problem bootstrapping the sys0.ji file

[spott]

I'm running into an issue with compiling julia:

https://gist.github.com/3107999

I'm compiling julia through homebrew, however it appears to be a julia problem that I'm having:

staticfloat/homebrew-julia#1

Let me know if I need to reproduce the problem somewhere else.

-Andrew

[pao]

Mac people, assemble! cc @ViralBShah @StefanKarpinski

(Do we need a "mac" tag?)

[ViralBShah]

I think a mac tag is a good idea, and might as well add linux while we are at it. We already have windows.

-viral

On 14-Jul-2012, at 6:17 AM, pao wrote:

Mac people, assemble! cc @ViralBShah @StefanKarpinski

(Do we need a "mac" tag?)

---

Reply to this email directly or view it on GitHub:
#1052 (comment)

[pao]

Mac tag done (in brushed metal!)

[staticfloat]

Last time I had a problem like this, it was actually @JeffBezanson who knew what was going on, presumably due to his expertise in the whole sys0.ji business. :)

[nolta]

@spott, can you try building julia without using homebrew, and see if you run into the same problem.

[spott]

@nolta, I had done this before, and it worked, however, when I just did it, it failed. I'm not sure what changed in my environment since it worked before (well, a lot of things probably changed, it went from a brand new computer, to having a bunch of custom homebrew install stuff, but I don't know specifically what might have caused a problem.):

LINK usr/bin/julia-release-basic
LINK usr/bin/julia-release-readline
LINK usr/bin/julia-release-webserver
JULIA usr/lib/julia/sys0.ji
/bin/sh: line 1: 85008 Segmentation fault: 11  /Users/spott/Documents/Code/julia/usr/bin/julia-release-readline -b stage0.jl
make[1]: *** [/Users/spott/Documents/Code/julia/usr/lib/julia/sys0.ji] Error 139
make: *** [release] Error 2 

I'm not sure how to get more verbose information out of the compilation, so I can't give you any more than that.

[samueljohn]

I can build julia fine with the same homebrew version as @spott.

[nolta]

I'm not sure what changed in my environment since it worked before:

Is this a fresh clone + build? If not, can you try git pull && make cleanall test?

And if that doesn't fix it, what's the output of:

$ cd base
$ gdb ../julia
...
(gdb) r -b stage0.jl
...
(gdb) bt

I gather you're trying this on a new retina macbook pro?

[spott]

This was a brand new git clone ### and make -j 7 (later replicated with just a regular make, though not completely rebuilt, should I do that?)

(gdb) r -b stage0.jl
Starting program: /Users/spott/Documents/Code/julia/julia -b stage0.jl
Reading symbols for shared libraries +++.+................................................................................................. done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000000
0x0000000100058fb8 in jl_method_table_assoc_exact ()
(gdb) bt
# 0  0x0000000100058fb8 in jl_method_table_assoc_exact ()
# 1  0x0000000100058e9c in jl_apply_generic ()
# 2  0x0000000100061096 in jl_show ()
# 3  0x00000001000014a7 in exec_program ()
# 4  0x0000000100001721 in true_main ()
# 5  0x00000001000861ad in julia_trampoline ()
# 6  0x0000000100001ded in main ()

Let me know if you need anything else.

(these are short code snippets, if I should start putting them in gists instead of pasting them here, let me know.)

[pao: avoid autolinking issues 1-6]

[nolta]

What's the output of git rev-parse HEAD?

And is this an ivy bridge mac?

