Re: Linux-2.6.30.9 build failure fixed new problem show up
Ken Moffat wrote: 2009/12/7 Mykal Funk mykalf...@gmail.com: Ken Moffat wrote: Do you have CONFIG_COMPAT_VDSO=y ? If so, try turning it off. whoops, if not try turning it on. I'm not sure where it appears in menuconfig, but the help says: config COMPAT_VDSO def_bool y prompt Compat VDSO support depends on X86_32 || IA32_EMULATION ---help--- Map the 32-bit VDSO to the predictable old-style address too. ---help--- Say N here if you are running a sufficiently recent glibc version (2.3.3 or later), to remove the high-mapped VDSO mapping and to exclusively use the randomized VDSO. If unsure, say Y. LFS-5.0 used gcc-2.3.2 so it will need this set to Y. ĸen No. I compiled the kernel without COMPAT_VDSO. Do you think it needs included? The host is running Glibc 2.3.3. I think system may be a LFS 5.1. I didn't write a file anywhere to tell me, but I know it was build April 2004 with the then stable book. Thank, Mykal Funk -- Yes, all the main hits on google for inconsistency detected by ld.so appear to suggest that will fix it. Sorry for the thinko in my earlier reply. You can determine the version of glibc by running /lib/libc.so.6 ĸen Thanks a bunch, Ken. I got a kernerl working! Now to build the LFS 6.5 and see how everything goes. If it all goes well, I can go back to lurking. :) -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
Mykal Funk wrote: Thanks a bunch, Ken. I got a kernerl working! Now to build the LFS 6.5 and see how everything goes. If it all goes well, I can go back to lurking. :) But first, be sure to keep us informed on how things all come out! Mike -- p=p=%c%s%c;main(){printf(p,34,p,34);};main(){printf(p,34,p,34);} Oppose globalization and One World Governments like the UN. This message made from 100% recycled bits. You have found the bank of Larn. I speak only for myself, and I am unanimous in that! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
Ken Moffat wrote: 2009/12/5 William Immendorf will.immend...@gmail.com: Try building the generic EIDE/PATA driver into the kernel. -- William Immendorf The ultimate in free computing. Messages in plain text, please, no HTML. -- Yeah, I think William has probably hit the nail on the head here - you mentioned this was very old hardware. Thanks for the pointers. Once I got the configuration right it would go all the way to loading Init. However, it is now givining an error Inconsistency detected by ld.so: rtld.c: 1180: dl_main: Assertion `(void *) ph-p_vaddr == _rtld_local._dl_sysinfo_dso’ failed! and the kernel panics. I'm considering recompiling Glibc but am unsure if that would fix the problem or cause more. Thanks in advance, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
2009/12/7 Mykal Funk mykalf...@gmail.com: Thanks for the pointers. Once I got the configuration right it would go all the way to loading Init. However, it is now givining an error Inconsistency detected by ld.so: rtld.c: 1180: dl_main: Assertion `(void *) ph-p_vaddr == _rtld_local._dl_sysinfo_dso’ failed! and the kernel panics. I'm considering recompiling Glibc but am unsure if that would fix the problem or cause more. Thanks in advance, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support Do you have CONFIG_COMPAT_VDSO=y ? If so, try turning it off. I'm not sure where it appears in menuconfig, but the help says: config COMPAT_VDSO def_bool y prompt Compat VDSO support depends on X86_32 || IA32_EMULATION ---help--- Map the 32-bit VDSO to the predictable old-style address too. ---help--- Say N here if you are running a sufficiently recent glibc version (2.3.3 or later), to remove the high-mapped VDSO mapping and to exclusively use the randomized VDSO. If unsure, say Y. LFS-5.0 used gcc-2.3.2 so it will need this set to Y. ĸen -- After tragedy, and farce, OMG poneys! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
