On Thu, Dec 07, 2023 at 11:12:39AM +0100, Richard Biener wrote:
> > 2023-12-07 Jakub Jelinek <ja...@redhat.com>
> >
> > PR middle-end/112411
> > * params.opt (-param=min-nondebug-insn-uid=): Add
> > IntegerRange(0, 1073741824).
> > * lra.cc (check_and_expand_insn_recog_data): Use 3U rather than 3
> > in * 3 / 2 computation and if the result is smaller or equal to
> > index, use index + 1.
> >
> > * gcc.dg/params/blocksort-part.c: Add dg-skip-if for
> > --param min-nondebug-insn-uid=1073741824.
>
> what's this change for? Do we test the actual param limit? Can you
> skip for the param without specifying the actual upper bound?
params.exp iterates over all params which have a range selected and tries
to compile the testcase(s) with both the minimum and if any maximum of the
range.
I think it is useful to test normally with --param min-nondebug-insn-uid=0
the minimum, that means it is off, it is just the maximum which either
doesn't work or requires those hundreds of gigabytes of memory (guess I
should look at what needs that much).
I don't know how else to skip just the maximum test for the param except
to specify the exact value; if params.opt changes that value, people will
notice FAILs of the test and the test can be adjusted too (unless the
maximum is lowered into something so small that it works well even on low
memory 32-bit hosts).
Jakub
