On Thursday, February 26th, 2026 at 7:17 PM, Jakub Jelinek <[email protected]> wrote:
> On Thu, Feb 26, 2026 at 10:26:07PM +0800, Yuao Ma wrote: > > On Thu, Feb 26, 2026 at 10:23 PM Tobias Burnus <[email protected]> wrote: > > > > > > Yuao Ma wrote: > > > > This patch makes OMP_WAIT_POLICY accept the pvsched > > > > option. Tests and documentation have also been added. > > > > […] > > > > +As an extension, if the value is @code{PVSCHED}, the runtime library > > > > will > > > > +employ a para-virtualized scheduling guided barrier synchronization > > > > mechanism, > > > > +though this feature has not yet been implemented. > > > > > > I prefer waiting for the actual feature implementation before adding > > > (and documenting) a stub implementation. > > > > > > In order to make it less likely that the patch gets lost, I added it to > > > https://gcc.gnu.org/wiki/SummerOfCode#:~:text=new%20value%2C%20pvsched > > > > > > > That works for me - thanks for the quick response! > > Also, I wonder if it is a good idea to add an extension to a standard OMP_* > env var value rather than adding it to a GOMP_* prefixed env var (existing > one or new). > > Jakub > @Yuao Thanks for the patch. I agree with Jakub. Instead of adding a new value for OMP_WAIT_POLICY, proposing a GOMP prefixed version makes more sense for now. @Jakub @Tobias How about simply calling it GOMP_WAIT_POLICY? Thanks, Himadri P. S. @Andrea has joined the project as a co-mentor. He will be helping us with the eBPF implementation of Phantom Tracker. Adding him to this thread.
