This is the mail archive of the
mailing list for the glibc project.
Re: [PATCHv2 0/2] Tunables for glibc
- From: Torvald Riegel <triegel at redhat dot com>
- To: Andrew Pinski <pinskia at gmail dot com>
- Cc: Siddhesh Poyarekar <sid at reserved-bit dot com>, GNU C Library <libc-alpha at sourceware dot org>, Roland McGrath <roland at hack dot frob dot com>, "Carlos O'Donell" <carlos at redhat dot com>, Andi Kleen <andi at firstfloor dot org>, Tulio Magno Quites Machado Filho <tuliom at linux dot vnet dot ibm dot com>, "Paul E. Murphy" <murphyp at linux dot vnet dot ibm dot com>, munroesj at linux dot vnet dot ibm dot com
- Date: Thu, 12 May 2016 13:39:37 +0200
- Subject: Re: [PATCHv2 0/2] Tunables for glibc
- Authentication-results: sourceware.org; auth=none
- References: <20160116185503 dot GA17754 at devel dot intra dot reserved-bit dot com> <CA+=Sn1k63+=qNriuR5J5POX4KQBocqsYePZNXzrpf9aSQY4g3w at mail dot gmail dot com>
On Wed, 2016-05-11 at 19:21 -0700, Andrew Pinski wrote:
> I have a case where I want to have
> a tunable parameter for pthread_mutex_lock where we spin a little bit
> in userspace before calling futex as people use pthread_mutex's as
> normal locks and in the case of huge number of cores we should not
> call futex if the time spent inside the lock is small.
This (ie, spinning and back-off on all mutex types) is something I
planned to do for all archs. If you have experimented and found
spinning/back-off numbers that work well for a particular arch, please
let me know.
I'd also be happy to review any patches you might have.