Re: [lfs-support] Still fighting with Network Madness...
On Thu, 2012-01-12 at 01:50 +, Andrew Benton wrote: > That reminds me of http://xkcd.com/349/ > > Andy Yes, that comic should sound familiar to all LFS users... :) Simon. signature.asc Description: This is a digitally signed message part -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Error during Glibc Installation
On Jan 12, 2012 5:43 AM, "Uthayanan" wrote: > > > Hi Group, > > I got the following error during my Glibc installation process. Is this a known issue ? Please help me on this. > > > > > mawk: scripts/gen-sorted.awk: line 19: regular expression compile failed (bad class -- [], [^] or [) > /[^ > > > Thanks. > > -- > Regards, > s uthay > someone, sorry I can't remember who, pointed out that mawk wasn't very good with glibc. Uninstall mawk and install gawk on your host system and you should be fine (run configure again) -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
[lfs-support] Error during Glibc Installation
Hi Group, I got the following error during my Glibc installation process. Is this a known issue ? Please help me on this. mawk: scripts/gen-sorted.awk: line 19: regular expression compile failed (bad class -- [], [^] or [) /[^ mawk: scripts/gen-sorted.awk: line 19: syntax error at or near ] mawk: scripts/gen-sorted.awk: line 19: runaway regular expression /, "", subd ... make[1]: *** No rule to make target `/mnt/lfs/sourcenew/glibc-build/Versions.all', needed by `/mnt/lfs/sourcenew/glibc-build/abi-versions.h'. Stop. make[1]: Leaving directory `/mnt/lfs/sourcenew/glibc-2.14.1' make: *** [install] Error 2 lfsuthayan@ubuntu:/mnt/lfs/sourcenew/glibc-build$ Thanks. -- Regards, s uthay -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LFS 7.0 Chapter 6.17 Gcc-4.6.1 compilation - test results
On Wed, Jan 11, 2012 at 10:33:25PM -0400, Franck Chuiton wrote: > Hi everybody ! > I had first tried to build my LFS with a live CD but I eventually ended > up installing this Crunbang on the hard drive and starting over for a > new build. > So far, I've had no problem (seems to me, at least...). > I am not even sure I have one today : > The gcc tests ended up with this : > [ snipping except the unexpected failures ] > === libmudflap tests === > > > Running target unix > FAIL: libmudflap.c/fail40-frag.c output pattern test > -- > === libmudflap Summary === > > # of expected passes 1406 > # of unexpected failures 21 [...] > > So, please, is it a PASS ? Can I carry on ? > > Thanks in advance. > > Franck Chuiton > I no longer build on i686, only x86_64, but that seems *remarkably* good to me - no failures at all in gcc itself! On x86_64 I see failures in the gcc torture test, and rather fewer failures than you are getting in libmudflap. But, I suspect the libmudflap testsuite is best not discussed in polite company! Unless anyone else has very different i686 results, I think you are probably fine to carry on. As always, what really matters is not what the testsuite says, but whether the resulting system is good enough to build the packages you wish to run. If you saw dozens of failures all over the suite I would say there was a problem. If you saw up to a dozen (in different tests), there *might* be a problem. To get an apparently large number of failures from ONE libmudflap test doesn't convince me that anything is wrong. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
[lfs-support] LFS 7.0 Chapter 6.17 Gcc-4.6.1 compilation - test results
Hi everybody ! I had first tried to build my LFS with a live CD but I eventually ended up installing this Crunbang on the hard drive and starting over for a new build. So far, I've had no problem (seems to me, at least...). I am not even sure I have one today : The gcc tests ended up with this : === make[1]: Leaving directory `/sources/gcc-build' make: *** [do-check] Error 2 make: Target `check' not remade because of errors. === I summed it up and checked with this compile log (http://www.linuxfromscratch.org/lfs/build-logs/7.0/core2duo/test-logs/080-gcc) I don't have the same CPU nor the same figures... So I don't really know if it's OK... To me, it doesn't look so bad, what do you think ? Here are my results : === === g++ Summary === # of expected passes27266 # of expected failures 164 # of unsupported tests 139 /sources/gcc-build/gcc/testsuite/g++/../../g++ version 4.6.1 (GCC) === gcc tests === -- === gcc Summary === # of expected passes77323 # of expected failures 212 # of unsupported tests 679 /sources/gcc-build/gcc/xgcc version 4.6.1 (GCC) === libgomp tests === -- === libgomp Summary === # of expected passes1063 === libmudflap tests === Running target unix FAIL: libmudflap.c/fail40-frag.c output pattern test -- === libmudflap Summary === # of expected passes1406 # of unexpected failures21 === libstdc++ tests === Running target unix -- === libstdc++ Summary === # of expected passes8191 # of expected failures 84 # of unsupported tests 161 Compiler version: 4.6.1 (GCC) Platform: i686-pc-linux-gnu So, please, is it a PASS ? Can I carry on ? Thanks in advance. -- Franck Chuiton -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Still fighting with Network Madness...
jasonps...@jegas.com wrote: > The init scripts during boot say "trying to bring up interface eth0" > than I'm WARNED eth0 doesn't exist. In your LFS system, run 'dmesg > kernel.log'. Then in another system with network connectivity, go to http://pastebin.com/, sign up and paste the log there. Send us the link. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Still fighting with Network Madness...
On Wed, 11 Jan 2012 16:20:00 -0700 wrote: > I get the feeling that this is one of those areas that just works for > people, and I'm admittedly frustrated. > > Additionally, to "Evade this issue" when I do a full build in VirtualBox > using just default Kernel config, > I get a kernel panic about can't find init (And I checked the boot > loader and ftab to make sure the settings were right for the boot drive) > > I was so hoping to be deciding what Window Manager I wanted the most by > now :( That reminds me of http://xkcd.com/349/ Andy -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Still fighting with Network Madness...
