this problem. Time for 0.51-5? Matt Would you check chroot first? I'm not sure if it was something weird I was doing or if it really was broken (I was using 0.52pre out of cvs). I assume that you are refering to it segfaulting when passed no command is specified? It looks like it calls shell_main with argc=0, argv={NULL}, which the shell doesn't like. I don't think this should be a problem for the boot-floppies since we don't include the BusyBox shell so it shouldn't call this code. Matt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Mon, May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: On Sat, May 12, 2001 at 11:18:06AM -0600, Erik Andersen wrote: I'm building a new one now. Rather then waiting for 0.52 to stabalize, I'm backporting the critical bugfixes into 0.51. Is that 0.51-3? You havn't picked up the md5sum fix, which is necessary for big endian: slab:/usr/src/x# diff busybox-0.51/md5sum.c ../busybox-0.51/md5sum.c 184c184,190 #define SWAP(n) (n) --- /* Handle endian-ness */ #if __BYTE_ORDER == __LITTLE_ENDIAN #define SWAP(n) (n) #else #define SWAP(n) ((n 24) | ((n65280)8) | ((n16711680)8) | (n24)) #endif Correct, this one didn't make it into the release. It is in CVS though. I built my own from 0.51-1 source last week and found zcat didn't work if you fed it input in stdin - just gave a usage msg. Havn't checked if that is fixed in cvs yet, or if it was a local problem. This is fixed in 0.51-3. Currently using a 0.50 build with the md5sum fix in it, but in that sed doesn't work, sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' $reldest only spits out the md5sum: line of the release file and none of the following lines. Again, havn't checked cvs yet. As is this one. If Erik is willing, the following patch against 0.51-3 should fix the SWAP problem and set things up for a 0.51-4 release. Matt diff -ur busybox-0.51/debian/changelog busybox/debian/changelog --- busybox-0.51/debian/changelog Sat May 12 17:47:27 2001 +++ busybox/debian/changelogMon May 14 10:37:38 2001 @@ -1,3 +1,9 @@ +busybox (1:0.51-4) unstable; urgency=high + + * Fix md5sum endianness issue. + + -- Erik Andersen [EMAIL PROTECTED] Mon, 14 May 2001 10:37:35 -0600 + busybox (1:0.51-3) unstable; urgency=low * This is the I am an idiot release. diff -ur busybox-0.51/md5sum.c busybox/md5sum.c --- busybox-0.51/md5sum.c Wed Mar 7 14:45:13 2001 +++ busybox/md5sum.cMon May 14 10:26:44 2001 @@ -181,7 +181,11 @@ //end of md5.h // -#define SWAP(n) (n) +#if __BYTE_ORDER == __LITTLE_ENDIAN +# define SWAP(n) (n) +#else +# define SWAP(n) ((n 24) | ((n65280)8) | ((n16711680)8) | (n24)) +#endif /* This array contains the bytes used to pad the buffer to the next 64-byte boundary. (RFC 1321, 3.1: Step 1) */ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Mon May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: I built my own from 0.51-1 source last week and found zcat didn't work if you fed it input in stdin - just gave a usage msg. Havn't checked if that is fixed in cvs yet, or if it was a local problem. That is fixed in 0.51-3. 0.51-3 is basically 0.51 plus some bugfixes backported from CVS. I'll be making a 0.51-4 in a few minutes that also backports the endianness fix for md5sum. Currently using a 0.50 build with the md5sum fix in it, but in that sed doesn't work, sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' $reldest only spits out the md5sum: line of the release file and none of the following lines. Again, havn't checked cvs yet. Could you send me what $reldest expands to so I can test this and fix it if needed? -Erik -- Erik B. Andersen email: [EMAIL PROTECTED], [EMAIL PROTECTED] --This message was written using 73% post-consumer electrons-- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Mon, May 14, 2001 at 01:35:20PM -0600, Erik Andersen wrote: On Mon May 14, 2001 at 05:00:34PM +0100, Richard Hirst wrote: Currently using a 0.50 build with the md5sum fix in it, but in that sed doesn't work, sed -n '/^[Mm][Dd]5[Ss][Uu][Mm]/,/^[^ ]/p' $reldest only spits out the md5sum: line of the release file and none of the following lines. Again, havn't checked cvs yet. Could you send me what $reldest expands to so I can test this and fix it if needed? I think that this was bug 91758, which was fixed in 0.51-1. And I think $reldest expands to one of the Release files, such as http://http.us.debian.org/debian/dists/woody/Release . Matt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
