1 |
Apparently, though unproven, at 23:24 on Monday 11 October 2010, Daniel |
2 |
Pielmeier did opine thusly: |
3 |
|
4 |
> Alan McKinnon schrieb am 11.10.2010 22:39: |
5 |
> > Uh-oh. |
6 |
> > |
7 |
> > genlop started failing today with the mysterious error "Illegal |
8 |
> > instruction", and it's consistent - every time. That's all the message, |
9 |
> > nothing else: |
10 |
> > |
11 |
> > $ genlop -t portage |
12 |
> > Illegal instruction |
13 |
> > |
14 |
> > Now emerge dbus-glib fails similarly: |
15 |
> > |
16 |
> > /bin/sh: line 21: 1084 Illegal instruction /usr/bin/gtkdoc-rebase -- |
17 |
> > relative --dest-dir=/var/tmp/portage/dev-libs/dbus-glib-0.88/image/ |
18 |
> > --html- dir=${installdir} |
19 |
> > |
20 |
> > |
21 |
> > I don't really know where to start looking..... |
22 |
> > I just know Google is going to give me millions of useless hits with that |
23 |
> > search, but I'll hope over to b.g.o. meanwhile and poke around unless |
24 |
> > someone else has a better idea. |
25 |
> |
26 |
> Google has something to say about this. |
27 |
> |
28 |
> Recently changed CFLAGS. |
29 |
> Wrong CFLAGS. |
30 |
> Compiler has problems with march native. |
31 |
> Glibc corruption. |
32 |
|
33 |
|
34 |
It's none of those apparently. I checked CFLAGS set by the ebuild in the |
35 |
emerge log before posting and they looked fine. gcc was last updated a month |
36 |
ago and the machine gets updated almost daily. |
37 |
|
38 |
glibc seems possible but it's a moot point, especially as after investigating |
39 |
memory at Mark's suggestion, genlop runs fine now, world updates successfully |
40 |
and 2 ./configure errors about aclocal (that I didn't even mention before) |
41 |
have gone away. |
42 |
|
43 |
I should probably start treating this poor machine more like a notebook and |
44 |
less like a high performance machine - running flat out almost 24/7 is |
45 |
probably outside of it's design spec :-) |
46 |
|
47 |
|
48 |
-- |
49 |
alan dot mckinnon at gmail dot com |