This is the mail archive of the libc-alpha@sources.redhat.com mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Build failure of 2.2.3 and 2.2.4pre2 on sparc-unknown-linux


On Sun, 05 Aug 2001, Andreas Jaeger stated:
> FYI: with-elf, with-fp, enable-shared, disable-bounded are set by
> default, no need to give them - but no harm done either.

Crazy paranoia on my part :)

> Please do not use --enable-omitfp, especially if you report problems.

Pardon me, historical inertia from my IA32 days. I tested without
--enable-omitfp as well, and it still happened (but forgot to mention
that in the report, whoops).

In any case, omitting --enable-omitfp didn't help :(

(Especially on smaller IA32 boxes, not using --enable-omitfp is not
feasible; long the past, I benchmarked substantial performance losses,
up to 20% (!), from omitting that switch on those platforms. This *was*
a while ago and things may have changed, but I don't know that they
have.

Yes, this was intentionally pathological code that did a silly number of
calls to the libc, but still...

... oh, and btw, if problems with --enable-omitfp are never reported,
how will they be known about to be fixed? ;) I know, the reasoning is
that compiler bugs are likely with -fomit-frame-pointer; but they're not
all that much more likely; the likelihood is almost lost in the noise.
I can think of only one GCC release with a fubared -fomit-frame-pointer
on a common platform, and that's GCC-3.0.)

-- 
`It's all about bossing computers around. Users have to say "please".
Programmers get to say "do what I want NOW or the hard disk gets it".'
                        -- Richard Heathfield on the nature of programming


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]