On Wed, 11 Jan 2012 16:20:00 -0700 wrote: > It does make an entry in the /etc/udev/rules.d/70-persistent-net.rules > file for my card and uses eth0, > it even has a comment above it displaying a PCI device ID and in parens > (tg3) > > THIS IS PROBABLY GOOD I THINK Yes, that looks good. > I do not know where to go from here. Bruce explained there might be a > file in /sys/ folder, but I don't know what to do with it if I find id. > > The init scripts during boot say "trying to bring up interface eth0" > than I'm WARNED eth0 doesn't exist. That doesn't look good. Does eth0 exist? What do you get when you type: ip addr If eth0 doesn't exist it's a kernel problem. Maybe it needs recompiling, maybe it needs firmware. Andy -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Still fighting with Network Madness...
On Wed, Jan 11, 2012 at 04:20:00PM -0700, jasonps...@jegas.com wrote: > > In page 7.2.1 of the LFS 7.0 book, there is a script that you are > supposed to run. It doesn't say if you should run it on the host, after > you boot up or not but I've been running it once chroot'd and my > /mnt/lfs/dev has been populated via a bind. > > It does make an entry in the /etc/udev/rules.d/70-persistent-net.rules > file for my card and uses eth0, > it even has a comment above it displaying a PCI device ID and in parens > (tg3) > > THIS IS PROBABLY GOOD I THINK > If you have multiple ethernet devices, then fixing the names is useful. For a single network device, as the book says, it is optional. I believe that running it in chroot is the correct thing to do - otherwise we would have put a note there. For real hardware you need to look at the host's dmesg to see what it said, and compare it to what your new lfs kernel reported. (Virtual machines probably work very differently). You might also need to ensure that the driver and any fixes is indeed upstream in the kernel, rather than relying on distro patches. Similarly, if your host is running a newer kernel than the one you built, then it's possible you might need to build the newer kernel (e.g. if new PCI IDs were added to an existing driver - although those seem to get backported to stable kernels). ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
[lfs-support] Still fighting with Network Madness...
Hello Folks, I'm still trying to get my nic card to work. We've established the driver is in the kernel, (Broadcom 57788, known as tg3 by linux). I've run lspci while the host is running to establish its there. In page 7.2.1 of the LFS 7.0 book, there is a script that you are supposed to run. It doesn't say if you should run it on the host, after you boot up or not but I've been running it once chroot'd and my /mnt/lfs/dev has been populated via a bind. It does make an entry in the /etc/udev/rules.d/70-persistent-net.rules file for my card and uses eth0, it even has a comment above it displaying a PCI device ID and in parens (tg3) THIS IS PROBABLY GOOD I THINK I do not know where to go from here. Bruce explained there might be a file in /sys/ folder, but I don't know what to do with it if I find id. The init scripts during boot say "trying to bring up interface eth0" than I'm WARNED eth0 doesn't exist. I get the feeling that this is one of those areas that just works for people, and I'm admittedly frustrated. Additionally, to "Evade this issue" when I do a full build in VirtualBox using just default Kernel config, I get a kernel panic about can't find init (And I checked the boot loader and ftab to make sure the settings were right for the boot drive) I was so hoping to be deciding what Window Manager I wanted the most by now :( I've received help from many of you, but nothing yet has helped me solve this just yet, athough I know more about the subject now... I'm still feeling stumped. --Jason -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
Ken Moffat wrote: > Going to grub2 was somewhat painful. There is a bit of a learning curve, but if you use a separate boot partition and specify it in fstab for each /, it's really quite easy in LFS: 1. Place the kernel in /boot 2. Edit /boot/grub/grub.cfg and add menuentry "Descriptive text" { linux /kernel-filename root=/dev/sda ro } 3. Reboot There are a lot of gotchas when the user wants to do something more difficult like use a serial terminal, pxe boot, use a 3TB drive, etc. Basically I don't see a need for the boot loader to go into graphical mode when the user only sees it for maybe 5 seconds. There isn't really a requirement for a mouse when only a couple of arrow down/up keystrokes and a return are needed (if not just using the default). My biggest gripe is not GRUB itself, but the fact that some distros think they own grub.cfg and will wipe out what you have there. Remember that the boot loader really only does two things: load a file from a location (disk, network, etc) into memory and start executing that file it just loaded. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
On Wed, Jan 11, 2012 at 05:41:11PM +, Firerat wrote: > On Wed, Jan 11, 2012 at 11:20 AM, wrote: [ snipped - people, PLEASE trim what you are replying to ] > > lilo is plain and simple to use > but at least for me there is no contest , grub wins hands down As a long-time user of lilo, I have to agree that grub2 is better. All bootloaders are nasty, and I used to find lilo's simplicity suited me. On my (x86_64) desktops I moved once grub2 came into the book, and after a learning curve it now runs reliably for me. My server used to have an old install and I was stuck on a very old kernel because of lilo (the root disk moved to sd? from hda when the kernel went to the new ATA code [ might have been sda, or might have been sdc, I forget ] and lilo refused to write the fixed-up entry because in the old kernel the sd drive didn't exist. Going to grub2 was somewhat painful. Needed a rescue CD a few times for when it failed to boot. On the desktops I had to make a symlink to sort out my separate /boot partition. On the server I also had to search long and hard to get a 1024x768 framebuffer after I moved to a new motherboard - changed to vesafb from radeonfb and had to add things to grub: setgfxmode=1024x768 setgfxpaylod=keep # hmm, that looks like a typo insmod gfxterm insmod vbe Note that both my desktops are happily using radeonfb and kms without having to fiddle with grub.cfg, but these sort of "works there, but not here" issues are common in bootloaders. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Problem on compiling and installing Binutils-2.21.1a