Ken Moffat wrote: 2009/12/7 Mykal Funk mykalf...@gmail.com: Thanks for the pointers. Once I got the configuration right it would go all the way to loading Init. However, it is now givining an error Inconsistency detected by ld.so: rtld.c: 1180: dl_main: Assertion `(void *) ph-p_vaddr == _rtld_local._dl_sysinfo_dso’ failed! and the kernel panics. I'm considering recompiling Glibc but am unsure if that would fix the problem or cause more. Thanks in advance, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support Do you have CONFIG_COMPAT_VDSO=y ? If so, try turning it off. I'm not sure where it appears in menuconfig, but the help says: config COMPAT_VDSO def_bool y prompt Compat VDSO support depends on X86_32 || IA32_EMULATION ---help--- Map the 32-bit VDSO to the predictable old-style address too. ---help--- Say N here if you are running a sufficiently recent glibc version (2.3.3 or later), to remove the high-mapped VDSO mapping and to exclusively use the randomized VDSO. If unsure, say Y. LFS-5.0 used gcc-2.3.2 so it will need this set to Y. ĸen No. I compiled the kernel without COMPAT_VDSO. Do you think it needs included? The host is running Glibc 2.3.3. I think system may be a LFS 5.1. I didn't write a file anywhere to tell me, but I know it was build April 2004 with the then stable book. Thank, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
2009/12/7 Mykal Funk mykalf...@gmail.com: Ken Moffat wrote: Do you have CONFIG_COMPAT_VDSO=y ? If so, try turning it off. whoops, if not try turning it on. I'm not sure where it appears in menuconfig, but the help says: config COMPAT_VDSO def_bool y prompt Compat VDSO support depends on X86_32 || IA32_EMULATION ---help--- Map the 32-bit VDSO to the predictable old-style address too. ---help--- Say N here if you are running a sufficiently recent glibc version (2.3.3 or later), to remove the high-mapped VDSO mapping and to exclusively use the randomized VDSO. If unsure, say Y. LFS-5.0 used gcc-2.3.2 so it will need this set to Y. ĸen No. I compiled the kernel without COMPAT_VDSO. Do you think it needs included? The host is running Glibc 2.3.3. I think system may be a LFS 5.1. I didn't write a file anywhere to tell me, but I know it was build April 2004 with the then stable book. Thank, Mykal Funk -- Yes, all the main hits on google for inconsistency detected by ld.so appear to suggest that will fix it. Sorry for the thinko in my earlier reply. You can determine the version of glibc by running /lib/libc.so.6 ĸen -- After tragedy, and farce, OMG poneys! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Linux-2.6.30.9 build failure fixed new problem show up
Ken Moffat wrote: The slightly longer-winded version is to build several versions of gcc and binutils, using each to build a newer version. As always, the version of binutils needs to be suitable for the version of gcc but these things are never documented. Looking back to my old notes, I guess the following versions might be worth exploring e.g. first one in /opt/tools1, second in /opt/tools2 - during toolchain builds set PATH to /opt/tools3:/opt/tools2:/opt/tools1: and then the normal PATH after that. binutils-2.14 and gcc-3.3.5 (based loosely on 5.1) binutils-2.16.1 and gcc-4.0.2 binutils-2.17 and gcc-4.1.2 - I would very much hope that is recent enough to build a current system (and particularly a current kernel). Thanks Ken that got the 2.6.30.9 kernel compiled. But I can't seem to boot it. It complains about not having an NFS mount then asks for a root floppy. If I hit a key, the kernel panics. I haven't figured out how to capture the output. I've googled around and am not finding anything. I've checked my Grub settings and they seem fine. I've double-checked the kernel config and it includes support for reiserfs, which is what the LFS 5 is on. Any ideas on what to do next? Thanks in Advance, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
On Sat, Dec 5, 2009 at 4:16 PM, Mykal Funk mykalf...@gmail.com wrote: Thanks Ken that got the 2.6.30.9 kernel compiled. But I can't seem to boot it. It complains about not having an NFS mount then asks for a root floppy. If I hit a key, the kernel panics. I haven't figured out how to capture the output. I've googled around and am not finding anything. I've checked my Grub settings and they seem fine. I've double-checked the kernel config and it includes support for reiserfs, which is what the LFS 5 is on. Try building the generic EIDE/PATA driver into the kernel. -- William Immendorf The ultimate in free computing. Messages in plain text, please, no HTML. -- Every nonfree program has a lord, a master -- and if you use the program, he is your master. Richard Stallman -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure fixed new problem show up