Can you try the following:

$ make debug
$ cd base
$ gdb ../usr/bin/julia-debug-basic
...
(gdb) b jl_raise
(gdb) r -b stage0.jl
...
(gdb) bt

[samueljohn]

He has an ivy bridge mac. Retina MBP. Lucky one :-)

[spott]

Yes, I am running an IvyBridge retina MacBook Pro. :)

git rev-parse HEAD:
3f0d689fa48689ddeaf7498e85f1035c0b0ff9ee

make debug dies with:

    LINK usr/bin/julia-debug-readline
    LINK usr/bin/julia-debug-webserver
    JULIA usr/lib/julia/sys0.ji
/bin/sh: line 1: 86201 Segmentation fault: 11  /Users/spott/Documents/Code/julia/usr/bin/julia-debug-readline -b stage0.jl
make[1]: *** [/Users/spott/Documents/Code/julia/usr/lib/julia/sys0.ji] Error 139
make: *** [debug] Error 2

(gdb) b jl_raise
Breakpoint 1 at 0x4189374b18ea2c: file task.c, line 521.
(gdb) r -b stage0.jl
Starting program: /Users/spott/Documents/Code/julia/usr/bin/julia-debug-basic -b stage0.jl
Reading symbols for shared libraries ++++................................................................................................. done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done
Reading symbols for shared libraries . done

Breakpoint 1, jl_raise (e=0x10c5927b0) at task.c:521
521     jl_task_t *eh = jl_current_task->state.eh_task;
(gdb) bt
#0  jl_raise (e=0x10c5927b0) at task.c:521
#1  0x000000010001bb82 in jl_errorf (fmt=0x1008d0467 "could not load module %s: %s") at builtins.c:48
#2  0x000000010007a2b6 in jl_load_dynamic_library (modname=0x10c5ad9b8 "libm") at dlload.c:105
#3  0x0000000103111246 in ?? ()
#4  0x0000000100014624 in jl_apply (f=0x102d331a0, args=0x7fff5fbfdaa8, nargs=1) at julia.h:846
#5  0x0000000100017460 in jl_apply_generic (F=0x10294b9a0, args=0x7fff5fbfdaa8, nargs=1) at gf.c:1241
#6  0x0000000100074974 in jl_apply (f=0x10294b9a0, args=0x7fff5fbfdaa8, nargs=1) at julia.h:846
#7  0x00000001000748e4 in do_call (f=0x10294b9a0, args=0x10c59dc38, nargs=1, locals=0x0, nl=0) at interpreter.c:53
#8  0x0000000100074d7e in eval (e=0x10c5a7c40, locals=0x0, nl=0) at interpreter.c:114
#9  0x0000000100074eaa in eval (e=0x10c5a7c20, locals=0x0, nl=0) at interpreter.c:125
#10 0x0000000100074458 in jl_interpret_toplevel_expr (e=0x10c5a7c20) at interpreter.c:16
#11 0x0000000100087e0a in jl_toplevel_eval_flex (e=0x10c5a7be0, fast=1, plineno=0x7fff5fbfe444) at toplevel.c:240
#12 0x000000010008812f in jl_parse_eval_all (fname=0x10c59daf0 "math_libm.jl") at toplevel.c:286
#13 0x0000000100088394 in jl_load (fname=0x10c59daf0 "math_libm.jl") at toplevel.c:325
#14 0x00000001000883e0 in jl_load_ (str=0x10c592770) at toplevel.c:332
#15 0x00000001031014ac in ?? ()