On Wed, Jan 11, 2012 at 09:07:47AM -0800, Qrux wrote: > > On Jan 10, 2012, at 3:58 PM, Andrew Benton wrote: > > > Fix this. Glibc does not like mawk. Use gawk. > > Also, can you please stop top posting and trim your replies to the > > relevant content? > > Building binutils (pass 3?): > > http://www.linuxfromscratch.org/lfs/view/stable/chapter06/binutils.html > > It's hard to tell if my observations are related to the previous binutils > posts from the excessively long quoting...But, I've run into issues with both > the compilation and the testing. Specifically, > > make tooldir=/usr > > fails (looks like subdirs without Makefiles, or something like that): > > config.status: creating po/Makefile.in > config.status: executing depfiles commands > config.status: executing libtool commands > config.status: executing default-1 commands > config.status: executing bfd_stdint.h commands > config.status: executing default commands > make[3]: Leaving directory `/sources/binutils-build/bfd/po' > make[3]: Entering directory `/sources/binutils-build/bfd/po' > make[3]: Nothing to be done for `info'. > make[3]: Leaving directory `/sources/binutils-build/bfd/po' > make[3]: Entering directory `/sources/binutils-build/bfd' > make[3]: Nothing to be done for `info-am'. > make[3]: Leaving directory `/sources/binutils-build/bfd' > make[2]: *** [info-recursive] Error 1 > make[2]: Leaving directory `/sources/binutils-build/bfd' > make[1]: *** [all-bfd] Error 2 > make[1]: Leaving directory `/sources/binutils-build' > make: *** [all] Error 2 > > Even if this is acceptable, it's rather disconcerting. Could a patch be made > to avoid descending into bfd, etc? It isn't acceptable. I've never seen anything like this. The report of 'Error 1' without an actual error message makes me wonder if you caught stdout in a file but didn't catch stderr ? Apart from that, please make sure that you delete any build directory as well as the extracted source after you have installed each package. So, in this case, please delete the binutils-build and binutils-2.21.1a directories before you try again. Did you fail to install texinfo in chapter 5 ? A quick test for that is to try running 'info' or 'info libc'. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Q: Diskless boot and initramfs final mount using LFS
On Wed, Jan 11, 2012 at 7:32 PM, Firerat wrote: > > On Jan 11, 2012 5:22 PM, "Li, David" wrote: >> >> Hi, >> >> >> >> I am totally new to LFS but very interested to see how I can use it for my >> requirements. >> >> >> >> I ‘ll need to custom build a Linux distribution that eventually has to >> meet requirements: >> >> >> >> 1. Smaller than 200MB (kernel + intramfs image files). >> > > Embeded lfs (see CLFS ) may be more suited to your needs > Been a while since I read that ,I think it details things like busybox to > keep the size down uClibc instead of glibc is also helping. It's definitely possible. There's a project called Tor, they have a 5.3M ramdisk with linux and their Tor server in it, so it should be possible to build a system with 200 MB. Matijn -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Q: Diskless boot and initramfs final mount using LFS
On Jan 11, 2012 5:22 PM, "Li, David" wrote: > > Hi, > > > > I am totally new to LFS but very interested to see how I can use it for my requirements. > > > > I ‘ll need to custom build a Linux distribution that eventually has to meet requirements: > > > > 1. Smaller than 200MB (kernel + intramfs image files). > Embeded lfs (see CLFS ) may be more suited to your needs Been a while since I read that ,I think it details things like busybox to keep the size down > 2. Be able to support KVM-based virtualization (Redhat). > part of the kernel > 3. Can be pxebooted on a x86_64 architecture > doubt LFS will be a constraint your Nic will need to support it You will need a server to serve the boot images > 4. Can mount the initramfs as its final root file system (no disk, no NFS). > > Kernel has initramfs, and as far as I remember scripts to create them > > Is this something doable using LFS? > In conclusion Yes and No As it stands LFS will probably be too large for your needs , as mentioned CLFS may help with that The PXE is irrelevant, unless you are building an LFS to act as the server Initramfs, read the kernel docs -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Q: Diskless boot and initramfs final mount using LFS
>Hi, > >I am totally new to LFS but very interested to see how I can use it for my >requirements. > >I ‘ll need to custom build a Linux distribution that eventually has to meet >requirements: > >1. Smaller than 200MB (kernel + intramfs image files). >2. Be able to support KVM-based virtualization (Redhat). >3. Can be pxebooted on a x86_64 architecture >4. Can mount the initramfs as its final root file system (no disk, no >NFS). > >Is this something doable using LFS? > >Thanks. > >David > Hello David and Welcome. I'm SO NEW here myself I don't really feel qualified to answer with authority, so keep that in mind. Also, I'm hoping where I'm wrong, I'll promptly be corrected someone more knowledgeable. 1: I DO think you can make a smaller than 200meg Linux system. I don't know how big the KVM stuff would be though Also I think some "trimming" on your part would be needed possibly stripping out any packages you don't require and other packages aren't dependant on. 2: This part I'm not sure. FRankly you'll need a Working LFS system, and the the KVM sources to try to compile KVM tools and see...I think. 3: I don't know what that means: pxebootedI guess... Maybe? Did you missspell? Prebooted? I'm Leaning toward yes in that case. 4: I'm no expert hert but it's my understanding that working with the RAM File System requires some kernel know how. From what I can tell so far, you have to get your "file system" zipped up and small so that the kernel can use it. I THINK this zipped file needs to be sort of compiled into the kernel - so if I'm right - you probably can but this is well out side the realms of the LFS "Scope", that doesn't meen its not possible. It's just many of us are happy to make a linux files system that boots, you want to then that and squeeze it into yet another kernel and boot it there. I think you have some serious work/time ahead of you on this one, but I KNOW generally speaking this is possible. Some "Rescue Linux" and even "Puppy" run their kernel and a small filesystem in RAM... Puppy does it for speed gains. Good Luck and Hope others chime in to correct me where I'm likely wrong so David here gets a good idea. --Jason -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
