| Message ID | 20241107120739.415743-1-mheib@redhat.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 2cd78740effc587610ca2be6d803e3f61fc87ef6 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] net: atlantic: use irq_update_affinity_hint() | expand |
On Thu, 7 Nov 2024 14:07:39 +0200 Mohammad Heib wrote:
> Subject: [PATCH net] net: atlantic: use irq_update_affinity_hint()
Hi Mohammad, I see you're sending more off these. For future postings
please make sure you use net-next as subject tag (instead of just net).
We use net as a tag for fixes, and this is not a fix AFAIU.
No need to repost the 3 patches you already posted just for this.
On Thu, Nov 07, 2024 at 02:07:39PM +0200, Mohammad Heib wrote: > irq_set_affinity_hint() is deprecated, Use irq_update_affinity_hint() > instead. This removes the side-effect of actually applying the affinity. > > The driver does not really need to worry about spreading its IRQs across > CPUs. The core code already takes care of that. when the driver applies the > affinities by itself, it breaks the users' expectations: > > 1. The user configures irqbalance with IRQBALANCE_BANNED_CPULIST in > order to prevent IRQs from being moved to certain CPUs that run a > real-time workload. > > 2. atlantic device reopening will resets the affinity > in aq_ndev_open(). > > 3. atlantic has no idea about irqbalance's config, so it may move an IRQ to > a banned CPU. The real-time workload suffers unacceptable latency. > > Signed-off-by: Mohammad Heib <mheib@redhat.com> Reviewed-by: Simon Horman <horms@kernel.org>
Hello: This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Thu, 7 Nov 2024 14:07:39 +0200 you wrote: > irq_set_affinity_hint() is deprecated, Use irq_update_affinity_hint() > instead. This removes the side-effect of actually applying the affinity. > > The driver does not really need to worry about spreading its IRQs across > CPUs. The core code already takes care of that. when the driver applies the > affinities by itself, it breaks the users' expectations: > > [...] Here is the summary with links: - [net] net: atlantic: use irq_update_affinity_hint() https://git.kernel.org/netdev/net-next/c/2cd78740effc You are awesome, thank you!
diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c b/drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c index 43c71f6b314f..08630ee94251 100644 --- a/drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c +++ b/drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c @@ -162,8 +162,8 @@ int aq_pci_func_alloc_irq(struct aq_nic_s *self, unsigned int i, self->msix_entry_mask |= (1 << i); if (pdev->msix_enabled && affinity_mask) - irq_set_affinity_hint(pci_irq_vector(pdev, i), - affinity_mask); + irq_update_affinity_hint(pci_irq_vector(pdev, i), + affinity_mask); } return err; @@ -187,7 +187,7 @@ void aq_pci_func_free_irqs(struct aq_nic_s *self) continue; if (pdev->msix_enabled) - irq_set_affinity_hint(pci_irq_vector(pdev, i), NULL); + irq_update_affinity_hint(pci_irq_vector(pdev, i), NULL); free_irq(pci_irq_vector(pdev, i), irq_data); self->msix_entry_mask &= ~(1U << i); }
irq_set_affinity_hint() is deprecated, Use irq_update_affinity_hint() instead. This removes the side-effect of actually applying the affinity. The driver does not really need to worry about spreading its IRQs across CPUs. The core code already takes care of that. when the driver applies the affinities by itself, it breaks the users' expectations: 1. The user configures irqbalance with IRQBALANCE_BANNED_CPULIST in order to prevent IRQs from being moved to certain CPUs that run a real-time workload. 2. atlantic device reopening will resets the affinity in aq_ndev_open(). 3. atlantic has no idea about irqbalance's config, so it may move an IRQ to a banned CPU. The real-time workload suffers unacceptable latency. Signed-off-by: Mohammad Heib <mheib@redhat.com> --- drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)