At 12:46 PM -0700 5/11/01, David Whedon wrote: Fri, May 11, 2001 at 01:56:34PM -0500 wrote: The next major stumbling block I ran into was that pump doesn't work (bts# 94176). Pump has annoyed me for a number of reasons including bts#64092 and its kin. I think there was some discussion some time ago about perhaps using dhcp-client. So I hacked dhcp-client to work. It works very well. The downside is that dhcp-client takes up a bit more space than pump: 137930 (more if kernel 2.0 support is required) vs. I don't think we need to directly support 2.0.x on boot-floppies, though if we decide not to we should make a note in the docs that dhcp won't work if you replace the kernel with a 2.0 serias. 58112. It's easy enough to #def this into/out of dbootstrap, but not quite so trivial to maintain both versions of the EXTRACT files. Let me know if you want the patch, even if just for dbootstrap. I've got an idea. Why don't we use dhcp-client-udeb? It is destined for use with debian-installer and is a bit smaller than the .deb: ruff:davidw$ dpkg -c dhcp-client-udeb_2.0pl5-5_i386.udeb drwxr-xr-x root/root 0 2001-05-11 12:38:14 ./ drwxr-xr-x root/root 0 2001-05-11 12:38:11 ./sbin/ -rwxr-xr-x root/root107356 2001-05-11 12:38:11 ./sbin/dhclient drwxr-xr-x root/root 0 2001-05-11 12:38:08 ./etc/ -rwxr-xr-x root/root 5816 2001-05-11 12:38:08 ./etc/dhclient-script ruff:davidw$ The udeb already exists in the archive, the only thing to do would be to modify the apt magic so it looks for it. Saves us a little space, and a lot of people complain about pump. Sounds like a good plan. /sbin/dhclient in the deb is a script that calls either dhclient-2.2.x or dhclient-2.0.x depending on uname -r. I just dropped dhclient-2.0.x and dhclient from the EXTRACT lists. I wonder if dhclient-2.2.x is renamed dhclient in the udeb? I suppose I'll just have to see. Should I fix dbootstrap to look for dhclient and failing that pump? That would give us more flexibility and it shouldn't be hard to do. -- Stephen R. MarenkaIf life's not fun, you're not doing it right! [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
Stephen R Marenka [EMAIL PROTECTED] writes: At 12:46 PM -0700 5/11/01, David Whedon wrote: Fri, May 11, 2001 at 01:56:34PM -0500 wrote: The next major stumbling block I ran into was that pump doesn't work (bts# 94176). Pump has annoyed me for a number of reasons including bts#64092 and its kin. I think there was some discussion some time ago about perhaps using dhcp-client. So I hacked dhcp-client to work. It works very well. The downside is that dhcp-client takes up a bit more space than pump: 137930 (more if kernel 2.0 support is required) vs. I don't think we need to directly support 2.0.x on boot-floppies, though if we decide not to we should make a note in the docs that dhcp won't work if you replace the kernel with a 2.0 serias. No, kernel 2.0 not needed, no one uses it. 58112. It's easy enough to #def this into/out of dbootstrap, but not quite so trivial to maintain both versions of the EXTRACT files. Let me know if you want the patch, even if just for dbootstrap. I've got an idea. Why don't we use dhcp-client-udeb? Um, I'm a bit hesitant to add more special cases and wierd handling (such as udeb) to the boot-floppies. Are there any udebs in Woody? Remember, this needs to be buildable by pure woody. Sounds like a good plan. /sbin/dhclient in the deb is a script that calls either dhclient-2.2.x or dhclient-2.0.x depending on uname -r. I just dropped dhclient-2.0.x and dhclient from the EXTRACT lists. I wonder if dhclient-2.2.x is renamed dhclient in the udeb? I suppose I'll just have to see. Again, I'm fine with moving to dhclient -- I would prefer to use the proper package rather than the udeb, space be damned. Should I fix dbootstrap to look for dhclient and failing that pump? That would give us more flexibility and it shouldn't be hard to do. Sure, please. -- .Adam Di [EMAIL PROTECTED]URL:http://www.onshored.com/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
David Whedon [EMAIL PROTECTED] writes: Fri, May 11, 2001 at 01:56:34PM -0500 wrote: That got me all the way to installing base, but that slammed to a halt quickly since it appears that md5sum is broken (another busybox problem? ppc-only? needs more research). So, I commented out the md5 check in debootstrap to see how much further I could go. i I remember this one. Are you using busybox cvs? At the moment busybox cvs is working, bb 0.51-1 (the current Debian package) doesn't work for a number of reasons. That got me to mount -t proc proc /proc which failed miserably. Interestingly, /proc was already (successfully?) mounted. I don't see a good way to get around that so there it ends. I guess I'll be checking out the busybox cvs next. oh, yes, you should be using cvs. Could someone *please* just NMU busybox, urgency high? I am wanting to build this boot-floppies *now* and I don't want to wait anymore... :) -- .Adam Di [EMAIL PROTECTED]URL:http://www.onshored.com/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