On Wed, Jan 11, 2012 at 11:20 AM, wrote: > First of all, let me say that from everything I've read, Grub2 in > particular is the best boot loader out there, but with its > sophistication comes larger size and slightly more complicated > configuration (for me anyways.) And I've read that there have been > LILO/BRUB "flame Wars" in here - so PLEASE, this post isn't about that. > It's more about I don't know grub well enough and the book helps > sometimes.. I'll explain. > > When I am using Slackware as my Host, it has Lilo. When I finish making > my LFS 7.0 OS on my /mnt/lfs partition, all I have to do to guarantee a > boot is: > > 1: Copy the /mnt/lfs/boot/vmlinuz-3.1-lfs-7.0 kernel file to the host > /boot/ > > 2: edit /etc/lilo.conf and copy the last entry in the file and paste it, > change the label, the name of the kernel file and the drive where where > my LFS is located. > > EXAMPLE: At end of lilo config file I see something like (comments added > for clarity) > > #--BEGIN > image=/boot/vmlinuz > root=/dev/sda2 > label=Linux > read-only > #--END > > I copy-n-paste this and change the values for my LFS: > > #--BEGIN > image=/boot/vmlinuz-3.1-lfs-7.0 > root=/dev/sda3 > label=LFS > read-only > #--END > > Then I REBOOT - and I have the Slackware/Lilo Menu Screen, I select LFS > and it boots everytime. > Now, if your host has LILO, you might find this is pretty slick. You > MUST remember that if you make a new kernel you need to copy it to your > host's /boot/ folder before rebooting or things might not act like you > expect. > > ---Moving On > > Barry mentioned that LILO has some issues that can be problematic, and I > don't doubt it for a second. I do however have trouble with GRUB. The > Book tells us some steps.. but I get differing results everytime I try > GRUB and in all cases I'm not able to run the HOST again. > > 1: I am under the impression we are to install grub on the host machine > because it isn't clear in the book, further trying to install grub while > chrooted into the LFS system hasn't worked for me yet. Compiling errors > if memory serves me correctly. > > 2: All attempts to configure the grub menu to launch the HOST fail for > me. (note I always get one shot at this, A rescue disk would help here > surely) > > Usually when grub seems like its going to work, I get a menu.. and I > think only a couple times have I personally had luck making it actually > boot. I often have to edit the menu option here and tweak it to work, it > seems so mangle the linux line on me and I have to put it back > together - then I can boot. > > Usually I get a grub rescue prompt, and I'm sorry but I've googled the > heck out of that, read all sorts of documentation... that thing is as > useless as [insert some funny remark here]. > > The bummer for me is I like the way Grub allows you set timing, splash > screens, fire off windows partitions and much more. Lilo does to but has > less features I suppose. > > Either Way, I have no idea yet how I could make a installable ISO just > yet that would boot, start a setup program to gather information, and > then install grub or lilo for that matter on a target machine without > messing it up. This is something I surely what to be able to do. > > So this message is not so much a whine, or an ask for help as it is my > current place in this stormy ocean at this point. I've made tons of > progress, I'm hoping my Lilo Trick for folks with Lilo based hosts is > somewhat useful... but I, Like Barry and the Book recommend, would love > to use Grub without feeling sorta fearful I'm going trash the machine > everytime I try. > > I know it's likely my lack of knowledge in this Grubby Area, however I > still felt like sharing the bits that have worked flawless for me as > well. > > --Jason lilo is plain and simple to use but at least for me there is no contest , grub wins hands down regards iso booting, this is easy in both lilo and grub ( tbh even easier in grub using grml scripts ) a quick google of "lilo iso boot" gets you this http://www.linuxquestions.org/questions/linux-general-1/boot-iso-image-from-hard-disk-294744/ -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Q: Diskless boot and initramfs final mount using LFS
Li, David wrote: > Hi, > > I am totally new to LFS but very interested to see how I can use it for my > requirements. > > I 'll need to custom build a Linux distribution that eventually has to meet > requirements: > > > 1. Smaller than 200MB (kernel + intramfs image files). You'll have to delete some things, but that's possible. > 2. Be able to support KVM-based virtualization (Redhat). Isn't that a function of the kernel? I don't know if you need extra programs for that or not. > 3. Can be pxebooted on a x86_64 architecture I don't see why not. AFAIK, pxe just transfers files (bitstreams). > 4. Can mount the initramfs as its final root file system (no disk, no NFS). You would have to rewrite the boot scripts. Note that LFS doesn't directly support any initramfs, so you are on your own there. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
Eleanore Boyd wrote: > May I add that I use the GRUB bootloader found in Puppy 5.28? It finds > everything installed on the system, and you can easily edit the menu.lst > to change the boot order and options. Just a note. If you are using menu.lst, you are using GRUB Legacy, not GRUB 2. GRUB Legacy may do exactly what you want, but it is nowhere as flexible as GRUB 2. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
[lfs-support] Q: Diskless boot and initramfs final mount using LFS
