> Luke Palmer <[EMAIL PROTECTED]> wrote:
>
> > It appears JIT is entirely broken. It broke recently, as it was
> > working well for me just a couple days ago.
>
> > I'm running i686 (P3) Linux, gcc-3.2.2
>
> > I get segfaults with both imcc -Oj and parrot -j (with assemble.pl) on
> > mandel.pasm and a bunch of others.
>
> Works for me. And could you be more detailed on "entirely broken"?
>
> $ perl t/harness -j t/*/*.t
> $ IMCC=languages/imcc/imcc perl t/harness -j t/*/*.t
Doesn't work for me. It fails almost half the tests due to core
dumps. That's what I mean by "entirely broken". Only two of the
examples run, bsr and local_label.
It's that JIT_CGP thingy (in jit/i386/jit_emit.h) that's causing all
the trouble on my box. I comment out the define of that and it
works. Indeed, 1.62 used an incorrect #if, and it didn't break until
it was corrected. I am using gcc-3.2.2, and using computed goto...
Maybe that special case was for gcc-2.9's? Maybe I have a screwy
system for some reason (it's a clean gentoo-1.4 ... don't know why it would
be screwy). Hmm, we did configure our kernel to have a non-executable
stack. I don't know much about JIT, but could that be the problem?
Here's my test report:
% perl t/harness -j t/*/*.t
...
Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/op/arithmetics.t 31 7936 38 31 81.58% 3-10 14-26 28 30-38
t/op/bitwise.t 10 2560 20 10 50.00% 11-20
t/op/conv.t 4 1024 5 4 80.00% 1-4
t/op/gc.t 5 1280 5 5 100.00% 1-5
t/op/integer.t 19 4864 38 19 50.00% 3-9 22-32 38
t/op/interp.t 1 256 2 1 50.00% 1
t/op/jit.t 7 1792 57 7 12.28% 44-47 51 53-54
t/op/lexicals.t 4 1024 6 4 66.67% 2 4-6
t/op/macro.t 2 512 15 2 13.33% 7 9
t/op/number.t 10 2560 37 10 27.03% 3-4 6 9-10 25-27
34-35
t/op/rx.t 13 3328 23 13 56.52% 1 4-5 7-8 10 12
14-17 19 21
t/op/stacks.t 9 2304 40 9 22.50% 7 30-32 34-35 37
39-40
t/op/string.t 15 3840 102 15 14.71% 8 81-82 84-86 90-91
93 96-100
10
t/op/time.t 1 256 4 1 25.00% 3
t/op/trans.t 18 4608 18 18 100.00% 1-18
t/op/types.t 1 256 2 1 50.00% 1
t/pmc/array.t 4 1024 9 4 44.44% 6-9
t/pmc/coroutine.t 2 512 3 2 66.67% 2-3
t/pmc/iter.t 1 256 2 1 50.00% 2
t/pmc/multiarray.t 2 512 3 2 66.67% 2-3
t/pmc/perlarray.t 8 2048 22 8 36.36% 1-3 15 19-22
t/pmc/perlhash.t 11 2816 25 11 44.00% 5 7-9 15 17-19 22-24
t/pm/perlint.t 2 512 5 2 40.00% 4-5
t/pmc/perlstring.t 4 1024 13 4 30.77% 2-4 7
t/pmc/pmc.t 44 11264 83 44 53.01% 12-28 34 36 39-42
44-45 55-58
60-72 81 83
t/pmc/prop.t 5 1280 6 5 83.33% 1-3 5-6
t/pmc/scratchpad.t 4 1024 4 4 100.00% 1-4
t/pmc/sub.t 5 1280 5 5 100.00% 1-5
20 subtests skipped.
Failed 28/49 test scripts, 42.86% okay. 242/699 subtests failed,
65.38% okay.
Luke
> leo