#16 0x0000000100014624 in jl_apply (f=0x102905e60, args=0x7fff5fbfe5d8, nargs=1) at julia.h:846
#17 0x0000000100017460 in jl_apply_generic (F=0x1028cfe00, args=0x7fff5fbfe5d8, nargs=1) at gf.c:1241
#18 0x0000000100074974 in jl_apply (f=0x1028cfe00, args=0x7fff5fbfe5d8, nargs=1) at julia.h:846
#19 0x00000001000748e4 in do_call (f=0x1028cfe00, args=0x10c59db38, nargs=1, locals=0x0, nl=0) at interpreter.c:53
#20 0x0000000100074d7e in eval (e=0x10c5a7bc0, locals=0x0, nl=0) at interpreter.c:114
#21 0x0000000100074458 in jl_interpret_toplevel_expr (e=0x10c5a7bc0) at interpreter.c:16
#22 0x0000000100087e0a in jl_toplevel_eval_flex (e=0x10c5a7ba0, fast=1, plineno=0x7fff5fbfec24) at toplevel.c:240
#23 0x000000010008812f in jl_parse_eval_all (fname=0x10282ad70 "sysimg.jl") at toplevel.c:286
---Type <return> to continue, or q <return> to quit---
#24 0x0000000100088394 in jl_load (fname=0x10282ad70 "sysimg.jl") at toplevel.c:325
#25 0x00000001000883e0 in jl_load_ (str=0x1028fba60) at toplevel.c:332
#26 0x00000001031014ac in ?? ()
#27 0x000000010001cd04 in jl_apply (f=0x102905e60, args=0x7fff5fbfee68, nargs=1) at julia.h:846
#28 0x000000010001ebd3 in jl_trampoline (F=0x102905e60, args=0x7fff5fbfee68, nargs=1) at builtins.c:659
#29 0x0000000100014624 in jl_apply (f=0x102905e60, args=0x7fff5fbfee68, nargs=1) at julia.h:846
#30 0x0000000100017460 in jl_apply_generic (F=0x1028cfe00, args=0x7fff5fbfee68, nargs=1) at gf.c:1241
#31 0x0000000100074974 in jl_apply (f=0x1028cfe00, args=0x7fff5fbfee68, nargs=1) at julia.h:846
#32 0x00000001000748e4 in do_call (f=0x1028cfe00, args=0x10290df78, nargs=1, locals=0x0, nl=0) at interpreter.c:53
#33 0x0000000100074d7e in eval (e=0x102905e40, locals=0x0, nl=0) at interpreter.c:114
#34 0x0000000100074458 in jl_interpret_toplevel_expr (e=0x102905e40) at interpreter.c:16
#35 0x0000000100087e0a in jl_toplevel_eval_flex (e=0x1029044c0, fast=1, plineno=0x7fff5fbff704) at toplevel.c:240
#36 0x0000000100086fbf in jl_eval_module_expr (ex=0x1028d0040, plineno=0x7fff5fbff704) at toplevel.c:64
#37 0x00000001000879a6 in jl_toplevel_eval_flex (e=0x1028d0040, fast=1, plineno=0x7fff5fbff704) at toplevel.c:182
#38 0x000000010008812f in jl_parse_eval_all (fname=0x7fff5fbffbaf "stage0.jl") at toplevel.c:286
#39 0x0000000100088394 in jl_load (fname=0x7fff5fbffbaf "stage0.jl") at toplevel.c:325
#40 0x0000000100001ccf in exec_program () at repl.c:137
#41 0x00000001000020c2 in true_main (argc=1, argv=0x7fff5fbffa20) at repl.c:195
#42 0x000000010007c77a in julia_trampoline (argc=1, argv=0x7fff5fbffa20, pmain=0x100001f90 <true_main>) at init.c:231
#43 0x00000001000023bf in main (argc=1, argv=0x7fff5fbffa20) at repl.c:248