2009/12/5 William Immendorf will.immend...@gmail.com: On Sat, Dec 5, 2009 at 4:16 PM, Mykal Funk mykalf...@gmail.com wrote: Thanks Ken that got the 2.6.30.9 kernel compiled. But I can't seem to boot it. It complains about not having an NFS mount then asks for a root floppy. If I hit a key, the kernel panics. I haven't figured out how to capture the output. I've googled around and am not finding anything. I've checked my Grub settings and they seem fine. I've double-checked the kernel config and it includes support for reiserfs, which is what the LFS 5 is on. Try building the generic EIDE/PATA driver into the kernel. -- William Immendorf The ultimate in free computing. Messages in plain text, please, no HTML. -- Yeah, I think William has probably hit the nail on the head here - you mentioned this was very old hardware. One of the problems with 2.6 was its very long gestation (ISTR in excess of 70 versions of the 2.5 development series) and an enormous number of things changed so there might be a lot more to change once you have it booting. If that turns out to be the case, a description of how to get to 2.6 from 2.4 is in fact at http://www.linuxfromscratch.org/hints/downloads/files/OLD/kernel-2_6-migration.txt To use modules, I *think* the important thing was the 'moveold' which *should* allow both 2.4 and 2.6 kernels to run. As a precaution, copy all the programs from modutils before this, just in case they get overwritten. You should also check that your 2.4 kernel can boot to a useful state without loading any modules, just in case this goes wrong. If that works, you can use modules, then modprobe them as necessary until you find which are necessary. You can find module-init-tools-3.1 at http://www.kernel.org/pub/linux/kernel/people/rusty/modules/ HTH, and good luck! ĸen -- After tragedy, and farce, OMG poneys! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
On Wed, 2009-12-02 at 23:21 -0500, Mykal Funk wrote: The machine has collected dust for the last 5 years. As it is a 486DX, it will take a couple days to see if your suggestions work. And yes, I think I left this one a bit too long. But I like a challenge. Thats why I bother with an old 486 in the first place. Wow... if you *do* get a new LFS build running on that, I'd be curious to know how long it took... 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: Linux-2.6.30.9 build failure
On 12/3/09, Simon Geard wrote: Wow... if you *do* get a new LFS build running on that, I'd be curious to know how long it took... My first guess is 11 days or so. Calculating from: http://www.linuxfromscratch.org/~sbu MHz=100 one_sbu=5848 lfs_6_5_sbus=153 seconds=899422 time=10,9:50:22 -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
linux fan wrote: On 12/3/09, Simon Geard wrote: Wow... if you *do* get a new LFS build running on that, I'd be curious to know how long it took... My first guess is 11 days or so. You don't necessarily have to build on that machine. However, I realize that may be part of the challenge. Mike -- p=p=%c%s%c;main(){printf(p,34,p,34);};main(){printf(p,34,p,34);} Oppose globalization and One World Governments like the UN. This message made from 100% recycled bits. You have found the bank of Larn. I speak only for myself, and I am unanimous in that! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
On 12/3/09, Mike McCarty wrote: You don't necessarily have to build on that machine. However, I realize that may be part of the challenge. Hmm, if LFS 6.5 cross compiles, could you build it on a fast machine for the slow machine and then put it on the slow machine with rsync or something? -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
