1 |
Michael Haubenwallner wrote: |
2 |
> On Mon, 2008-07-21 at 12:10 +0200, Michael Haubenwallner wrote: |
3 |
>> On Thu, 2008-07-17 at 11:52 -0500, Greg wrote: |
4 |
>>> Jeremy Olexa wrote: |
5 |
>>> |
6 |
>>>> On Thu, Jul 17, 2008 at 11:11 AM, Greg <trigggl@×××××.com> wrote: |
7 |
>>>> |
8 |
>>>> |
9 |
>>>> |
10 |
>>>>> Out of pure optimism, I'm trying to install mplayer. I doubt it will work, |
11 |
>>>>> but I'm going to try. I've been using --nodeps on a package by package |
12 |
> |
13 |
>> It is the x-modular.eclass pulling in either binutils or odcctools. |
14 |
>> You might want to allow sys-devel/native-cctools too to suffice the |
15 |
>> binutils dependency. IFF that really works, we can commit something like |
16 |
>> this: |
17 |
> |
18 |
> Greg, I've fixed and keyworded all the prerequisites of mplayer for |
19 |
> '~ppc-aix' now, but not mplayer itself: |
20 |
|
21 |
Thanks. |
22 |
> |
23 |
> The first problem is that GNU 'uname -p' gives 'unknown' for the |
24 |
> cpu-type, while '/bin/uname -p' correctly states 'powerpc' - so |
25 |
> mplayer's configure correctly detects 'ppc' architecture and does its |
26 |
> job if you (temporarily) remove $EPREFIX/usr/bin/uname and |
27 |
> $EPREFIX/bin/uname. This really should be fixed in GNU coreutils. |
28 |
> |
29 |
> But when compiling, there are some assembler errors: |
30 |
> line 23: Instruction stvx is not implemented in the current assembly mode PPC. |
31 |
|
32 |
Mine stops at: |
33 |
|
34 |
powerpc-ibm-aix5.2.0.0-gcc -Wdisabled-optimization -Wno-pointer-sign |
35 |
-Wdeclaration-after-statement -I. -Wall -Wno-switch -Wpointer-arith |
36 |
-Wredundant-decls -O4 -pipe -ffast-math -fomit-frame-pointer |
37 |
-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -DHAVE_CONFIG_H |
38 |
-I/home/greg/gentoo/usr/include/freetype2 |
39 |
-I/home/greg/gentoo/usr/include -c -o libmpeg2/cpu_state.o |
40 |
libmpeg2/cpu_state.c |
41 |
Assembler: |
42 |
line 14: invalid opcode or pseudo-op |
43 |
line 16: invalid opcode or pseudo-op |
44 |
line 18: invalid opcode or pseudo-op |
45 |
line 20: invalid opcode or pseudo-op |
46 |
line 22: invalid opcode or pseudo-op |
47 |
line 24: invalid opcode or pseudo-op |
48 |
line 26: invalid opcode or pseudo-op |
49 |
line 28: invalid opcode or pseudo-op |
50 |
line 30: invalid opcode or pseudo-op |
51 |
line 32: invalid opcode or pseudo-op |
52 |
line 34: invalid opcode or pseudo-op |
53 |
line 35: invalid opcode or pseudo-op |
54 |
line 55: invalid opcode or pseudo-op |
55 |
line 57: invalid opcode or pseudo-op |
56 |
line 59: invalid opcode or pseudo-op |
57 |
line 61: invalid opcode or pseudo-op |
58 |
line 63: invalid opcode or pseudo-op |
59 |
line 65: invalid opcode or pseudo-op |
60 |
line 67: invalid opcode or pseudo-op |
61 |
line 69: invalid opcode or pseudo-op |
62 |
line 71: invalid opcode or pseudo-op |
63 |
line 73: invalid opcode or pseudo-op |
64 |
line 75: invalid opcode or pseudo-op |
65 |
line 76: invalid opcode or pseudo-op |
66 |
make: *** [libmpeg2/cpu_state.o] Error 1 |
67 |
make: *** Waiting for unfinished jobs.... |
68 |
In file included from libfaad2/syntax.c:32: |
69 |
libfaad2/common.h:354:1: warning: "_ISOC99_SOURCE" redefined |
70 |
|
71 |
The use of '-ffast-math' kind of concerns me. |
72 |
> |
73 |
> After some searching, the native (aix5.3 here) assembler would accept |
74 |
> this with '-m970', but gcc doesn't support this earlier than in gcc-4.3 |
75 |
> on aix5.3, and only with '-mcpu=970' or '-mcpu=G5'. |
76 |
|
77 |
I wonder if it will accept '-mcpu=power3'? |
78 |
> |
79 |
> Will leave mplayer for now, although it _is_ designed to basically work |
80 |
> on ppc-aix. |
81 |
|
82 |
|
83 |
|
84 |
Greg Trigg |
85 |
Jacksonville, AR |