Sounds like a good plan. /sbin/dhclient in the deb is a script that calls either dhclient-2.2.x or dhclient-2.0.x depending on uname -r. I just dropped dhclient-2.0.x and dhclient from the EXTRACT lists. I wonder if dhclient-2.2.x is renamed dhclient in the udeb? I suppose I'll just have to see. Again, I'm fine with moving to dhclient -- I would prefer to use the proper package rather than the udeb, space be damned. Fine by me, we don't save that much space with the udeb anyway, pretty much the only difference is the compiler options are -Os. Should I fix dbootstrap to look for dhclient and failing that pump? That would give us more flexibility and it shouldn't be hard to do. Sure, please. Sounds good. David -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Sat May 12, 2001 at 01:05:55PM -0400, Adam Di Carlo wrote: David Whedon [EMAIL PROTECTED] writes: Fri, May 11, 2001 at 01:56:34PM -0500 wrote: That got me all the way to installing base, but that slammed to a halt quickly since it appears that md5sum is broken (another busybox problem? ppc-only? needs more research). So, I commented out the md5 check in debootstrap to see how much further I could go. i I remember this one. Are you using busybox cvs? At the moment busybox cvs is working, bb 0.51-1 (the current Debian package) doesn't work for a number of reasons. That got me to mount -t proc proc /proc which failed miserably. Interestingly, /proc was already (successfully?) mounted. I don't see a good way to get around that so there it ends. I guess I'll be checking out the busybox cvs next. oh, yes, you should be using cvs. Could someone *please* just NMU busybox, urgency high? I am wanting to build this boot-floppies *now* and I don't want to wait anymore... :) I'm building a new one now. Rather then waiting for 0.52 to stabalize, I'm backporting the critical bugfixes into 0.51. -Erik -- Erik B. Andersen email: [EMAIL PROTECTED], [EMAIL PROTECTED] --This message was written using 73% post-consumer electrons-- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
Erik Andersen [EMAIL PROTECTED] writes: I'm building a new one now. Rather then waiting for 0.52 to stabalize, I'm backporting the critical bugfixes into 0.51. Thank you . Please let us know when it's up. I can do a source release tomorrow and then we can have working boot-floppies source package at least in woody. -- .Adam Di [EMAIL PROTECTED]URL:http://www.onshored.com/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
On Sat May 12, 2001 at 12:10:26PM -0600, Erik Andersen wrote: It is now in incoming. Lemme know if you have any problems with it. I did a rather quick hack-n-slash job to put this together. It should be fine now, and I think I have all bugs bothering the boot-floppies fixed. Hopefully I didn't create any new problems... It turns out I screwed up cp and mv. Another upload that should actually work this time is now in incoming. Sorry about that. -Erik -- Erik B. Andersen email: [EMAIL PROTECTED], [EMAIL PROTECTED] --This message was written using 73% post-consumer electrons-- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: powerpc woody bf installation status
Fri, May 11, 2001 at 01:56:34PM -0500 wrote: On ppc I can now get all the way to extracting base. glad to head it :-) Sorry about breaking Sparc -- I thought I recalled seeing __KERNEL__ in at least asm-i386/unaligned.h, but it must have been my fuzzy eyes. no biggie. I have a couple simple patches to commit -- hopefully tonight -- one fixes a build error and the other adds the pmac drivers disk to the dbootstrap list of known driver disks. The next major stumbling block I ran into was that pump doesn't work (bts# 94176). Pump has annoyed me for a number of reasons including bts#64092 and its kin. I think there was some discussion some time ago about perhaps using dhcp-client. So I hacked dhcp-client to work. It works very well. The downside is that dhcp-client takes up a bit more space than pump: 137930 (more if kernel 2.0 support is required) vs. I don't think we need to directly support 2.0.x on boot-floppies, though if we decide not to we should make a note in the docs that dhcp won't work if you replace the kernel with a 2.0 serias. 58112. It's easy enough to #def this into/out of dbootstrap, but not quite so trivial to maintain both versions of the EXTRACT files. Let me know if you want the patch, even if just for dbootstrap. I've got an idea. Why don't we use dhcp-client-udeb? It is destined for use with debian-installer and is a bit smaller than the .deb: ruff:davidw$ dpkg -c dhcp-client-udeb_2.0pl5-5_i386.udeb drwxr-xr-x root/root 0 2001-05-11 12:38:14 ./ drwxr-xr-x root/root 0 2001-05-11 12:38:11 ./sbin/ -rwxr-xr-x root/root107356 2001-05-11 12:38:11 ./sbin/dhclient drwxr-xr-x root/root 0 2001-05-11 12:38:08 ./etc/ -rwxr-xr-x root/root 5816 2001-05-11 12:38:08 ./etc/dhclient-script ruff:davidw$ The udeb already exists in the archive, the only thing to do would be to modify the apt magic so it looks for it. Saves us a little space, and a lot of people complain about pump. That got me all the way to installing base, but that slammed to a halt quickly since it appears that md5sum is broken (another busybox problem? ppc-only? needs more research). So, I commented out the md5 check in debootstrap to see how much further I could go. i I remember this one. Are you using busybox cvs? At the moment busybox cvs is working, bb 0.51-1 (the current Debian package) doesn't work for a number of reasons. That got me to mount -t proc proc /proc which failed miserably. Interestingly, /proc was already (successfully?) mounted. I don't see a good way to get around that so there it ends. I guess I'll be checking out the busybox cvs next. oh, yes, you should be using cvs. -David -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]