Hi, I am totally new to LFS but very interested to see how I can use it for my requirements. I 'll need to custom build a Linux distribution that eventually has to meet requirements: 1. Smaller than 200MB (kernel + intramfs image files). 2. Be able to support KVM-based virtualization (Redhat). 3. Can be pxebooted on a x86_64 architecture 4. Can mount the initramfs as its final root file system (no disk, no NFS). Is this something doable using LFS? Thanks. David -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Problem on compiling and installing Binutils-2.21.1a
On Jan 10, 2012, at 3:58 PM, Andrew Benton wrote: > Fix this. Glibc does not like mawk. Use gawk. > Also, can you please stop top posting and trim your replies to the > relevant content? Building binutils (pass 3?): http://www.linuxfromscratch.org/lfs/view/stable/chapter06/binutils.html It's hard to tell if my observations are related to the previous binutils posts from the excessively long quoting...But, I've run into issues with both the compilation and the testing. Specifically, make tooldir=/usr fails (looks like subdirs without Makefiles, or something like that): config.status: creating po/Makefile.in config.status: executing depfiles commands config.status: executing libtool commands config.status: executing default-1 commands config.status: executing bfd_stdint.h commands config.status: executing default commands make[3]: Leaving directory `/sources/binutils-build/bfd/po' make[3]: Entering directory `/sources/binutils-build/bfd/po' make[3]: Nothing to be done for `info'. make[3]: Leaving directory `/sources/binutils-build/bfd/po' make[3]: Entering directory `/sources/binutils-build/bfd' make[3]: Nothing to be done for `info-am'. make[3]: Leaving directory `/sources/binutils-build/bfd' make[2]: *** [info-recursive] Error 1 make[2]: Leaving directory `/sources/binutils-build/bfd' make[1]: *** [all-bfd] Error 2 make[1]: Leaving directory `/sources/binutils-build' make: *** [all] Error 2 Even if this is acceptable, it's rather disconcerting. Could a patch be made to avoid descending into bfd, etc? In addition, testing: make -k check also seems to fail, perhaps also due to empty subdirs (or something similar): make[4]: Leaving directory `/sources/binutils-build/bfd' make[3]: Leaving directory `/sources/binutils-build/bfd' make[2]: Leaving directory `/sources/binutils-build/bfd' /bin/sh: line 3: cd: ./opcodes: No such file or directory make[1]: *** [check-opcodes] Error 1 /bin/sh: line 3: cd: ./binutils: No such file or directory make[1]: *** [check-binutils] Error 1 /bin/sh: line 3: cd: ./etc: No such file or directory make[1]: *** [check-etc] Error 1 /bin/sh: line 3: cd: ./gas: No such file or directory make[1]: *** [check-gas] Error 1 /bin/sh: line 3: cd: ./gprof: No such file or directory make[1]: *** [check-gprof] Error 1 make[2]: Entering directory `/sources/binutils-build/intl' make[2]: Nothing to be done for `check'. make[2]: Leaving directory `/sources/binutils-build/intl' /bin/sh: line 3: cd: ./ld: No such file or directory make[1]: *** [check-ld] Error 1 make[2]: Entering directory `/sources/binutils-build/libiberty' make[3]: Entering directory `/sources/binutils-build/libiberty/testsuite' gcc -DHAVE_CONFIG_H -g -O2 -I.. -I../../../binutils-2.21.1/libiberty/testsuite/../../include -o test-demangle \ ../../../binutils-2.21.1/libiberty/testsuite/test-demangle.c ../libiberty.a ./test-demangle < ../../../binutils-2.21.1/libiberty/testsuite/demangle-expected ./test-demangle: 827 tests, 0 failures gcc -DHAVE_CONFIG_H -g -O2 -I.. -I../../../binutils-2.21.1/libiberty/testsuite/../../include -DHAVE_CONFIG_H -I.. -o test-pexecute \ ../../../binutils-2.21.1/libiberty/testsuite/test-pexecute.c ../libiberty.a ./test-pexecute gcc -DHAVE_CONFIG_H -g -O2 -I.. -I../../../binutils-2.21.1/libiberty/testsuite/../../include -DHAVE_CONFIG_H -I.. -o test-expandargv \ ../../../binutils-2.21.1/libiberty/testsuite/test-expandargv.c ../libiberty.a ./test-expandargv PASS: test-expandargv-0. PASS: test-expandargv-1. PASS: test-expandargv-2. PASS: test-expandargv-3. PASS: test-expandargv-4. PASS: test-expandargv-5. PASS: test-expandargv-6. make[3]: Leaving directory `/sources/binutils-build/libiberty/testsuite' make[2]: Leaving directory `/sources/binutils-build/libiberty' make[1]: Target `check-host' not remade because of errors. make[1]: Nothing to be done for `check-target'. make[1]: Leaving directory `/sources/binutils-build' make: *** [do-check] Error 2 make: Target `check' not remade because of errors. There are indications it's working. Something about 'test-demangle' passing 827 tests and failing 0 and later some PASS messages. But, when the book says: "The test suite for Binutils in this section is considered critical. Do not skip it under any circumstances." and the testing fails it's unnerving. == I'm working under the assumption that the tests are succeeding, and that descending the subdirs could be eliminated. It's possible that I'm somehow compiling things incorrectly (though it's not my first trip around the LFS block). It's also possible that the tests are actually failing, and somehow it's just showing up now--which seems odd, since I'm no longer using the "cross-compiled" toolchain, but rather the tools in /bin after the chroot-jail. Anything jump out at anyone? Q -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfr
Re: [lfs-support] LFS 7.0 - No Devices in /dev on boot up - Kernel then shuts off PC
On Wednesday 11 January 2012 05:04:31 am Simon Geard wrote: > On Tue, 2012-01-10 at 02:20 -0700, jasonps...@jegas.com wrote: > > Well, I opend the "previous untouchable" RC file and tossed readline all > > over it ... and when I booted .. GUESS WHAT? I was able to step through > > the actual boot process one command at a time... and I saw the offending > > error for the first time... and bot this is embarrassing > > > > /etc/rc.f/rcS.d/S10udev Line 78 /sbin/udevd - No Such File > > > > I looked in my scripts that build LFS, and realized I forgot to UNTAR > > UDEV! > > Next lesson - when doing scripted builds, error handling is important. > You want the entire build to stop immediately when something goes wrong, > instead of trying to continue, making the original error hard to find. > > Personally, my preferred means of dealing with that is to simply add > "set -e" near the top of my scripts. The has the effect that if any > command run by the script exits with a non-zero status (i.e fails), the > script aborts immediately. > > Simon. I use the following: #!/bin/bash -e shopt -s -o pipefail the shopt makes sure that the error makes it through a pipe. so if you redirect to a log (which I do) it will stop on the error. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
