multi sub with where clause + temp crashing on MoarVM

[p6rt]

Migrated from rt.perl.org#123686 (status was 'resolved')

Searchable as RT123686$

[p6rt]

From @dwarring

Consider​:

  my $x;
  multi sub foo($n where True) { temp $x; }
  foo($_) for 1 ... 1000; # crashing here on MoarVM
  say 'ok';

This runs to completion on JVM and Parrot, but crashes on MoarVM 2015.01-30-gd671638. Typically at around 100 iterations.

[Golfed from flapping CSS​::Writer tests]

[p6rt]

From @dwarring

A test case has been added to integration/weird-errors.t

Fudged for rakudo.moar
On Tue Jan 27 12​:03​:24 2015, david.warring wrote​:

Consider​:

my $x;
multi sub foo($n where True) { temp $x; }
foo($_) for 1 ... 1000; # crashing here on MoarVM
say 'ok';

This runs to completion on JVM and Parrot, but crashes on MoarVM
2015.01-30-gd671638. Typically at around 100 iterations.

[Golfed from flapping CSS​::Writer tests]

[p6rt]

From @usev6

The test passes now and was unfudged with commit Raku/roast@4954d43df4

I'm closing this ticket now. (whoever fixed this)++

[p6rt]

From @usev6

The test passes now and was unfudged with commit Raku/roast@4954d43df4

I'm closing this ticket now. (whoever fixed this)++

[p6rt]

@usev6 - Status changed from 'new' to 'resolved'

[p6rt]

From @dwarring

On Sun Feb 01 12​:21​:33 2015, bartolin@​gmx.de wrote​:

The test passes now and was unfudged with commit
Raku/roast@4954d43df4

I'm closing this ticket now. (whoever fixed this)++

This bug seems to be back :-| e.g.
https://github.com/coke/perl6-roast-data/blob/f327f9a3387549f96d160b2e23993be2518c4d51/log/rakudo.moar-jit_summary.out#L3147

Reverted Raku/roast@4954d43df4 to refudge.

I might try bisecting when I get a chance, to see when this was reintroduced.

[p6rt]

@dwarring - Status changed from 'resolved' to 'open'

[p6rt]

From @lizmat

On 01 Feb 2015, at 21​:21, Christian Bartolomaeus via RT <bugs-comment@​bugs6.perl.org> wrote​:

The test passes now and was unfudged with commit Raku/roast@4954d43df4

I'm closing this ticket now. (whoever fixed this)++

Seems #​124318 is related to this.

Liz

[p6rt]

From @jnthn

On Thu Apr 16 13​:19​:13 2015, elizabeth wrote​:

On 01 Feb 2015, at 21​:21, Christian Bartolomaeus via RT <bugs-
comment@​bugs6.perl.org> wrote​:

The test passes now and was unfudged with commit
Raku/roast@4954d43df4

I'm closing this ticket now. (whoever fixed this)++

Seems #​124318 is related to this.

It was, though the dynamic optimizer was not involved, as #​124318 suggested; it just rearranged memory such that we saw it further. It was a bug involving reading bogus memory, so could easily have appeared to go away, then come back. Now it should be fixed for good. Test re-enabled.

[p6rt]

@jnthn - Status changed from 'open' to 'resolved'