Date: Fri, 21 Jul 2000 12:20:19 -0600 From: Nick Jones To: pgcc AT delorie DOT com Subject: Re: Problems compiling kernel with pgcc 2.95.2 Message-ID: <20000721122018.A32663@ductape.net> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.2i In-Reply-To: ; from strasbur@chkw386.ch.pwr.wroc.pl on Fri, Jul 21, 2000 at 10:50:28AM +0000 X-Operating-System: Linux K6II 2.4.0-test3 Reply-To: pgcc AT delorie DOT com Errors-To: nobody AT delorie DOT com X-Mailing-List: pgcc AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk You may want to try '-mno-mmx' to ensure that mmx code isn't being dropped in. Brian mentioned that he was running a non-mmx pentium chip. PGCC may be assuming that either since mmx is supported on the host, that it is supported on the target machine too. Pentiums swing two ways... PGCC also could be assuming that all Pentiums support mmx. -=-=-=-=-=-=-=-Previous Message(s)-=-=-=-=-=-=-=- > Brian Foddy wrote: > >with pgcc 2.95.2 loaded. I am trying to build a kernel on this machine > >for a Pentium 100 (no MMX) running Mandrake 6.1. The result of every > >normal compile is a kernel that will not boot. I see the loading > >message after lilo with most of the dots, then the machine reboots. > Pgcc probably inserts P6 instructions into the code, even if it should not > do it. I have seen similar problem with the userland program. It ran fine > on P6, but failed to work on P5 (compiled with -march=pentium, of course). > So, "Pentium GCC" became "Pentium Pro GCC":]. > >Also, if I build a PentiumPro kernel for the P2, it seems to work fine > >also. > They are both based on the P6 core. > For pentiums, I downgraded to pgcc 1.0.3a. It seems to be the latest > version with bugs documented well. It produces working code, if the > "unstable" compiler options are not used. I also don't trust the 2.91.66 > release (egcs/pgcc 1.1.2). Egcs 2.91.66 gives bogus code, when the quantum > chemistry program Gaussian is compiled with it. I had also problems with > other floating point intensive code compiled with pgcc-1.1. > Krzysztof