This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [PATCH 4/16] LTTng 0.6.36 for 2.6.18 : atomic UP operations on SMP
- From: Christoph Hellwig <hch at infradead dot org>
- To: Mathieu Desnoyers <mathieu dot desnoyers at polymtl dot ca>
- Cc: linux-kernel at vger dot kernel dot org, Christoph Hellwig <hch at infradead dot org>, Andrew Morton <akpm at osdl dot org>, Ingo Molnar <mingo at redhat dot com>, Greg Kroah-Hartman <gregkh at suse dot de>, Thomas Gleixner <tglx at linutronix dot de>, Tom Zanussi <zanussi at us dot ibm dot com>, Karim Yaghmour <karim at opersys dot com>, Paul Mundt <lethal at linux-sh dot org>, Jes Sorensen <jes at sgi dot com>, Richard J Moore <richardj_moore at uk dot ibm dot com>, "Martin J. Bligh" <mbligh at mbligh dot org>, Michel Dagenais <michel dot dagenais at polymtl dot ca>, Douglas Niehaus <niehaus at eecs dot ku dot edu>, ltt-dev at shafik dot org, systemtap at sources dot redhat dot com
- Date: Mon, 27 Nov 2006 16:56:43 +0000
- Subject: Re: [PATCH 4/16] LTTng 0.6.36 for 2.6.18 : atomic UP operations on SMP
- References: <20061124215518.GE25048@Krystal>
On Fri, Nov 24, 2006 at 04:55:18PM -0500, Mathieu Desnoyers wrote:
> This patch adds a UP flavor of SMP operations which is intended to provide
> atomic modification of per-cpu data without suffering from the LOCK of memory
> barrier performance cost. Note that extreme care must be taken when accessing
> this data from different CPUs : smp_wmb() and smp_rmb() must be used
> explicitely. As this last scenario happens very rarely in LTTng, it provides a
We already have local_t in asm/local.h for this purposed. Unfortunately
several architecture implementations are rather suboptimal, but I'm sure
the architecture maintainers would be interested in patches to optimize
the various implementations.