[1.9/trunk] Bus error / abort trop with latest sources

Ruby
1

Hello,

I’m compiling Ruby 1.9 from the SVN tree, latest changeset is 12384 on
OS
X/PPC 10.4.9.

It fails during compilation with the following message:

…
gcc -O -pipe -pipe -fno-common -L. main.o libruby-static.a
-lpthread -ldl -lobjc -o miniruby
: – stack frame ------------
0000 (0x126000): 00000000 <- lfp <- dfp
– control frame ----------
c:0001 p:---- s:0001 b:-001 l:000000 d:000000 ------

[BUG] Bus Error
ruby 1.9.0 (2007-05-25) [powerpc-darwin8.9.0]

Running miniruby and pressing ^D leads to the same message. miniruby -v,
same.

Am I the only one? I’ve tried make clean, rerunning configure, changing
-O2
into -O, no success.

Thanks.

2

Hi,

At Fri, 25 May 2007 21:43:54 +0900,
Ollivier R. wrote in [ruby-talk:252956]:

[BUG] Bus Error
ruby 1.9.0 (2007-05-25) [powerpc-darwin8.9.0]

What command failed?

Running miniruby and pressing ^D leads to the same message. miniruby -v,
same.

It’s a known bug.

Index: compile.c

— compile.c (revision 12384)
+++ compile.c (working copy)
@@ -145,5 +145,8 @@ rb_iseq_compile(VALUE self, NODE *node)
GetISeqPtr(self, iseq);

  • if (nd_type(node) == NODE_SCOPE) {
  • if (node == 0) {
  • COMPILE(ret, “nil”, node);
  • }
  • else if (nd_type(node) == NODE_SCOPE) {
    /* iseq type of top, method, class, block */
    set_local_table(iseq, node->nd_tbl);
    @@ -198,7 +201,4 @@ rb_iseq_compile(VALUE self, NODE *node)
    COMPILE(ret, “defined guard”, node);
    }
  • else if (node == 0) {
  •  COMPILE(ret, "nil", node);
  • }
    else {
    rb_bug(“unknown scope”);
3

In article
[email protected],
Nobuyoshi N. [email protected] wrote:

What command failed?

My guess is that it is miniruby which fails.

Running miniruby and pressing ^D leads to the same message. miniruby -v,
same.

It’s a known bug.

Is it committed in trunk?

Any news on the bug where trying to use “-r debug” does not work because
a hook is not yet available (according to ko1)? I can not switch (and
therefore test) 1.9 is the debugger is not usable…

Thanks.

4

Hi,

At Wed, 30 May 2007 23:15:09 +0900,
Ollivier R. wrote in [ruby-talk:253586]:

What command failed?

My guess is that it is miniruby which fails.

Yes, obviously. What I asked is what miniruby was about to do.

Running miniruby and pressing ^D leads to the same message. miniruby -v,
same.

It’s a known bug.

Is it committed in trunk?

Yes.

Sat May 26 00:38:21 2007 Nobuyoshi N. [email protected]

  • eval.c (ruby_exec_internal): do nothing if no code.

  • compile.c (rb_iseq_compile): check node if NULL before check
    nd_type. [ruby-talk:252956]

Any news on the bug where trying to use “-r debug” does not
work because a hook is not yet available (according to ko1)?
I can not switch (and therefore test) 1.9 is the debugger is
not usable…

Thank you, I’ve investigated it, and will fix it.

5

Nobuyoshi N. wrote:

Yes, obviously. What I asked is what miniruby was about to do.

I think the very first invocation of miniruby during the compilation
process is failling.

Is it committed in trunk?

The compilation still faisl with the bus error but as for miniruby -v:

Now, it is just looping away, eating CPU cycles :frowning:

gcc -O -pipe -pipe -fno-common -I. -I.ext/include/powerpc-darwin8.9.0
-I…/include -I… -DRUBY_EXPORT -c …/main.c
gcc -O -pipe -pipe -fno-common -L. main.o libruby-static.a
-lpthread -ldl -lobjc -o miniruby
: – stack frame ------------
0000 (0x128000): 00000000 <- lfp <- dfp
– control frame ----------
c:0001 p:---- s:0001 b:-001 l:000000 d:000000 ------

[BUG] Bus Error
ruby 1.9.0 (2007-06-10) [powerpc-darwin8.9.0]

make: *** [.rbconfig.time] Abort trap
353 [10:50] roberto@robertotmp-al:yarv/build> ./miniruby -v
^D
^D

^C^C^C^C^Zload: 0.28 cmd: miniruby 27571 running 0.28u 3.49s
load: 0.28 cmd: miniruby 27571 running 0.39u 4.78s

^C^C^C^Z

I have to kill it in another window. It is a plain 1.9 at changeset
#12515.

Thank you, I’ve investigated it, and will fix it.

As soon as I get a working ruby, I’ll test it, thanks.

6

Ollivier R. wrote:

[BUG] Bus Error
ruby 1.9.0 (2007-06-10) [powerpc-darwin8.9.0]

Interestingly enough, the same sources on an Intel Mac (MBP) does not
dump core.

I have to kill it in another window. It is a plain 1.9 at changeset
#12515.

This happens on the macbook pro as well though.

Thank you, I’ve investigated it, and will fix it.

As soon as I get a working ruby, I’ll test it, thanks.

It does not work.