2022-08-10 18:55:06

by William McVicker

[permalink] [raw]
Subject: [PATCH v2 0/2] PCI: dwc: Add support for 64-bit MSI target addresses

Hi All,

I have updated the patch series based on the build error reported by the
kernel test robot which found an issues with the new inline function when
PCIE_DW_HOST isn't set. Additionally, I have updated the DMA alloc error
handling for the first patch which was reported to me directly by Isaac
Manjarres. Basically, we don't need to check for a DMA mapping error since
dma_alloc_coherent() will free the memory when a mapping error occurs.

Please take a look and let me know if you have any questions or concerns.

Thanks,
Will

Reported-by: kernel test robot <[email protected]>
Reported-by: Isaac J. Manjarres <[email protected]>

Will McVicker (2):
PCI: dwc: drop dependency on ZONE_DMA32
PCI: dwc: add support for 64-bit MSI target address

v2:
* Fixed build error caught by kernel test robot
* Fixed error handling reported by Isaac Manjarres

.../pci/controller/dwc/pcie-designware-host.c | 37 +++++++++++--------
drivers/pci/controller/dwc/pcie-designware.c | 9 +++++
drivers/pci/controller/dwc/pcie-designware.h | 1 +
3 files changed, 31 insertions(+), 16 deletions(-)


base-commit: d4252071b97d2027d246f6a82cbee4d52f618b47
--
2.37.1.559.g78731f0fdb-goog


2022-08-10 19:24:28

by William McVicker

[permalink] [raw]
Subject: [PATCH v2 1/2] PCI: dwc: drop dependency on ZONE_DMA32

Re-work the msi_msg DMA allocation logic to use dma_alloc_coherent()
which uses the coherent DMA mask to try and return an allocation within
the DMA mask limits. This allows kernel configurations that disable
ZONE_DMA32 to continue supporting a 32-bit DMA mask. Without this patch,
the PCIe host device will fail to probe when ZONE_DMA32 is disabled.

Fixes: 35797e672ff0 ("PCI: dwc: Fix MSI msi_msg DMA mapping")
Reported-by: Isaac J. Manjarres <[email protected]>
Signed-off-by: Will McVicker <[email protected]>
---
.../pci/controller/dwc/pcie-designware-host.c | 23 ++++++++-----------
1 file changed, 9 insertions(+), 14 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
index 7746f94a715f..8f2222f51671 100644
--- a/drivers/pci/controller/dwc/pcie-designware-host.c
+++ b/drivers/pci/controller/dwc/pcie-designware-host.c
@@ -272,9 +272,9 @@ static void dw_pcie_free_msi(struct dw_pcie_rp *pp)
struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
struct device *dev = pci->dev;

- dma_unmap_page(dev, pp->msi_data, PAGE_SIZE, DMA_FROM_DEVICE);
- if (pp->msi_page)
- __free_page(pp->msi_page);
+ dma_free_coherent(dev, PAGE_SIZE, pp->msi_page, pp->msi_data);
+ pp->msi_data = 0;
+ pp->msi_page = NULL;
}
}

@@ -375,22 +375,17 @@ static int dw_pcie_msi_host_init(struct dw_pcie_rp *pp)
dw_chained_msi_isr, pp);
}

- ret = dma_set_mask(dev, DMA_BIT_MASK(32));
+ ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
if (ret)
dev_warn(dev, "Failed to set DMA mask to 32-bit. Devices with only 32-bit MSI support may not work properly\n");

- pp->msi_page = alloc_page(GFP_DMA32);
- pp->msi_data = dma_map_page(dev, pp->msi_page, 0,
- PAGE_SIZE, DMA_FROM_DEVICE);
- ret = dma_mapping_error(dev, pp->msi_data);
- if (ret) {
- dev_err(pci->dev, "Failed to map MSI data\n");
- __free_page(pp->msi_page);
- pp->msi_page = NULL;
+ pp->msi_page = dma_alloc_coherent(dev, PAGE_SIZE, &pp->msi_data,
+ GFP_KERNEL);
+ if (!pp->msi_page) {
+ dev_err(dev, "Failed to alloc and map MSI data\n");
pp->msi_data = 0;
dw_pcie_free_msi(pp);
-
- return ret;
+ return -ENOMEM;
}

return 0;
--
2.37.1.559.g78731f0fdb-goog

2022-08-10 21:54:37

by Rob Herring (Arm)

[permalink] [raw]
Subject: Re: [PATCH v2 1/2] PCI: dwc: drop dependency on ZONE_DMA32

