Re: sbcl build host breakage

Gary Byers <[email protected]> Thu, 8 Sep 2005 16:55:14 -0600 (MDT)
Newsgroups gmane.lisp.openmcl.bugs
Message-ID <[email protected]>
I can provoke it (usually) by

? (dotimes (i 20000) (ccl::ensure-svar (gensym)))

(Sometimes it takes more than 20000 tries.)

I think that it's a bug having to do with how hash tables
are handled by the GC.

There's a quick fix and a slow one.  I'll try to check
in the quick fix soon (so that we can get other things
done without worrying about Hash Table Bug du Jour), and
consider the more extensive fix later.

On Thu, 8 Sep 2005, bryan o'connor wrote:

>> (incf (nhash.count hash)) ; or some variant
>
> that worked.  i'm not seeing the second issue any more, so
> i guess it was related.
>
> but.. now i'm getting an unhandled exception 11 even later
> on in the build.
>
> Unhandled exception 11 at 0x4043790, context->regs at #xf01353f8
> Read operation to unmapped address 0x5d157e06
> While executing: #<Function ENSURE-SVAR-IDX #x08025f06>
> ? for help
> [1353] OpenMCL kernel debugger: :b
> [1353] OpenMCL kernel debugger:
>
> (#xF01358E0) #x00005E90 : (subprimitive _ret1valn)
> (#xF01358F0) #x04043A58 : #<Function %ENSURE-SVAR #x0802607e> + 308
> (#xF0135900) #x04043988 : #<Function %ENSURE-SVAR #x0802607e> + 100
> (#xF0135910) #x04042FC0 : #<Function %SYM-VALUE #x080259f6> + 44
> (#xF0135920) #x04042DCC : #<Function BOUNDP #x08025896> + 32
> (#xF0135930) #x08691D30 : #<Function PROCESS-TYPE-DECL #x0869207e> + 188
> (#xF0135940) #x0869A41C : #<Function PROCESS-1-DECL #x0869a5ee> + 728
> (#xF0135950) #x00005E90 : (subprimitive _ret1valn)
> (#xF0135960) #x08699E38 : #<Function PROCESS-DECLS #x08699ede> + 228
> (#xF0135970) #x00005E90 : (subprimitive _ret1valn)
> (#xF0135980) #x0869F6E8 : #<Function IR1-CONVERT-LAMBDA #x0869fefe> + 436
> (#xF0135990) #x08689FBC : #<Function MAKE-FUNCTIONAL-FROM-TOPLEVEL-LAMBDA 
> #x0868
> ...
>
> rcontext = 0x300F60
> nargs = 4
> r16 (fn) = #<Function ENSURE-SVAR-IDX #x08025f06>
> r23 (arg_z) = 390422402
> r22 (arg_y) = #<HASH-TABLE @#x0802bea6>
> r21 (arg_x) = (*INTERRUPT-LEVEL*)
> r20 (temp0) = #<1-element vector subtag = 2F @#x04000406 (CODE-VECTOR)>
> r19 (temp1/next_method_context) = 0
> r18 (temp2/nfn) = #<Function ENSURE-SVAR-IDX #x08025f06>
> r17 (temp3/fname) = #<1-element vector subtag = 8A @#x0802be9e ()>
> r31 (save0) = G1781
> r30 (save1) = ()
> r29 (save2) = ()
> r28 (save3) = 390422402
> r27 (save4) = MAKE-LEXENV
> r26 (save5) = (INDEX G1781)
> r25 (save6) = (#<(LAMBDA-VAR BASIC-VAR LEAF STRUCTURE!OBJECT) @#x096a9f96> 
> #<(LAMBDA-VAR BASIC-VAR LEAF STRUCTURE!OBJECT) @#x096a9f46> #<(LAMBDA-VAR 
> BASIC-VAR LEAF STRUCTURE!OBJECT) @#x096a9ef6>)
> r24 (save7) = STRUCTURE-TYPEP
>
> _______________________________________________
> Bug-openmcl mailing list
> [email protected]
> http://clozure.com/mailman/listinfo/bug-openmcl
>
>