linux fan wrote: On 12/3/09, Mike McCarty wrote: You don't necessarily have to build on that machine. However, I realize that may be part of the challenge. Hmm, if LFS 6.5 cross compiles, could you build it on a fast machine for the slow machine and then put it on the slow machine with rsync or something? That's the idea, yes. I haven't personally carried out the process, myself, but I've done cross compiles in general for embedded machines, and for Windows machines with Linux, etc. There are those here who have done the compiles on VMs and then installed on real machines later via this technique, and they could probably advise you better than I. Mike -- p=p=%c%s%c;main(){printf(p,34,p,34);};main(){printf(p,34,p,34);} Oppose globalization and One World Governments like the UN. This message made from 100% recycled bits. You have found the bank of Larn. I speak only for myself, and I am unanimous in that! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
linux fan wrote: On 12/3/09, Mike McCarty wrote: You don't necessarily have to build on that machine. However, I realize that may be part of the challenge. Hmm, if LFS 6.5 cross compiles, could you build it on a fast machine for the slow machine and then put it on the slow machine with rsync or something? Yes, as long as you have a 32-bit host system and compile the kernel with the appropriate drivers. I recommend tar. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Linux-2.6.30.9 build failure
I am trying to upgrade an LFS 5 system so that I can build an updated LFS 6.5 system. However the compile fails with this output. MODPOST vmlinux.o WARNING: modpost: Found 5 section mismatch(es). To see full details build you kernel with: 'make CONFIG_DEBUG_SECTION_MISMATCH=y' GEN .version CHK include/linux/compile.h dnsdomainname: Host name lookup failure UPD include/linux/compile.h CC init/version.o LD int/built-in.o LD .tmp_vmlinux1 ld:arch/x86/kernel/vmlinux.lds:473: parse error make: *** [.tmp_vmlinux1] Error 1 Current kernel is linux-2.4.22-openmosix-2, if that helps. I've googled around and haven't found anything. I can't get this kernel to compile and I'm not sure why. Thanks in Advance, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
On Wed, Dec 2, 2009 at 4:06 PM, Mykal Funk mykalf...@gmail.com wrote: Current kernel is linux-2.4.22-openmosix-2, if that helps. I've googled around and haven't found anything. I can't get this kernel to compile and I'm not sure why. Sorry to dissapoint you, but you need a 2.6.18 kernel or up to build LFS 6.5. The only solution is to upgrade your host system. -- William Immendorf The ultimate in free computing. Messages in plain text, please, no HTML. -- Every nonfree program has a lord, a master -- and if you use the program, he is your master. Richard Stallman -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
Mykal Funk wrote: I am trying to upgrade an LFS 5 system so that I can build an updated LFS 6.5 system. However the compile fails with this output. MODPOST vmlinux.o WARNING: modpost: Found 5 section mismatch(es). To see full details build you kernel with: 'make CONFIG_DEBUG_SECTION_MISMATCH=y' GEN .version CHK include/linux/compile.h dnsdomainname: Host name lookup failure UPD include/linux/compile.h CC init/version.o LD int/built-in.o LD .tmp_vmlinux1 ld:arch/x86/kernel/vmlinux.lds:473: parse error make: *** [.tmp_vmlinux1] Error 1 Current kernel is linux-2.4.22-openmosix-2, if that helps. I've googled around and haven't found anything. I can't get this kernel to compile and I'm not sure why. It's been a long time since I head of anyone still using 2.4.x. When LFS first transitioned to the 2.6.x kernels, we didn't have a way to build it other than to load a commercial distro and build from there. Of course someone did the first build to 2.6, but that was back in 2003. We don't have a build path for 2.4.x-2.6.x and we now specify at least 2.6.18. I'd recommend getting ubuntu, fedora, suse, or another commercial distro and load that. From there, you can build LFS-6.5 or LFS-dev. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
