This is the mail archive of the
mailing list for the glibc project.
Re: [RFC][PATCH 0/2] Port of lock elision on z/Architecture
- From: Dominik Vogt <vogt at linux dot vnet dot ibm dot com>
- To: libc-alpha at sourceware dot org
- Date: Tue, 6 Aug 2013 13:15:39 +0200
- Subject: Re: [RFC][PATCH 0/2] Port of lock elision on z/Architecture
- References: <20130806090627 dot GA20664 at linux dot vnet dot ibm dot com> <20130806102208 dot GA6777 at domone dot kolej dot mff dot cuni dot cz>
- Reply-to: libc-alpha at sourceware dot org
On Tue, Aug 06, 2013 at 12:22:08PM +0200, OndÅej BÃlka wrote:
> On Tue, Aug 06, 2013 at 11:06:27AM +0200, Dominik Vogt wrote:
> > The following set of patches ports the implementation of the Intel
> > lock elision patches (without major changes to the original
> > logic). This is only a request for comments, please do not commit
> > any part of the patches (until my colleague who takes care of the
> > legal stuff is back from his holidays).
> Are there also other architectures that want to add lock elision?
> >From my perspective I would like not duplicate much code, at least for
> user visible interface handling. Could be these bits be factored out?
I've tried to stick to the Intel patches as closely as possible,
and - yes - I think some of the infrastructure can and should be
factored out. At the moment it's difficult for me to decide
_which_ parts can be extracted as I don't know how transactional
memory works on other platforms, e.g. Power.
> Also a second patch contains several formatting fixes for first patch,
> do not be surprised when somebody will point that issues in first patch.
Ups, that looks like I merged some changes that into the wrong
patch. I'll post updated version of the patches in a minute.
Dominik ^_^ ^_^