> (replying to self is silly : call this an update to the bug report)
>
> On Wed, 23 Jul 1997, Teunis Peters wrote:
>
> FWIW the recompile of X fixed XWindows..... (it works :)
> If anyone wants I compiled VGA16, SVGA, and S3... but as a warning: I use
> glibc-2.0.4 and pentium-gcc-970706....
>
You might be interested in a bug I found/reported and am wasting my
time trying to fix :) It is present in every version later than 522.
-mno-schedule-prologue or -fno-peephole will supress this. The problem
looks like it's in cse though as..
-O1 -fcse-skip-blocks -fno-defer-pop -fno-opt-reg-use -fno-thread-jumps
-fno-sjlj-exceptions -fno-keep-static-consts
-fno-reduce-index-givs -fno-function-cse -dap -o xx af_inet.cpp..
still has the problem.
inet_ioctl:
pushl %ebp / 750 movsi-2
movl %esp,%ebp / 752 movsi+1/1
pushl %eax / 754 movhicc_1+1
pushl %eax <=== the problem
pushl %edi / 755 movsi-2
pushl %esi / 756 movsi-2
pushl %ebx / 757 movsi-2
movl 16(%ebp),%ebx / 8 movsi+1/2
movl 8(%ebp),%eax / 670 movsi+1/2
movl 24(%eax),%eax / 14 movsi+1/2
movl %eax,-8(%ebp) / 673 movsi+1/1
movl 12(%ebp),%edx / 676 movsi+1/2
addl $-35073,%edx / 643 addsi3+1/1
cmpl $128,%edx / 644 cmpsi_1/1
ja .L139 / 645 bgtu+1
jmp *.L142(,%edx,4) / 648 tablejump
-Mike