This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: tree-ssa: a weird (mis)optimization
- From: Geert Bosch <bosch at adacore dot com>
- To: Richard Henderson <rth at redhat dot com>
- Cc: gcc at gcc dot gnu dot org, Andrew Haley <aph at redhat dot com>
- Date: Wed, 1 Dec 2004 12:35:51 -0500
- Subject: Re: tree-ssa: a weird (mis)optimization
- References: <16812.39590.439732.460777@cuddles.cambridge.redhat.com> <08C7A98A-42FB-11D9-91E1-000A959A128E@adacore.com> <16812.47260.320238.536151@cuddles.cambridge.redhat.com> <20041130224704.GA4960@redhat.com> <16813.39257.780795.560709@cuddles.cambridge.redhat.com> <20041201171934.GB12839@redhat.com>
On Dec 1, 2004, at 12:19, Richard Henderson wrote:
I still think that the Java front end should be doing something in
order to indicate that these loads should not be removed, beyond
just marking them as trapping -- or rather, not marking them as not
trapping. Otherwise -fnon-call-exceptions will suddenly generate
*significantly* worse code for everyone else.
I strongly agree with this statement. Also, it shouldn't be too hard
to do this is a non-intrusive way. One possibility would be to have
a volatile nop construct that is a use of the result of the value
loaded, but does not generate any code by itself. I imagine we'd
already have a construct like that for supporting inline asm.
-Geert