1 |
On 10/11/2014 07:57, "Paweł Hajdan, Jr." wrote: |
2 |
[snip] |
3 |
> I'd like to ask for every possible help with |
4 |
> <https://bugs.gentoo.org/show_bug.cgi?id=461954> blockers on amd64 and |
5 |
> x86. There are harder problems on arches like MIPS, but newer gcc is not |
6 |
> as urgent there I think. |
7 |
|
8 |
I've whittled down my MIPS problem (PR61538 // bug #516548) down to something |
9 |
screwy in the newer gcc-4.8 atomics for MIPS that appears to only affect |
10 |
R10000-based processors (so far). I am assuming it doesn't affect newer |
11 |
MIPS32/MIPS64 chips because someone would have caught that by now. It's got |
12 |
something to do with the new c++11 memory models. |
13 |
|
14 |
gcc-4.7 and earlier used __sync_lock_test_and_set() as a placeholder for |
15 |
atomic_exchange_acq(), which is an acquire memory op. That works fine. But in |
16 |
gcc-4.8, they added actual assembler for atomic_exchange_acq(), and that is |
17 |
somehow not playing nice with glibc's futex code on an R10000. Kernel-side, it |
18 |
gets stuck somewhere deep in freezable_schedule() and won't exit until you |
19 |
ctrl+c the stuck process. |
20 |
|
21 |
gcc upstream seems uninterested in addressing it right now, so I need to go |
22 |
poke libc-alpha next, I guess. Probably going to need a sharper stick. |
23 |
|
24 |
MIPS has no stable, so don't let it hold you guys back. Just keep gcc-4.7.x in |
25 |
the tree for a few more years. I've got time until glibc requires gcc-4.8 to |
26 |
get this bug figured out and fixed :) |
27 |
|
28 |
-- |
29 |
Joshua Kinard |
30 |
Gentoo/MIPS |
31 |
kumba@g.o |
32 |
4096R/D25D95E3 2011-03-28 |
33 |
|
34 |
"The past tempts us, the present confuses us, the future frightens us. And our |
35 |
lives slip away, moment by moment, lost in that vast, terrible in-between." |
36 |
|
37 |
--Emperor Turhan, Centauri Republic |