2009/12/2 Bruce Dubbs bruce.du...@gmail.com: Mykal Funk wrote: I am trying to upgrade an LFS 5 system so that I can build an updated LFS 6.5 system. However the compile fails with this output. MODPOST vmlinux.o WARNING: modpost: Found 5 section mismatch(es). To see full details build you kernel with: 'make CONFIG_DEBUG_SECTION_MISMATCH=y' GEN .version CHK include/linux/compile.h dnsdomainname: Host name lookup failure UPD include/linux/compile.h CC init/version.o LD int/built-in.o LD .tmp_vmlinux1 ld:arch/x86/kernel/vmlinux.lds:473: parse error make: *** [.tmp_vmlinux1] Error 1 Current kernel is linux-2.4.22-openmosix-2, if that helps. I've googled around and haven't found anything. I can't get this kernel to compile and I'm not sure why. In this case, it appears your binutils is too old. It's been a long time since I head of anyone still using 2.4.x. When LFS first transitioned to the 2.6.x kernels, we didn't have a way to build it other than to load a commercial distro and build from there. Of course someone did the first build to 2.6, but that was back in 2003. We don't have a build path for 2.4.x-2.6.x and we now specify at least 2.6.18. Much as I hesitate to disagree, many people built 2.6 kernels from then-recent LFS builds. AFAIR the big issue was installing module init tools with the correct workarounds so that 2.4 kernels could still be supported. But since then, everything has moved on and I'd be surprised if the recent versions of module-init-tools still support the workaround (and anyway I've forgotten the details of it).. So Bruce's suggestion to use a distro (or the last LFS Live CD - plus a current version of the book and the packages to go with that) is the most efficient use of your time. The alternative is to build a toolchain for the kernel (in /usr/local or ~/ or /opt/something - you only need binutils and gcc's (kernel only needs C, but otehr things in chapter 5 will need C++ so best to build both), build the version of the 2.6 kernel you intend to use (but without modules), and then see if that toolchain is good enough. Unfortunately, picking suitable old versions is guesswork - I thought there were tests in the kernel build system, particularly for gcc, but all I can find from a quick look is Documentation/Changes which still claims gcc-3.2 and binutils-2.12 are adequate - on x86 I'm certain they are unlikely to be adequate. The slightly longer-winded version is to build several versions of gcc and binutils, using each to build a newer version. As always, the version of binutils needs to be suitable for the version of gcc but these things are never documented. Looking back to my old notes, I guess the following versions might be worth exploring e.g. first one in /opt/tools1, second in /opt/tools2 - during toolchain builds set PATH to /opt/tools3:/opt/tools2:/opt/tools1: and then the normal PATH after that. binutils-2.14 and gcc-3.3.5 (based loosely on 5.1) binutils-2.16.1 and gcc-4.0.2 binutils-2.17 and gcc-4.1.2 - I would very much hope that is recent enough to build a current system (and particularly a current kernel). If you want to try this, you could even see if the more recent toolchain will build on your current system. Unfortunately, you will be totally on your own for this, there are no guarantees and it's even possibly that the old glibc might cause problems during the build or the testing. It's also possible that something else entirely is causing your version of ld to complain. The vmlinux.lds file is generated from your kernel .config so we've no idea what is on the line in question, and in any case I doubt there is anyone here who is familiar enough with the toolchain to diagnose the exact problem / required package version. :( In general, once you build LFS you are responsible for updating it, and rebuilding it in due course. I think you've left this one a bit too long. ĸen -- After tragedy, and farce, OMG poneys! -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page
Re: Linux-2.6.30.9 build failure
Ken Moffat wrote: In general, once you build LFS you are responsible for updating it, and rebuilding it in due course. I think you've left this one a bit too long. ĸen The machine has collected dust for the last 5 years. As it is a 486DX, it will take a couple days to see if your suggestions work. And yes, I think I left this one a bit too long. But I like a challenge. Thats why I bother with an old 486 in the first place. Thanks, Mykal Funk -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page