>> I know it's likely my lack of knowledge in this Grubby Area, however I >> still felt like sharing the bits that have worked flawless for me as >> well. >> >> --Jason > >May I add that I use the GRUB bootloader found in Puppy 5.28? It finds >everything installed on the system, and you can easily edit the menu.lst >to change the boot order and options. > >Elly > Of course you can! (BTW I love Puppy 5.28 and Puppy Mariner too) I tried Woof2 for my "distro" adventure but found it was quite buggy. I think Woof (version 1) was maybe more ironed out. Woof2 seemed in a state of flux. On that tangent, I tried T1 a little and looked at their sample *.ISO and Found Linux From Scratch the way to go if your serious about building your own system or learning how to build your own distro etc. You really learn the parts along the way .. I think so anyway - I didn't learn every granule but I sure learned a lot so far along the way. Now I just learned from you can mix-n-match compiled Grub's if you like, cool! I have Puppy 5.28 handy too - I'll have to experiment with this. Thank You Elly -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LFS 7.0 - No Devices in /dev on boot up - Kernel then shuts off PC
>In the kernel (3.2), >make menuconfig >Device Drivers ---> >[*] Network device support ---> >[*] Ethernet driver support ---> >[*] Broadcom devices >[*] Broadcom Tigon3 support > >Andy > I missed that! I Selected it and other in that category for good measure And the other Ethernet Series below it - my cards are getting fast! :) I think now with this and the suggestions from Barry and Matijn I should be in good shape! Thank You! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
On 1/11/2012 5:20 AM, jasonps...@jegas.com wrote: > First of all, let me say that from everything I've read, Grub2 in > particular is the best boot loader out there, but with its > sophistication comes larger size and slightly more complicated > configuration (for me anyways.) And I've read that there have been > LILO/BRUB "flame Wars" in here - so PLEASE, this post isn't about that. > It's more about I don't know grub well enough and the book helps > sometimes.. I'll explain. > > When I am using Slackware as my Host, it has Lilo. When I finish making > my LFS 7.0 OS on my /mnt/lfs partition, all I have to do to guarantee a > boot is: > > 1: Copy the /mnt/lfs/boot/vmlinuz-3.1-lfs-7.0 kernel file to the host > /boot/ > > 2: edit /etc/lilo.conf and copy the last entry in the file and paste it, > change the label, the name of the kernel file and the drive where where > my LFS is located. > > EXAMPLE: At end of lilo config file I see something like (comments added > for clarity) > > #--BEGIN > image=/boot/vmlinuz > root=/dev/sda2 > label=Linux > read-only > #--END > > I copy-n-paste this and change the values for my LFS: > > #--BEGIN > image=/boot/vmlinuz-3.1-lfs-7.0 > root=/dev/sda3 > label=LFS > read-only > #--END > > Then I REBOOT - and I have the Slackware/Lilo Menu Screen, I select LFS > and it boots everytime. > Now, if your host has LILO, you might find this is pretty slick. You > MUST remember that if you make a new kernel you need to copy it to your > host's /boot/ folder before rebooting or things might not act like you > expect. > > ---Moving On > > Barry mentioned that LILO has some issues that can be problematic, and I > don't doubt it for a second. I do however have trouble with GRUB. The > Book tells us some steps.. but I get differing results everytime I try > GRUB and in all cases I'm not able to run the HOST again. > > 1: I am under the impression we are to install grub on the host machine > because it isn't clear in the book, further trying to install grub while > chrooted into the LFS system hasn't worked for me yet. Compiling errors > if memory serves me correctly. > > 2: All attempts to configure the grub menu to launch the HOST fail for > me. (note I always get one shot at this, A rescue disk would help here > surely) > > Usually when grub seems like its going to work, I get a menu.. and I > think only a couple times have I personally had luck making it actually > boot. I often have to edit the menu option here and tweak it to work, it > seems so mangle the linux line on me and I have to put it back > together - then I can boot. > > Usually I get a grub rescue prompt, and I'm sorry but I've googled the > heck out of that, read all sorts of documentation... that thing is as > useless as [insert some funny remark here]. > > The bummer for me is I like the way Grub allows you set timing, splash > screens, fire off windows partitions and much more. Lilo does to but has > less features I suppose. > > Either Way, I have no idea yet how I could make a installable ISO just > yet that would boot, start a setup program to gather information, and > then install grub or lilo for that matter on a target machine without > messing it up. This is something I surely what to be able to do. > > So this message is not so much a whine, or an ask for help as it is my > current place in this stormy ocean at this point. I've made tons of > progress, I'm hoping my Lilo Trick for folks with Lilo based hosts is > somewhat useful... but I, Like Barry and the Book recommend, would love > to use Grub without feeling sorta fearful I'm going trash the machine > everytime I try. > > I know it's likely my lack of knowledge in this Grubby Area, however I > still felt like sharing the bits that have worked flawless for me as > well. > > --Jason > > > May I add that I use the GRUB bootloader found in Puppy 5.28? It finds everything installed on the system, and you can easily edit the menu.lst to change the boot order and options. Elly -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
[lfs-support] LILO and GRUB - LILO I can make work, GRUB often leaves me Grubby
