[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-patches
Subject:    Re: [PATCH]: fix PR testsuite/12325, inf unavailable on vax
From:       Andrew Pinski <pinskia () physics ! uc ! edu>
Date:       2007-01-13 19:47:25
Message-ID: 200701131947.l0DJlPii013672 () localhost ! localdomain
[Download RAW message or body]

> 
> In this PR, we get an error because some platforms (in this
> case vax-netbsdelf) don't support inf and GCC blathers about it causing
> the testsuite to flag the testcase as having an error.
> 
> The line in question actually folds away at the tree level and we never
> need to actually generate any infs.  So we can just ignore the warning.
> 
> I copied the dg-warning line from another testcase gcc.dg/builtin-inf-1.c,
> so the logic should be correct.  I don't have any vax systems, but I
> tested it on sparc with "make check" to ensure there were no typos.
> 
> The PR was reported against 3.4, okay for all active branches?

Most likely, we should make two target_supports macros, one for float infs
and one for double infs support and then use them here.

SPU is an example of where float does not support inf but double does.

If I get some time, I am also going to clean up ieee.exp into mutliple
different directories: ieee/float, ieee/double, ieee/long_double, ieee/all.
This is because SPU's float is not ieee but its double is.


Thanks,
Andrew Pinski
[prev in list] [next in list] [prev in thread] [next in thread]