Closed citypw closed 10 years ago
Thanks for the PR!
I did a bit of research and I'm wondering if we should just replace preempt_enable_no_resched() with preempt_enable() unconditionally. From my understanding, the only difference is that preempt_enable() will process pending reschedules whereas preempt_enable_no_resched() will not. Looking at its usage, I don't see why this would be an issue. Other Linux kernel patches regarding preempt_enable_no_resched() seem to advocate against it as well.
I've run a quick test compiling and running the module with preempt_enable() on 64-bit Ubuntu 12.10 (3.5.0-17-generic) with no issue so far. If you don't have any thoughts against it, I'll make the commit.
hey Michael,
I think what you said above is right. I saw there was some cleanup fix for preempt_enable_no_resched() issues in "a0fa1dd3cdbccec9597fe53b6177a9aa6e20f2f8" yesterday, which is already merged into 3.14. So I did a test on kernel-3.14.1 on Debian after did a little bit of research. It's working fine so far.
So, plz make the commit! Thanks!
Support for linux kernel 3.14.1