Re: [bug-mes] ARM mes on wip-arm - Other test failures

[Jan Nieuwenhuizen]

Danny Milosavljevic writes:

> Hi Janneke,
>
> On Tue, 12 Mar 2019 07:25:35 +0100
> Jan Nieuwenhuizen <address@hidden> wrote:
>
>> +if test $mes_libc = arm; then
>
> "= mes", right?

eh, I meant to say: $mes_cpu = arm, not sure
about restricting to $mes_libc = mes, WDYT?

>>  (define (remainder x y)
>> -  (- x (* (quotient x y) y)))
>> +  (abs (remainder n radix)))
>
> I thought about it but then I saw:
>
> $ guile
> GNU Guile 2.2.4
> [...]
> scheme@(guile-user)> (remainder -5 2)
> $1 = -1

Ah, right!

>> > FAIL: lib/tests/io/90-stat.c [no case for __arm__]
>
> This one is strange.  I've added a case for __arm__ and then:
> (1) When I "rm -rf gcc-lib" and then do "make check", it still fails
> (2) But when I then invoke "lib/tests/io/90-stat" manually it succeeds (!)

Do yo mean gcc-lib/lib/tests/io/90-stat ?

> It might be that the test timeout is too low - the division algorithm
> isn't exactly optimized ;)

Yeah, you can bump the timout in build-aux/test-c.sh.  Does the
gcc-lib/lib/tests/io/90-stat.log file say something interesting?

You can also try the somewhat experimental `make check recheck=true', if
that works :-)

janneke

-- 
Jan Nieuwenhuizen <address@hidden> | GNU LilyPond http://lilypond.org
Freelance IT http://JoyofSource.com | Avatar® http://AvatarAcademy.com