First of all, let me say that from everything I've read, Grub2 in particular is the best boot loader out there, but with its sophistication comes larger size and slightly more complicated configuration (for me anyways.) And I've read that there have been LILO/BRUB "flame Wars" in here - so PLEASE, this post isn't about that. It's more about I don't know grub well enough and the book helps sometimes.. I'll explain. When I am using Slackware as my Host, it has Lilo. When I finish making my LFS 7.0 OS on my /mnt/lfs partition, all I have to do to guarantee a boot is: 1: Copy the /mnt/lfs/boot/vmlinuz-3.1-lfs-7.0 kernel file to the host /boot/ 2: edit /etc/lilo.conf and copy the last entry in the file and paste it, change the label, the name of the kernel file and the drive where where my LFS is located. EXAMPLE: At end of lilo config file I see something like (comments added for clarity) #--BEGIN image=/boot/vmlinuz root=/dev/sda2 label=Linux read-only #--END I copy-n-paste this and change the values for my LFS: #--BEGIN image=/boot/vmlinuz-3.1-lfs-7.0 root=/dev/sda3 label=LFS read-only #--END Then I REBOOT - and I have the Slackware/Lilo Menu Screen, I select LFS and it boots everytime. Now, if your host has LILO, you might find this is pretty slick. You MUST remember that if you make a new kernel you need to copy it to your host's /boot/ folder before rebooting or things might not act like you expect. ---Moving On Barry mentioned that LILO has some issues that can be problematic, and I don't doubt it for a second. I do however have trouble with GRUB. The Book tells us some steps.. but I get differing results everytime I try GRUB and in all cases I'm not able to run the HOST again. 1: I am under the impression we are to install grub on the host machine because it isn't clear in the book, further trying to install grub while chrooted into the LFS system hasn't worked for me yet. Compiling errors if memory serves me correctly. 2: All attempts to configure the grub menu to launch the HOST fail for me. (note I always get one shot at this, A rescue disk would help here surely) Usually when grub seems like its going to work, I get a menu.. and I think only a couple times have I personally had luck making it actually boot. I often have to edit the menu option here and tweak it to work, it seems so mangle the linux line on me and I have to put it back together - then I can boot. Usually I get a grub rescue prompt, and I'm sorry but I've googled the heck out of that, read all sorts of documentation... that thing is as useless as [insert some funny remark here]. The bummer for me is I like the way Grub allows you set timing, splash screens, fire off windows partitions and much more. Lilo does to but has less features I suppose. Either Way, I have no idea yet how I could make a installable ISO just yet that would boot, start a setup program to gather information, and then install grub or lilo for that matter on a target machine without messing it up. This is something I surely what to be able to do. So this message is not so much a whine, or an ask for help as it is my current place in this stormy ocean at this point. I've made tons of progress, I'm hoping my Lilo Trick for folks with Lilo based hosts is somewhat useful... but I, Like Barry and the Book recommend, would love to use Grub without feeling sorta fearful I'm going trash the machine everytime I try. I know it's likely my lack of knowledge in this Grubby Area, however I still felt like sharing the bits that have worked flawless for me as well. --Jason -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Final GCC Compile Weirdness - 1 Time? No, 2 nd Time? Perfect!
consuming.. We're talking hours on a new i7 DELL XPS Studio (Quad Core with hyper-threading - its like having EIGHT CPU) >>> >>> Is 'hours' an exaggeration, or is that really what you're seeing? On a >>> modern machine like that, the gcc build should be something like five >>> minutes... >> >> Well 5 minutes is pretty quick. I have: >> >> Chapter 5 [gcc-pass1] 8 minutes and 22 seconds >> Chapter 5 [gcc-pass2] 11 minutes and 37 seconds >> Chapter 6 [gcc ] 78 minutes and 35 seconds >> >> Using only 1 core for consistent timing. >> > >Yes, but that's using a single core, which I'd never be doing on a >machine like the one described - running -j8 sounds perfectly >reasonable. > >However, it occurs to me that when I said 5 minutes, that would have >excluded tests, since my automated builds skip them. Can't remember how >effectively those can be parallelized, but I'd still not expect even the >Chapter 6 gcc build to take more than 20 minutes... > >Simon. I was trying to give you folks new numbers now that I seem to have my scripts almost in stone ... but I fell asleep for 12 hours (happens after all these 24+ hour marathons I do... programming, or doing this LFS stuff.. I tend to do that). For me, I've taken Barry's advice (He helped write the book so it might be his advice) Where is says these tests don't matter... I don't pay much attention. When the book says "Sanity Check - Run these tests - its critical? I just do it. Maybe I'll get daring after awhile and make a way to set a variable (export RUNLFSTEST=true) or something, and make the script make a decision on how it launches the makes and "configure" scripts. For now, I'm happy I can get the OS to compile and boot. I still have things I'm working on, but I'm pretty much overall pleased with the LFS book and the community. As for my gcc taking awhile, it is the tests and I didn't mean to exaggerate if I did - I would say its at least compile + testing .. around 45 minutes (double if I run twice because of errors) but that DOES include the testing which is what seems to take the longest. --Jason -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LFS 7.0 - No Devices in /dev on boot up - Kernel then shuts off PC
>> I looked in my scripts that build LFS, and realized I forgot to UNTAR >> UDEV! > >Next lesson - when doing scripted builds, error handling is important. >You want the entire build to stop immediately when something goes wrong, >instead of trying to continue, making the original error hard to find. > >Personally, my preferred means of dealing with that is to simply add >"set -e" near the top of my scripts. The has the effect that if any >command run by the script exits with a non-zero status (i.e fails), the >script aborts immediately. > >Simon. . I LIKE IT! KEEP EM COMING! set -e <--- Break on NON-ZERO EXIT STATUS = COOL (I look for exit results when spawning from within my programs but I didn't know this scripting trick. Thank you) -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] Final GCC Compile Weirdness - 1 Time? No, 2 nd Time? Perfect!
