| 1 |
From: Bo Ørsted Andresen <bo.andresen@××××.dk> |
| 2 |
Subject: Re: [gentoo-user] /etc/make.conf: changing CHOST on same system |
| 3 |
Date: Sat, 2 Sep 2006 11:30:55 +0200 |
| 4 |
|
| 5 |
> On Saturday 02 September 2006 10:58, Meino Christian Cramer wrote: |
| 6 |
> > So, is it correct / necessary in my case to do the following for |
| 7 |
> > upgrading to gcc-4.1: |
| 8 |
> > |
| 9 |
> > Correct the CHOST fault |
| 10 |
> > emerge -e world |
| 11 |
> > emerge glibc |
| 12 |
> > emerge -e world |
| 13 |
> > emerge gcc |
| 14 |
> > emerge -e world |
| 15 |
> |
| 16 |
> ...?. |
| 17 |
> |
| 18 |
> I guess you didn't follow this list too carefully the last couple of days... |
| 19 |
> Anyway, no you shouldn't emerge -e world more than once! |
| 20 |
> |
| 21 |
> 1. Correct the CHOST in make.conf |
| 22 |
> 2. Make sure the nptl and nptlonly use flags are enabled. |
| 23 |
> 3. *If* you want to upgrade your profile in the near future this may be a good |
| 24 |
> time to do so. This step is optional. You may have a look at [1] for more |
| 25 |
> info about this. |
| 26 |
> 4. Follow the gcc upgrade guide [2]. |
| 27 |
> |
| 28 |
> Following the gcc upgrade guide will upgrade glibc and pretty much everething |
| 29 |
> else that isn't quite up to date during the emerge -e world. Make sure that |
| 30 |
> you follow the "General Upgrade Instructions" carefully though. |
| 31 |
> |
| 32 |
> [1] http://www.gentoo.org/doc/en/gentoo-upgrading.xml |
| 33 |
> [2] http://www.gentoo.org/doc/en/gcc-upgrading.xml |
| 34 |
> -- |
| 35 |
> Bo Andresen |
| 36 |
|
| 37 |
Hi Bo ! |
| 38 |
|
| 39 |
...unfortunately I got stuck again. |
| 40 |
|
| 41 |
What I did: |
| 42 |
Changed CHOST from i386* i686*, ensure that nptl and nptlonly are USEd. |
| 43 |
(I am running an AMD Athlon(tm) 64 X2 Dual Core Processor 3800+) |
| 44 |
|
| 45 |
emerge -uav gcc |
| 46 |
|
| 47 |
The result was: |
| 48 |
The first package (it was not the gcc), what was tried to compile, failed with |
| 49 |
"Compiler cannot create output" (or similiar, dont remember |
| 50 |
exactly...) |
| 51 |
|
| 52 |
Looks like, as the new CHOST results in looking for |
| 53 |
/usr/lib/gcc/i686-pc-linux-gnu (which is not existing) instead of |
| 54 |
/usr/lib/gcc/i386-pc-linux-gnu (which still does exist). |
| 55 |
|
| 56 |
Ok, but when I reset CHOST to i386* again, the new glibc will fail to |
| 57 |
compile, since it insist of haveing a CPU better then i386* (which |
| 58 |
was the inital reason for posting my problems.) |
| 59 |
|
| 60 |
Do I have a "recursive Armageddon" here ? |
| 61 |
|
| 62 |
Since I will replace gcc, would it be a soultion to rename all |
| 63 |
occurences of i386-pc-linux-gnu in my root directory tree with |
| 64 |
i686-pc-linux-gnu ??? Or symlink it? Or do I shoot myself just in my |
| 65 |
other foot this way? |
| 66 |
I know that this "solution" would be not gentleman like but a hack of a |
| 67 |
rookie...but...when gentlemen does not succeed ??? ;) |
| 68 |
|
| 69 |
By the way: |
| 70 |
I looked for sane CFLAGS for my CPU by best performance. |
| 71 |
I am not sure for the -msse3 flag... |
| 72 |
The docs say: You can enable -msse3 for the CPU (mentioned above) |
| 73 |
when the "pni" entry is in /proc/cpuinfo (which is in my case). On |
| 74 |
the other side /proc/cpuinfo mentioned sse and sse2 but not sse3: |
| 75 |
|
| 76 |
processor : 0 |
| 77 |
vendor_id : AuthenticAMD |
| 78 |
cpu family : 15 |
| 79 |
model : 43 |
| 80 |
model name : AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ |
| 81 |
stepping : 1 |
| 82 |
cpu MHz : 2002.999 |
| 83 |
cache size : 512 KB |
| 84 |
physical id : 0 |
| 85 |
siblings : 2 |
| 86 |
core id : 0 |
| 87 |
cpu cores : 2 |
| 88 |
fdiv_bug : no |
| 89 |
hlt_bug : no |
| 90 |
f00f_bug : no |
| 91 |
coma_bug : no |
| 92 |
fpu : yes |
| 93 |
fpu_exception : yes |
| 94 |
cpuid level : 1 |
| 95 |
wp : yes |
| 96 |
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca |
| 97 |
cmov pat pse36 clflush mmx fxsr #sse# #sse2# ht syscall nx |
| 98 |
mmxext fxsr_opt lm 3dnowext 3dnow #pni# lahf_lm cmp_legacy ts |
| 99 |
fid vid ttp |
| 100 |
bogomips : 4012.21 |
| 101 |
|
| 102 |
Same goes for processor 1, but with slightly different bogomips |
| 103 |
(Sometimes even identical things seems to be different... ;O) . |
| 104 |
|
| 105 |
I dont know, how to get the information, whether my CPU has a |
| 106 |
Winchester, Manchester, Venice or whatever die... |
| 107 |
|
| 108 |
Last thing: |
| 109 |
When the gentoo docs talk about "AMD64" without mentioning, whether |
| 110 |
it is meant a 32bit- or 64bit-system....is it valid for 32bit AND |
| 111 |
64bit or only 64bit or... |
| 112 |
Sometimes I am not sure whether the docs are talking about "my |
| 113 |
system" since I am running a AMD64 -- a 64bit CPU -- in 32bit mode. |
| 114 |
|
| 115 |
|
| 116 |
Thanks a lot in advance for any help out of my homebrewed disaster... |
| 117 |
Keep hacking! |
| 118 |
mcc |
| 119 |
|
| 120 |
-- |
| 121 |
gentoo-user@g.o mailing list |
Archives