This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [glibc] Fix Linux sysconf(_SC_NPROCESSORS_[CONF|ONLN]) performanceproblem
- From: Paul Eggert <eggert at cs dot ucla dot edu>
- To: Linus Torvalds <torvalds at linux-foundation dot org>
- Cc: libc-alpha at sourceware dot org
- Date: Thu, 30 Jun 2011 10:19:21 -0700
- Subject: Re: [glibc] Fix Linux sysconf(_SC_NPROCESSORS_[CONF|ONLN]) performanceproblem
- References: <BANLkTi=ousV=JtYzAYEnMVqqX6fnF_1z9Q@mail.gmail.com> <20110616175811.GF5583@const.bordeaux.inria.fr> <BANLkTinbLEcMr6KhUghFBwB3qcGv2UZ5MA@mail.gmail.com> <BANLkTikSMrkhd7zH_7yXPDSUy8YcHR1GWQ@mail.gmail.com> <BANLkTi=GPPpsPTfSYs9R=79tec+87+1vzg@mail.gmail.com> <4DFA61E7.30107@cs.ucla.edu>
On 06/16/11 11:41, Linus Torvalds wrote:
> I'd be happy to discuss it once the glibc implementation isn't crap.
I didn't see it mentioned on this list, so I thought I'd bring it
to your attention that Ulrich committed to glibc a rate-limiting
performance improvement along the lines that we discussed:
http://sourceware.org/git/?p=glibc.git;a=commitdiff;h=852eb34d5c56bc75bdd82327fcf310d98655f6b0
along with a patch to use /sys/devices/system/cpu/online if available:
http://sourceware.org/git/?p=glibc.git;a=commitdiff;h=84e2a551a72c79b020694bb327e33b6d71b09b63
It'd be nice to have something even faster, but at least now the glibc
implementation isn't, ahem, well, in the same category as
Brown Material Road. I've been there, and I don't want to go back:
http://en.wikipedia.org/wiki/File:CA_SR33_at_Brown_Material_Road_in_Kern_County.JPG