[nolta]

On your laptop:

CLT: 4.3.0.0.1.1249367152
GCC-4.0: N/A
GCC-4.2: N/A

but on my laptop:

CLT: 1.0.0.9000000000.1.1249367152
GCC-4.0: N/A
GCC-4.2: build 5666

What's the output of gcc --version and gcc-4.2 --version?

[spott]

$gcc --version
i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.9.00)
Copyright (C) 2007 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

I don't have a gcc-4.2 I wonder why... I also wonder why I have an older version of gcc...

[nolta]

And just to be sure, what's the output of file /usr/lib/libm.dylib?

[spott]

$file /usr/lib/libm.dylib
/usr/lib/libm.dylib: Mach-O universal binary with 2 architectures
/usr/lib/libm.dylib (for architecture x86_64):  Mach-O 64-bit dynamically linked shared library x86_64
/usr/lib/libm.dylib (for architecture i386):    Mach-O dynamically linked shared library i386

[samueljohn]

I have the same gcc --version as @spott and I can compile julia. This is the llvm-gcc which is symlinked as gcc.

i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.9.00)
Copyright (C) 2007 Free Software Foundation, Inc.

This one comes with Xcode.

[spott]

@samueljohn: do you also not have a separate gcc-4.2?

Also, does your brew --config have a "N/A" after gcc-4.2?

[samueljohn]

Lion and current Xcode do not ship gcc. Only the llvm-gcc symlink.
Apple does not like the new GPL.

I can try the build (tomorrow) on a core 2 duo Mac with only the
command line tools (there I have no Xcode).

[samueljohn]

My (working) brew --config output. Julia compiles just fine with this.

HOMEBREW_VERSION: 0.9.2
HEAD: 03744647cc9271abc8c2cb368702fc328ee9a0ec
HOMEBREW_PREFIX: /usr/local
HOMEBREW_CELLAR: /usr/local/Cellar
CPU: 8-core 64-bit sandybridge
OS X: 10.7.4-x86_64
Xcode: 4.3.3
CLT: N/A
GCC-4.0: N/A
GCC-4.2: N/A
LLVM: build 2336
Clang: 3.1 build 318
MacPorts or Fink? false
X11: 2.7.2 @ /opt/X11
System Ruby: 1.8.7-357
Which Perl:   /usr/bin/perl
Which Python: /usr/bin/python
Which Ruby:   /usr/bin/ruby => /System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/bin/ruby

[staticfloat]

My (working) brew --config is very similar to @samueljohn's:

HOMEBREW_VERSION: 0.9.2
HEAD: 8f4469c781b41fb016e5f186d68661a66ffc0b3a
HOMEBREW_PREFIX: /Users/sabae/.homebrew
HOMEBREW_CELLAR: /Users/sabae/.homebrew/Cellar
CPU: dual-core 64-bit penryn
OS X: 10.7.4-x86_64
Xcode: 4.3.1
CLT: 1.0.0.9000000000.1.1249367152
GCC-4.0: N/A
GCC-4.2: build 5666
LLVM: build 2336
Clang: 3.1 build 318
MacPorts or Fink? false
X11: 2.6.4 @ /usr/X11
System Ruby: 1.8.7-357
Which Perl:   /usr/bin/perl
Which Python: /Users/sabae/.homebrew/bin/python => /Users/sabae/.homebrew/Cellar/python/2.7.3/Frameworks/Python.framework/Versions/2.7/bin/python2.7
Which Ruby:   /usr/bin/ruby => /System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/bin/ruby

[samueljohn]

Right now my only suspect is ivybridge.

[spott]

Aren't the different platforms supposed to be backward compatible?

On Jul 17, 2012, at 11:56 PM, Samuel John wrote:

Right now my only suspect is ivybridge.

---

Reply to this email directly or view it on GitHub:
#1052 (comment)

[samueljohn]

They should be. But on the other hand it would not be the first time for such an error to be cpu dependent.
@spott have you run brew doctor? Can you think of anything non-standard on your system's software?

[samueljohn]

@spott brew update and brew outdated just to assure that none of the dependencies is kinda too old?

[spott]

@samueljohn brew doctor only has one error, an unlinked keg that I manually installed (and doesn't need to be linked), brew update and brew outdated don't do anything, everything is up to date.

I unlinked all homebrew software not in the dependencies for julia, and I still get this error.

[spott]

SOLVED!!:

I had an old bashrc thing that appended to DYLD_FALLBACK_LIBRARY_PATH, and that was causing the problems.

[samueljohn]

Wow, cool!
I guess I'll report to he homebrew team to let the doctor check, if that variable is set!

[spott]

That seems like a decent idea, though no-one in their right mind should have it set... It was a weird case with an old library I was attempting to use. I never actually got it working, but forgot it was in my bashrc.

[samueljohn]

Homebrew/legacy-homebrew#13463