Date: Tue, 29 Jun 2010 22:09:55 -0600 (MDT) From: "M. Warner Losh" <imp@bsdimp.com> To: xcllnt@mac.com Cc: arm@freebsd.org, yanefbsd@gmail.com, marcel@freebsd.org Subject: Re: svn commit: r209604 - head/lib/libc/gmon Message-ID: <20100629.220955.926703890692696643.imp@bsdimp.com> In-Reply-To: <F642907C-9545-4DD4-9519-DEE245B012F6@mac.com> References: <3E9AF18F-9404-49A8-A3A6-CB1F8CF06A80@mac.com> <20100629.210549.175946041797646353.imp@bsdimp.com> <F642907C-9545-4DD4-9519-DEE245B012F6@mac.com>
next in thread | previous in thread | raw e-mail | index | archive | help
In message: <F642907C-9545-4DD4-9519-DEE245B012F6@mac.com>
Marcel Moolenaar <xcllnt@mac.com> writes:
:
: On Jun 29, 2010, at 8:05 PM, M. Warner Losh wrote:
:
: > In message: <3E9AF18F-9404-49A8-A3A6-CB1F8CF06A80@mac.com>
: > Marcel Moolenaar <xcllnt@mac.com> writes:
: > :
: > : On Jun 29, 2010, at 7:37 PM, M. Warner Losh wrote:
: > :
: > : > In message: <14807AD0-2BC9-4FF1-9C2B-9E47FA2F9A41@mac.com>
: > : > Marcel Moolenaar <xcllnt@mac.com> writes:
: > : > :
: > : > : On Jun 29, 2010, at 6:48 PM, Garrett Cooper wrote:
: > : > : >> Modified: head/lib/libc/gmon/gmon.c
: > : > : >> ==============================================================================
: > : > : >> --- head/lib/libc/gmon/gmon.c Wed Jun 30 01:10:08 2010 (r209603)
: > : > : >> +++ head/lib/libc/gmon/gmon.c Wed Jun 30 01:40:25 2010 (r209604)
: > : > : >> @@ -111,7 +111,7 @@ monstartup(lowpc, highpc)
: > : > : >>
: > : > : >> o = p->highpc - p->lowpc;
: > : > : >> if (p->kcountsize < o) {
: > : > : >> -#ifndef hp300
: > : > : >> +#if !defined(__powerpc__)
: > : > : >> s_scale = ((float)p->kcountsize / o ) * SCALE_1_TO_1;
: > : > : >> #else /* avoid floating point */
: > : > : >> int quot = o / p->kcountsize;
: > : > : >
: > : > : > Hi Marcel and Warner,
: > : > : > Should this perhaps be properly expanded to other architectures,
: > : > : > like arm and mips? I'm not sure if we have a FP emulator on FreeBSD or
: > : > : > not like Cavium does on Linux.
: > : > :
: > : > : Yes, this can be expanded. I opted for a conservative commit to
: > : > : see what the general response to this commit would be.
: > : >
: > : > My response: "Why doesn't the soft floating point make this
: > : > unnecessary?" and "What about G4's and G5's that have floating point?"
: > :
: > : See the commit log: it's most likely a problem with the expression
: > : or with the compiler in dealing with the expression.
: >
: > That's a little vague.
: >
: > But it does suggest that arm and mips won't have to worry.
:
: No it doesn't. In fact, profiling on ARM shows the same problems
: that I just fixed on PowerPC. It's best that people validate MIPS
: before dismissing this as a non-issue.
Then I don't think this is just a simple compiler bug, but something
more fundamental. the softfloat stuff should make this unnecessary.
Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100629.220955.926703890692696643.imp>