ACATS Random failure under GNAT GCC

ACATS Random failure under GNAT GCC

[Anh Vo]

I have noticed that ACATS randomly fails in gcc 4.0.x and 4.1.x. under
RH 8.0 and 9.0. Normally, it does happen when it is built and tested
the firtst time. After that if ACATS is tested again consecutively, for
example with three consecutive commands make -C gcc check-ada; make -C
gcc check; make -C gcc check-ada, a small number of ACATS failures show
up. So far, for gcc-4.0.x the tests below have failed.

FAIL:   c94001c
FAIL:   c954025
FAIL:   cd7101f
FAIL:   cd7103d
FAIL:   cd7204b
FAIL:   ce3804b

For gcc-4.1.x there are two ACATS tests have failed.

FAIL:   ce2401l
FAIL:   ce3906f

Has anyone expericenced with this situation or just me? I would like
hear your suggestion before reporting to gcc.  

AV

Re: ACATS Random failure under GNAT GCC

[Georg Bauhaus]

Anh Vo wrote:
> I have noticed that ACATS randomly fails in gcc 4.0.x and 4.1.x. under
> RH 8.0 and 9.0. 

> Has anyone expericenced with this situation or just me? I would like
> hear your suggestion before reporting to gcc.  

I vaguely remember recent discussions of possibly related issues
on the GCC list. (Like: which compiler is actually used at what
stage, should GCC be using static libs for checking Ada on PPC,
and when, etc.) Maybe you want to ask the GCC list?


-- Georg 

Re: ACATS Random failure under GNAT GCC

[Randy Brukardt]

"Anh Vo" <anhvofrcaus@gmail.com> wrote in message
news:1139766766.085134.134180@f14g2000cwb.googlegroups.com...
> I have noticed that ACATS randomly fails in gcc 4.0.x and 4.1.x. under
> RH 8.0 and 9.0. Normally, it does happen when it is built and tested
> the firtst time. After that if ACATS is tested again consecutively, for
> example with three consecutive commands make -C gcc check-ada; make -C
> gcc check; make -C gcc check-ada, a small number of ACATS failures show
> up. So far, for gcc-4.0.x the tests below have failed.
>
> FAIL:   c94001c
> FAIL:   c954025
> FAIL:   cd7101f
> FAIL:   cd7103d
> FAIL:   cd7204b
> FAIL:   ce3804b
>
> For gcc-4.1.x there are two ACATS tests have failed.
>
> FAIL:   ce2401l
> FAIL:   ce3906f
>
> Has anyone expericenced with this situation or just me? I would like
> hear your suggestion before reporting to gcc.

Well, as someone experienced in running ACATS tests (but not with GNAT), the
list of tests suggests that there really are bugs here. The first two you
listed are tasking tests that might have timing errors (and thus there might
not be a real problem), but the rest are I/O tests and tests of the contents
of System. I can't think off-hand of how they could fail unless there is a
bug somewhere.

                                   Randy.

Re: ACATS Random failure under GNAT GCC

[Waldek Hebisch]

Anh Vo <anhvofrcaus@gmail.com> wrote:
> I have noticed that ACATS randomly fails in gcc 4.0.x and 4.1.x. under
> RH 8.0 and 9.0. Normally, it does happen when it is built and tested
> the firtst time. After that if ACATS is tested again consecutively, for
> example with three consecutive commands make -C gcc check-ada; make -C
> gcc check; make -C gcc check-ada, a small number of ACATS failures show
> up. So far, for gcc-4.0.x the tests below have failed.
> 
> FAIL:   c94001c
> FAIL:   c954025
> FAIL:   cd7101f
> FAIL:   cd7103d
> FAIL:   cd7204b
> FAIL:   ce3804b
> 
> For gcc-4.1.x there are two ACATS tests have failed.
> 
> FAIL:   ce2401l
> FAIL:   ce3906f
> 
> Has anyone expericenced with this situation or just me? I would like
> hear your suggestion before reporting to gcc.  
> 

I have no experience with ACATS. But I got random errors in C testsuite.
AFAIK the reason was bug in test environment (expect): it was unable
to correctly capture program output.

-- 
                              Waldek Hebisch
hebisch@math.uni.wroc.pl 

Re: ACATS Random failure under GNAT GCC

[Anh Vo]

Thank you all for your helpful info. It is interesting that I could not
duplicate the ACATS random failure on gnatgcc-4.1.0-20060219
(Prerelease). In fact, I built it on three different machines with
three flavors of Linux, Red Hat 8.0 running on i686, Red Hat 9.0
running on i686 and Yellow Dog 4.0 running on PPC G5. I will manually
run the two tests that gnatgcc-4.1 snapshot, released prior to
gnatgcc-4.1.0-20060219, failed before.

AV