On Tue, 2012-01-10 at 10:11 -0600, Bruce Dubbs wrote: > Simon Geard wrote: > > On Mon, 2012-01-09 at 08:57 -0700, jasonps...@jegas.com wrote: > >> Now, I wouldn't care that much except even running on a DEDICATED 32bit > >> Linux (Slackware 13.37) with MAKEFLAGS set to EIGHT to use all CORES... > >> it takes a while... So running the procedure twice is time > >> consuming.. We're talking hours on a new i7 DELL XPS Studio (Quad Core > >> with hyper-threading - its like having EIGHT CPU) > > > > Is 'hours' an exaggeration, or is that really what you're seeing? On a > > modern machine like that, the gcc build should be something like five > > minutes... > > Well 5 minutes is pretty quick. I have: > > Chapter 5 [gcc-pass1] 8 minutes and 22 seconds > Chapter 5 [gcc-pass2] 11 minutes and 37 seconds > Chapter 6 [gcc ] 78 minutes and 35 seconds > > Using only 1 core for consistent timing. > Yes, but that's using a single core, which I'd never be doing on a machine like the one described - running -j8 sounds perfectly reasonable. However, it occurs to me that when I said 5 minutes, that would have excluded tests, since my automated builds skip them. Can't remember how effectively those can be parallelized, but I'd still not expect even the Chapter 6 gcc build to take more than 20 minutes... Simon. signature.asc Description: This is a digitally signed message part -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] my latest project
>I will try out Jason’s scripts > >Alain Hey Alain. I have just put the latest of my scripts here ftp://jegas.net/os/lfs/misc/lfs7.jegas.0.2.tar.gz I unpack them in /mnt/lfs You'll end up with /mnt/lfs/lfs7 If done in right place. Then in /mnt/lfs/lfs7/Jegas/ You'll find the scripts in numbered order, each has instructions at the end for what to do next. The GCC script is clearly marked now as 11-chrootgcc.sh And one more thing: I RAN The Script in Win7->VirtualBox->Slackware 13.37 (32 bit) with four processors and the MAKEFLAGS set to 4 processors and it worked the first time, a second run wasn't needed. I believe now it IS a RACE condition that can happen... and running the script again in that situation, though a pain, solves the problem itself. (doesn't match my normal computer understanding, but I'll accept that FTW :) --Jason -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LFS 7.0 - No Devices in /dev on boot up - Kernel then shuts off PC
On Tue, 2012-01-10 at 02:20 -0700, jasonps...@jegas.com wrote: > Well, I opend the "previous untouchable" RC file and tossed readline all > over it ... and when I booted .. GUESS WHAT? I was able to step through > the actual boot process one command at a time... and I saw the offending > error for the first time... and bot this is embarrassing > > /etc/rc.f/rcS.d/S10udev Line 78 /sbin/udevd - No Such File > > I looked in my scripts that build LFS, and realized I forgot to UNTAR > UDEV! Next lesson - when doing scripted builds, error handling is important. You want the entire build to stop immediately when something goes wrong, instead of trying to continue, making the original error hard to find. Personally, my preferred means of dealing with that is to simply add "set -e" near the top of my scripts. The has the effect that if any command run by the script exits with a non-zero status (i.e fails), the script aborts immediately. Simon. signature.asc Description: This is a digitally signed message part -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: [lfs-support] LFS - 5.29 - perl-5.14.2 - possible errata
On Jan 10, 2012, at 12:56 AM, wrote: >>> On Mon, 2012-01-09 at 04:19 -0700, jasonps...@jegas.com wrote: >>> But my REAL puzzle is the PERL thing I'm addressing here happens pretty >>> early in the big picture, and AFTER a very global chown -R lfs /tools >>> and chown -R lfs /sources I'm building LFS-7.0, and I'm at the first Perl build: http://www.linuxfromscratch.org/lfs/view/stable/chapter05/perl.html I think there's an error here. I've been reading through the recent Perl-related scripts, and I'm not sure it's covered. Basically, the patch fails for me because /hints/linux.sh is read-only. Adding a line before the patch: chmod u+w hints/linux.sh patch -Np1 -i ../perl-5.14.2-libc-1.patch fixes the issue for me. (Everything else is fine...The symlink & related issues covered in the thread is just more BGFB stuff.) * * * On the off-chance that this is Jason's error...A very early chown/chmod may not solve anything, since the perms issue appears only *AFTER* you unpack the Perl tarball. Q -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page