[M3devel] ppc_linux hanging

[Jay K]

I'm afraid I just don't know.

Now I'm seeing:
Fatal Error: bad version stamps: TextClass.i3

version stamp mismatch: M3_BUILTIN.TEXT
  <61b2d19ae7136110> => TextClass.i3
  <bbef933ad0c2139d> => M3_BUILTIN.i3  
 *** execution of [<function _BuildGlobalFunction at 0x4820bbf0>, <function _Shi

and, really, you know, since I don't want it to be true and I don't want to debug it, I tried twice,
and we know I generally use the right boostrap order, and these things aren't changing either
so it shouldn't be very order dependent.


I wonder also about the change I made to use -build/host/target in native builds.
Maybe I should try without that.


So I'm going to try volatile again. Alas.
 If it helps, then I might try only volatile stores. It seems to me that volatile on store
  should have enough affect "no matter" why volatile helps.


Yes, it was a hang.
I can't reproduce it now. Very odd, since I had seen it a few times.
If I can, maybe I can debug, now that I "solved" one debugging problem (use gdb 7.1 or turn off -pie).


 - Jay

----------------------------------------
> From: hosking at cs.purdue.edu
> Date: Sat, 3 Jul 2010 20:29:00 -0400
> To: jay.krell at cornell.edu
> CC: m3devel at elegosoft.com
> Subject: Re: [M3devel] ppc_linux hanging
>
> Is there something about the memory model on PPC? It manifests as a hang, correct?
>
> On 3 Jul 2010, at 06:37, Jay K wrote:
>
> >
> > Alas, good news, bad news, restoring volatile on every load/store seems to fix this.
> > I wasn't even optimizing.
> >
> > Tony any ideas?
> >
> > I386_LINUX is ok. AMD64_DARWIN is ok.
> > SOLgnu is ok, but that doesn't count. I left volatile on there, was worried for some reason about interaction with stack walker, I forget.
> >
> > So far the debugger is being very uncooperative, can't set any breakpoints..it works for C.
> >
> > Aha..
> >
> > gcc -gstabs+ -fPIC -pie 2.c
> > (gdb) break main
> > Breakpoint 1 at 0x658: file 2.c, line 3.
> > (gdb) run
> > Starting program: /home/jay/a.out
> > Warning:
> > Cannot insert breakpoint 1.
> > Error accessing memory address 0x658: Input/output error.
> >
> >
> > (gdb) q
> > The program is running. Exit anyway? (y or n) y
> > jay at plin:~$ gcc -gstabs+ -fPIC 2.c
> > jay at plin:~$ gdb ./a.out
> > GNU gdb 6.8-debian
> > Copyright (C) 2008 Free Software Foundation, Inc.
> > License GPLv3+: GNU GPL version 3 or later 
> > This is free software: you are free to change and redistribute it.
> > There is NO WARRANTY, to the extent permitted by law. Type "show copying"
> > and "show warranty" for details.
> > This GDB was configured as "powerpc-linux-gnu"...
> > (gdb) break main
> > Breakpoint 1 at 0x10000448: file 2.c, line 3.
> > (gdb) run
> > Starting program: /home/jay/a.out
> >
> > Breakpoint 1, main () at 2.c:3
> > 3 }
> >
> >
> > So lame..
> >
> > - Jay
> >
> > ----------------------------------------
> >> From: jay.krell at cornell.edu
> >> To: m3devel at elegosoft.com
> >> Subject: ppc_linux hanging
> >> Date: Fri, 2 Jul 2010 08:10:10 +0000
> >>
> >>
> >> Alas, PPC_LINUX is very prone to hang on head.
> >> e.g. compiling m3front.
> >> It doesn't hang with @M3nogc.
> >> The stack at the time just shows junk, no symbols at all.
> >> and info threads shows nothing, maybe I mistyped it.
> >>
> >>
> >> I'll try to figure it out.
> >> The biggest recent change I think is removing volatile so I'll start
> >> by trying with that put back.
> >>
> >>
> >> Also I use Darwin more lately, which has its own thread suspension code.
> >> Could be something broke the others.
> >>
> >>
> >> Later,
> >> - Jay
> >>
> >>
> >
>