Re: [Gm2] Report on gm2-2010-11-07 on Solaris 10/sparc

[Gaius Mulley]

john o goyo <address@hidden> writes:

> Gmake check-gm2 on gm2-2010-11-07 Solaris 10/sparc reports the
> following.
>
>                 === gm2 Summary ===
>
> # of expected passes            9171
> # of unexpected failures        45
>
> (Full report to Gaius.)
>
> The extra failures are due to sigfig.mod but something is amiss in
> reporting.  The file gm2.sum contains the following.
>
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -g
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -O
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -O -g
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -Os
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -O3 -fomit-frame-
> pointer
> FAIL: gm2/isolib/run/pass/sigfig.mod execution,  -O3 -fomit-frame-
> pointer -finline-functions
>
> However, on their own, we have the following passes.
>
> johng: 591 [tst-gm2]=> gm2 -c -g sigfig.mod
> johng: 592 [tst-gm2]=> gm2 -g sigfig.mod
> johng: 593 [tst-gm2]=> ./a.out
>   passed   ToSigFig(12.3456789, 3) -> 12.3
>   passed   ToSigFig(12.3456789, 4) -> 12.35
>   passed   ToSigFig(12.3456789, 5) -> 12.346
>   passed   ToSigFig(12.3456789, 6) -> 12.3457
>   passed   ToSigFig(19.99, 3) -> 20.0
>   passed   ToSigFig(99.999, 3) -> 100
>   passed   ToSigFig(99.999, 3) -> 100
>
> The sigfig tests: passed
>
> Similarly, "-O", "-O -g", and "-Os" pass.
>
> The compilation option "-O3" causes failure regardless of what follows.
>
> johng: 603 [tst-gm2]=> gm2 -c -O3 -fomit-frame-pointer sigfig.mod
> johng: 604 [tst-gm2]=> gm2 -O3 -fomit-frame-pointer sigfig.mod
> johng: 605 [tst-gm2]=> ./a.out
> ../../../src/gcc-4.1.2/gcc/gm2/gm2-libs/StrLib.mod:117:3:array index
> out of bounds during dynamic ar
> ray access in implementation module M2EXCEPTION
> Abort (core dumped)
> johng: 606 [tst-gm2]=> gm2 -c -O3 sigfig.mod
> johng: 607 [tst-gm2]=> gm2 -O3 sigfig.mod
> johng: 608 [tst-gm2]=> ./a.out
> ../../../src/gcc-4.1.2/gcc/gm2/gm2-libs/StrLib.mod:117:3:array index
> out of bounds during dynamic ar
> ray access in implementation module M2EXCEPTION
> Abort (core dumped)
>
> john

Hi John,

thanks for the bug report - I've just fixed the bug exposed by sigfigs
now (a double free) - do you still get the bug exposed by StrLib ?

regards,
Gaius