Messages in this thread | | | Date | Fri, 26 Mar 1999 10:54:05 -0800 (PST) | From | Linus Torvalds <> | Subject | Re: lots of 2.2.4 oopses |
| |
On Fri, 26 Mar 1999, Horst von Brand wrote: > Valient Gough <vgough@pobox.com> said: > > I've been running Linux 2.2.4 compiled with gcc 2.7.2.3 for nearly two days > > now without problems. > > > > When I ran it with egcs-1.1.2, using exactly the same options, etc, > > I couldn't run for more then an hour before getting problems. > > > > So, it's starting to look more and more likely that the problem is due to > > the combination of egcs-1.1.2 and Linux 2.2.4... > > For me it works fine, and has for several -ac and even 2.2.5-pre1 versions > too. i586 UP, IDE only, haifa-enabled egcs-1.1.2, binutils-2.9.1.0.22b, > optimization flags changed to -O2 -fomit-frame-pointer -march=pentium.
Note that it's likely that there is just a few places (or even just _one_) where Linux and egcs disagree about something. So it can be very setup- dependent, and it could also be a question of timing.
For example, it could be a case where egcs actually does something that is correct ANSI C - re-orders stores to different memory locations because egcs did the alias analysis and determined that they cannot clash. And it may be that the kernel depends on the exact order of stores. That would be a kernel bug.
Equally possible is that egcs just gets the alias calculations wrong, or that there is some other bug that just makes it generate bad code under certain circumstances. Finding out exactly what the problem is can be _very_ hard: the oops that was posted looked basically like memory corruption, so the bug was not actually likely to be anywhere close to where the crash actually happened.
Linus
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |