| Message ID | 20241228184949.31582-14-yury.norov@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Delegated to: | Krzysztof Wilczyński |
| Headers | show |
| Series | cpumask: cleanup cpumask_next_wrap() implementation and usage | expand |
From: Yury Norov <yury.norov@gmail.com> Sent: Saturday, December 28, 2024 10:50 AM > > Calling cpumask_next_wrap_old() with starting CPU == nr_cpu_ids > is effectively the same as request to find first CPU, starting > from a given one and wrapping around if needed. > > cpumask_next_wrap() is a proper replacement for that. > > Signed-off-by: Yury Norov <yury.norov@gmail.com> > --- > drivers/pci/controller/pci-hyperv.c | 3 +-- > 1 file changed, 1 insertion(+), 2 deletions(-) > > diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c > index 86d1c2be8eb5..f8ebf98248b3 100644 > --- a/drivers/pci/controller/pci-hyperv.c > +++ b/drivers/pci/controller/pci-hyperv.c > @@ -1757,8 +1757,7 @@ static int hv_compose_multi_msi_req_get_cpu(void) > > spin_lock_irqsave(&multi_msi_cpu_lock, flags); > > - cpu_next = cpumask_next_wrap_old(cpu_next, cpu_online_mask, nr_cpu_ids, > - false); > + cpu_next = cpumask_next_wrap(cpu_next, cpu_online_mask); > cpu = cpu_next; > > spin_unlock_irqrestore(&multi_msi_cpu_lock, flags); > -- > 2.43.0 > I remember reviewing the patch that originally added this use of cpumask_next_wrap(). The two extra parameters were really hard to understand. Nice to see them go away! Reviewed-by: Michael Kelley <mhklinux@outlook.com>
On Sat, Dec 28, 2024 at 10:49:45AM -0800, Yury Norov wrote: > Calling cpumask_next_wrap_old() with starting CPU == nr_cpu_ids > is effectively the same as request to find first CPU, starting > from a given one and wrapping around if needed. > > cpumask_next_wrap() is a proper replacement for that. > > Signed-off-by: Yury Norov <yury.norov@gmail.com> s/switch/Switch/ in subject to match history. Since this depends on previous patches, I assume you'll merge them all together, so: Acked-by: Bjorn Helgaas <bhelgaas@google.com> > --- > drivers/pci/controller/pci-hyperv.c | 3 +-- > 1 file changed, 1 insertion(+), 2 deletions(-) > > diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c > index 86d1c2be8eb5..f8ebf98248b3 100644 > --- a/drivers/pci/controller/pci-hyperv.c > +++ b/drivers/pci/controller/pci-hyperv.c > @@ -1757,8 +1757,7 @@ static int hv_compose_multi_msi_req_get_cpu(void) > > spin_lock_irqsave(&multi_msi_cpu_lock, flags); > > - cpu_next = cpumask_next_wrap_old(cpu_next, cpu_online_mask, nr_cpu_ids, > - false); > + cpu_next = cpumask_next_wrap(cpu_next, cpu_online_mask); > cpu = cpu_next; > > spin_unlock_irqrestore(&multi_msi_cpu_lock, flags); > -- > 2.43.0 >
On Fri, Jan 03, 2025 at 11:45:43AM -0600, Bjorn Helgaas wrote: > On Sat, Dec 28, 2024 at 10:49:45AM -0800, Yury Norov wrote: > > Calling cpumask_next_wrap_old() with starting CPU == nr_cpu_ids > > is effectively the same as request to find first CPU, starting > > from a given one and wrapping around if needed. > > > > cpumask_next_wrap() is a proper replacement for that. > > > > Signed-off-by: Yury Norov <yury.norov@gmail.com> > > s/switch/Switch/ in subject to match history. > > Since this depends on previous patches, I assume you'll merge them all > together, so: > > Acked-by: Bjorn Helgaas <bhelgaas@google.com> Hi Bjorn, Thanks for review! Agree with everything you spotted out. I'll fix it in v2 if it will be needed, or inplace when applying. Thanks, Yury
diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c index 86d1c2be8eb5..f8ebf98248b3 100644 --- a/drivers/pci/controller/pci-hyperv.c +++ b/drivers/pci/controller/pci-hyperv.c @@ -1757,8 +1757,7 @@ static int hv_compose_multi_msi_req_get_cpu(void) spin_lock_irqsave(&multi_msi_cpu_lock, flags); - cpu_next = cpumask_next_wrap_old(cpu_next, cpu_online_mask, nr_cpu_ids, - false); + cpu_next = cpumask_next_wrap(cpu_next, cpu_online_mask); cpu = cpu_next; spin_unlock_irqrestore(&multi_msi_cpu_lock, flags);
Calling cpumask_next_wrap_old() with starting CPU == nr_cpu_ids is effectively the same as request to find first CPU, starting from a given one and wrapping around if needed. cpumask_next_wrap() is a proper replacement for that. Signed-off-by: Yury Norov <yury.norov@gmail.com> --- drivers/pci/controller/pci-hyperv.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)