On Thu, Jul 03, 2025 at 08:49:16PM +0000, Ruslan Bukin wrote: > The branch main has been updated by br: > > URL: > https://cgit.FreeBSD.org/src/commit/?id=906d7a4b521c19b2b1ae3ec844b5d4626f2fd529 > > commit 906d7a4b521c19b2b1ae3ec844b5d4626f2fd529 > Author: Ruslan Bukin <b...@freebsd.org> > AuthorDate: 2025-07-03 20:43:01 +0000 > Commit: Ruslan Bukin <b...@freebsd.org> > CommitDate: 2025-07-03 20:43:01 +0000 > > hwt: fix !SMP build. > > Sponsored by: UKRI > --- > sys/dev/hwt/hwt_ioctl.c | 2 ++ > sys/dev/hwt/hwt_vm.c | 2 ++ > 2 files changed, 4 insertions(+) > > diff --git a/sys/dev/hwt/hwt_ioctl.c b/sys/dev/hwt/hwt_ioctl.c > index 04133838d292..592db4931bb4 100644 > --- a/sys/dev/hwt/hwt_ioctl.c > +++ b/sys/dev/hwt/hwt_ioctl.c > @@ -305,9 +305,11 @@ hwt_ioctl_alloc_mode_cpu(struct thread *td, struct > hwt_owner *ho, > return (error); > > CPU_FOREACH_ISSET(cpu_id, &cpu_map) { > +#ifdef SMP > /* Ensure CPU is not halted. */ > if (CPU_ISSET(cpu_id, &hlt_cpus_mask)) > return (ENXIO); > +#endif > #if 0 > /* TODO: Check if the owner have this cpu configured already. */ > ctx = hwt_owner_lookup_ctx_by_cpu(ho, halloc->cpu); > diff --git a/sys/dev/hwt/hwt_vm.c b/sys/dev/hwt/hwt_vm.c > index a6799c30300d..6c55e218dcec 100644 > --- a/sys/dev/hwt/hwt_vm.c > +++ b/sys/dev/hwt/hwt_vm.c > @@ -213,9 +213,11 @@ hwt_vm_start_cpu_mode(struct hwt_context *ctx) > CPU_ZERO(&enable_cpus); > > CPU_FOREACH_ISSET(cpu_id, &ctx->cpu_map) { > +#ifdef SMP > /* Ensure CPU is not halted. */ > if (CPU_ISSET(cpu_id, &hlt_cpus_mask)) > continue; > +#endif > > hwt_backend_configure(ctx, cpu_id, cpu_id); >
This again raises a wonder: should we drop SMP option at all?
