On Sunday 01 January 2006 20:55, Dan Nicholson wrote: > On 1/1/06, Randy McMurchy <[EMAIL PROTECTED]> wrote: > > Allard Welter wrote these words on 01/01/06 12:06 CST: > > > I post this here as these errors occurred while compiling the > > > full gcc suite once rebooted into lfs-6.1.1. It is perhaps > > > relevant to lfs as well. The tests that fail are the ones > > > involving pre-compiled headers > > > (gcc-3.4.3/gcc/testsuite/gcc.dg/pch/) > > > > I show segmentation faults in my build of gcc-3.4.3 as well. I > > believe this pch issue is known about, and I think a patch was > > just added to the repo to fix it and that the errata for 6.1.1 > > has been updated to mention the fix. I may be wrong, but it seems > > Dan just did something about this. > > > > Check and see if this patch won't help: > > http://www.linuxfromscratch.org/patches/downloads/gcc/gcc-3.4.3-pch > >-1.patch > > I don't think the pch failures have anything to do with tmpfs in > gcc-3.4.3. Or, I should say that I hope that tmpfs hasn't caused new > failures. > > I assume you're running linux-2.6.12+? A new feature was introduced > called exec-shield-randomize or something that breaks pch. Take a > look at the patch and follow the links if you're curious. This has > been fixed in gcc-3.4.5 and gcc-4. Something will be going in the > lfs-6.1.1 errata within the week.
Actually no - it is tmpfs on /tmp. First of all my bad for not checking the errata. My first hunch was kernel versions as the system I chrooted from was running linux-2.6.10. I tried 2.11 and 2.10 with the same results before checking other things. I've just finished compiling and testing gcc-3.4.5 with tmpfs mounted on /tmp without problems. It's been a _long_ time since I've had the time to build a new system so I figured the odds are that I'm overlooking something. I'm now considering whether to start all over again with cvs as i've usually done in the past - I thought for a change lets build the stable versions but now I worry that somewhere a package will use precompiled headers. Seems there will always be something to puzzle out. Allard. -- http://linuxfromscratch.org/mailman/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
