Mail Archives: djgpp/2009/03/13/21:45:04
Hi,
On Jan 23, 11:32=A0pm, Rugxulo <rugx DOT DOT DOT AT gmail DOT com> wrote:
> Blah,
>
> On Dec 31 2008, 11:06=A0pm, Rugxulo <rugx DOT DOT DOT AT gmail DOT com> wrote:
>
> > Okay, after scratching my head recompiling things over and over again
> > for a while, I think it turns out that GNU pthreads 2.0.7 (beta only)
> > seems to require 686+ (yet another CMOV.. compile error). The /
> > current/ version works fine in DOSBox 0.72 unlike /beta/'s. Lemme go
> > test on my real 586 and see if the p7zip with "current" pth works
> > okay ... yup. (Michael Kostylev was the guy who previously compiled
> > 4.57 and 4.58, but he used old FSU Pthreads instead for whatever
> > reason.)
I tried again with FSU Pthreads, but it didn't compile p7zip
correctly.
> Ironically, there's some bug somewhere that makes it not work under
> FreeDOS by default, so I had to disable a bunch of MT (multithreading)
> defines. It worked fine in DR-DOS or WinXP without messing with that,
> though. And I tested various FD kernels, so whatever. I'm not
> experienced enough to really debug it either. =A0 :-/
>
> > Very annoying. Oh well. Plus, to make matters worse, I can't seem to
> > recompile pth 2.0.7, the instructions don't seem to work (autoconf
> > temp files "file not found" errors, ring a bell anyone?): "bash djgpp/
> > djconfig.sh".
>
> pconf=3D"/tmp/" (or whatever) needs to be changed to "/dev/env/DJDIR/
> tmp" although even that only compiles to PTHSEM.A and no other libs,
> so I'm not entirely sure how the pre-built packages were made. And I
> haven't tested that version much (if at all ... I forget).
Tried this too, also didn't compile p7zip.
> P.S. I have srcs, patches, and binaries on my site for this (using
> WATT-32 as socket lib). Michael Kostylev did compile a (smaller)
> version with FSU Pthreads (ugh, bitrotted, hard to get working). But
> neither works without an FPU (unlike his earlier 4.57 build,
> strangely). So who knows, bugs all around. Almost better to just use
> HX + latest 7ZA (which does work on my 486 Sx, go figure). =A0 :-P
Not sure exactly, but here's what a debug build of p7zip says (with or
without DJGPP emulation available) on my 486 Sx/25:
[ MS-DOS 6.22 ] Thu 02-26-2009>7za a blah \aut* \config*
Cannot continue from exception, Exiting due to signal SIGNOFP
Coprocessor not available at eip=3D000715b1
eax=3D00000000 ebx=3D00000000 ecx=3D00015f90 edx=3D00177310 esi=3D0021e084
edi=3D000714f0
ebp=3D0020d7a8 esp=3D0021dfc4 program=3DC:\TEMP\7ZA.EXE
cs: sel=3D00a7 base=3D10000000 limit=3D0022ffff
ds: sel=3D00af base=3D10000000 limit=3D0022ffff
es: sel=3D00af base=3D10000000 limit=3D0022ffff
fs: sel=3D008f base=3D00009d00 limit=3D0000ffff
gs: sel=3D00bf base=3D00000000 limit=3D0010ffff
ss: sel=3D00af base=3D10000000 limit=3D0022ffff
App stack: [0020d800..0018d800] Exceptn stack: [0018c9ec..0018aaac]
Call frame traceback EIPs:
0x000715b1 __pth_scheduler+193, file pth_sched.c
- Raw text -