Messages in this thread | | | From | Linus Torvalds <> | Date | Tue, 18 Sep 2012 13:27:50 -0700 | Subject | Re: [PATCH 3.6-rc6] cpufreq/powernow-k8: workqueue user shouldn't migrate the kworker to another CPU |
| |
On Tue, Sep 18, 2012 at 1:12 PM, Tejun Heo <tj@kernel.org> wrote: > So, with work_on_cpu() reimplementation just posted[1], we can do the > following instead. Functionally it's about the same but less ugly. > Ugly as it may be, I think the previous open coded version is better > suited as a fix and for -stable. Thoughts?
I have to say, since the work_on_cpu() reimplementation seems to seriously simplify the code, and removes more lines than it adds, and makes this patch smaller than your original patch, I would personally prefer this approach instead anyway.
It's what we want long-range, isn't it? And it's smaller and simpler. Sure, it might be a *conceptually* bigger change, but since it's both prettier and *practically* smaller, I do like it more. Even at this stage of -rc (and even for backporting to -stable).
Can we get some quick testing of this two-patch series from the people who saw the original K8 cpufreq issue? Duncan?
Linus
| |