| Message ID | 20220609161457.69614-1-jean-philippe@linaro.org (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 884c65e4daf3eab8730b2bbd5abc5a2c0403b3f3 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | amd-xgbe: Use platform_irq_count() | expand |
On Thu, Jun 9, 2022 at 10:20 AM Jean-Philippe Brucker <jean-philippe@linaro.org> wrote: > > The AMD XGbE driver currently counts the number of interrupts assigned > to the device by inspecting the pdev->resource array. Since commit > a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT > core") removed IRQs from this array, the driver now attempts to get all > interrupts from 1 to -1U and gives up probing once it reaches an invalid > interrupt index. > > Obtain the number of IRQs with platform_irq_count() instead. > > Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core") > Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org> > --- > drivers/net/ethernet/amd/xgbe/xgbe-platform.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > index 4ebd2410185a..4d790a89fe77 100644 > --- a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > +++ b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > @@ -338,7 +338,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) > * the PHY resources listed last > */ > phy_memnum = xgbe_resource_count(pdev, IORESOURCE_MEM) - 3; > - phy_irqnum = xgbe_resource_count(pdev, IORESOURCE_IRQ) - 1; > + phy_irqnum = platform_irq_count(pdev) - 1; > dma_irqnum = 1; > dma_irqend = phy_irqnum; > } else { > @@ -348,7 +348,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) > phy_memnum = 0; > phy_irqnum = 0; > dma_irqnum = 1; > - dma_irqend = xgbe_resource_count(pdev, IORESOURCE_IRQ); > + dma_irqend = platform_irq_count(pdev); This is some ugly code... This clause is for 'old style' DT (which actually looks more correct to me then 'new style' which seems influenced by ACPI). Is there a need to continue carrying support for old DTs? How many users of Seattle with DT are there? In any case, that's all a separate change on top of this I think. Acked-by: Rob Herring <robh@kernel.org> Rob
On 6/9/22 11:14, Jean-Philippe Brucker wrote: > The AMD XGbE driver currently counts the number of interrupts assigned > to the device by inspecting the pdev->resource array. Since commit > a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT > core") removed IRQs from this array, the driver now attempts to get all > interrupts from 1 to -1U and gives up probing once it reaches an invalid > interrupt index. > > Obtain the number of IRQs with platform_irq_count() instead. > > Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core") > Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> It looks like the CCP driver does a similar loop looking for IRQ resources (sp_get_irqs() in drivers/crypto/ccp/sp-platform.c), have you looked at fixing that driver, too? I can submit a patch if you hadn't planned on it. Thanks, Tom > --- > drivers/net/ethernet/amd/xgbe/xgbe-platform.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > index 4ebd2410185a..4d790a89fe77 100644 > --- a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > +++ b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c > @@ -338,7 +338,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) > * the PHY resources listed last > */ > phy_memnum = xgbe_resource_count(pdev, IORESOURCE_MEM) - 3; > - phy_irqnum = xgbe_resource_count(pdev, IORESOURCE_IRQ) - 1; > + phy_irqnum = platform_irq_count(pdev) - 1; > dma_irqnum = 1; > dma_irqend = phy_irqnum; > } else { > @@ -348,7 +348,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) > phy_memnum = 0; > phy_irqnum = 0; > dma_irqnum = 1; > - dma_irqend = xgbe_resource_count(pdev, IORESOURCE_IRQ); > + dma_irqend = platform_irq_count(pdev); > } > > /* Obtain the mmio areas for the device */
On Mon, Jun 13, 2022 at 11:39:48AM -0500, Tom Lendacky wrote: > On 6/9/22 11:14, Jean-Philippe Brucker wrote: > > The AMD XGbE driver currently counts the number of interrupts assigned > > to the device by inspecting the pdev->resource array. Since commit > > a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT > > core") removed IRQs from this array, the driver now attempts to get all > > interrupts from 1 to -1U and gives up probing once it reaches an invalid > > interrupt index. > > > > Obtain the number of IRQs with platform_irq_count() instead. > > > > Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core") > > Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org> > > Acked-by: Tom Lendacky <thomas.lendacky@amd.com> > > It looks like the CCP driver does a similar loop looking for IRQ resources > (sp_get_irqs() in drivers/crypto/ccp/sp-platform.c), have you looked at > fixing that driver, too? I can submit a patch if you hadn't planned on it. Ah no, I did very briefly look for similar issues but missed it, feel free to send the patch Thanks, Jean
Hello: This patch was applied to netdev/net.git (master) by Jakub Kicinski <kuba@kernel.org>: On Thu, 9 Jun 2022 17:14:59 +0100 you wrote: > The AMD XGbE driver currently counts the number of interrupts assigned > to the device by inspecting the pdev->resource array. Since commit > a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT > core") removed IRQs from this array, the driver now attempts to get all > interrupts from 1 to -1U and gives up probing once it reaches an invalid > interrupt index. > > [...] Here is the summary with links: - amd-xgbe: Use platform_irq_count() https://git.kernel.org/netdev/net/c/884c65e4daf3 You are awesome, thank you!
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c index 4ebd2410185a..4d790a89fe77 100644 --- a/drivers/net/ethernet/amd/xgbe/xgbe-platform.c +++ b/drivers/net/ethernet/amd/xgbe/xgbe-platform.c @@ -338,7 +338,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) * the PHY resources listed last */ phy_memnum = xgbe_resource_count(pdev, IORESOURCE_MEM) - 3; - phy_irqnum = xgbe_resource_count(pdev, IORESOURCE_IRQ) - 1; + phy_irqnum = platform_irq_count(pdev) - 1; dma_irqnum = 1; dma_irqend = phy_irqnum; } else { @@ -348,7 +348,7 @@ static int xgbe_platform_probe(struct platform_device *pdev) phy_memnum = 0; phy_irqnum = 0; dma_irqnum = 1; - dma_irqend = xgbe_resource_count(pdev, IORESOURCE_IRQ); + dma_irqend = platform_irq_count(pdev); } /* Obtain the mmio areas for the device */
The AMD XGbE driver currently counts the number of interrupts assigned to the device by inspecting the pdev->resource array. Since commit a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core") removed IRQs from this array, the driver now attempts to get all interrupts from 1 to -1U and gives up probing once it reaches an invalid interrupt index. Obtain the number of IRQs with platform_irq_count() instead. Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core") Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org> --- drivers/net/ethernet/amd/xgbe/xgbe-platform.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)