On Wed, Aug 10, 2022 at 06:35:34PM +0000, Will McVicker wrote:
> Re-work the msi_msg DMA allocation logic to use dma_alloc_coherent()
> which uses the coherent DMA mask to try and return an allocation within
> the DMA mask limits. This allows kernel configurations that disable
> ZONE_DMA32 to continue supporting a 32-bit DMA mask. Without this patch,
> the PCIe host device will fail to probe when ZONE_DMA32 is disabled.
>
> Fixes: 35797e672ff0 ("PCI: dwc: Fix MSI msi_msg DMA mapping")
> Reported-by: Isaac J. Manjarres <[email protected]>
> Signed-off-by: Will McVicker <[email protected]>
> ---
> .../pci/controller/dwc/pcie-designware-host.c | 23 ++++++++-----------
> 1 file changed, 9 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
> index 7746f94a715f..8f2222f51671 100644
> --- a/drivers/pci/controller/dwc/pcie-designware-host.c
> +++ b/drivers/pci/controller/dwc/pcie-designware-host.c
> @@ -272,9 +272,9 @@ static void dw_pcie_free_msi(struct dw_pcie_rp *pp)
> struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> struct device *dev = pci->dev;
>
> - dma_unmap_page(dev, pp->msi_data, PAGE_SIZE, DMA_FROM_DEVICE);
> - if (pp->msi_page)
> - __free_page(pp->msi_page);
> + dma_free_coherent(dev, PAGE_SIZE, pp->msi_page, pp->msi_data);
> + pp->msi_data = 0;
> + pp->msi_page = NULL;
> }
> }
>
> @@ -375,22 +375,17 @@ static int dw_pcie_msi_host_init(struct dw_pcie_rp *pp)
> dw_chained_msi_isr, pp);
> }
>
> - ret = dma_set_mask(dev, DMA_BIT_MASK(32));
> + ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
> if (ret)
> dev_warn(dev, "Failed to set DMA mask to 32-bit. Devices with only 32-bit MSI support may not work properly\n");
>
> - pp->msi_page = alloc_page(GFP_DMA32);
> - pp->msi_data = dma_map_page(dev, pp->msi_page, 0,
> - PAGE_SIZE, DMA_FROM_DEVICE);
> - ret = dma_mapping_error(dev, pp->msi_data);
> - if (ret) {
> - dev_err(pci->dev, "Failed to map MSI data\n");
> - __free_page(pp->msi_page);
> - pp->msi_page = NULL;
> + pp->msi_page = dma_alloc_coherent(dev, PAGE_SIZE, &pp->msi_data,
> + GFP_KERNEL);

You can use the managed version, dmam_alloc_coherent(), and avoid the
freeing yourself. Also with that, I think you don't need 'msi_page'?

Also, no need to alloc a whole page. A u32 or u64? should be fine. The
write never makes it to memory, so doesn't really matter.

> + if (!pp->msi_page) {
> + dev_err(dev, "Failed to alloc and map MSI data\n");
> pp->msi_data = 0;
> dw_pcie_free_msi(pp);
> -
> - return ret;
> + return -ENOMEM;
> }
>
> return 0;
> --
> 2.37.1.559.g78731f0fdb-goog
>

2022-08-10 23:17:48

by William McVicker

[permalink] [raw]
Subject: Re: [PATCH v2 1/2] PCI: dwc: drop dependency on ZONE_DMA32

On 08/10/2022, Rob Herring wrote:
> On Wed, Aug 10, 2022 at 06:35:34PM +0000, Will McVicker wrote:
> > Re-work the msi_msg DMA allocation logic to use dma_alloc_coherent()
> > which uses the coherent DMA mask to try and return an allocation within
> > the DMA mask limits. This allows kernel configurations that disable
> > ZONE_DMA32 to continue supporting a 32-bit DMA mask. Without this patch,
> > the PCIe host device will fail to probe when ZONE_DMA32 is disabled.
> >
> > Fixes: 35797e672ff0 ("PCI: dwc: Fix MSI msi_msg DMA mapping")
> > Reported-by: Isaac J. Manjarres <[email protected]>
> > Signed-off-by: Will McVicker <[email protected]>
> > ---
> > .../pci/controller/dwc/pcie-designware-host.c | 23 ++++++++-----------
> > 1 file changed, 9 insertions(+), 14 deletions(-)
> >
> > diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
> > index 7746f94a715f..8f2222f51671 100644
> > --- a/drivers/pci/controller/dwc/pcie-designware-host.c
> > +++ b/drivers/pci/controller/dwc/pcie-designware-host.c
> > @@ -272,9 +272,9 @@ static void dw_pcie_free_msi(struct dw_pcie_rp *pp)
> > struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> > struct device *dev = pci->dev;
> >
> > - dma_unmap_page(dev, pp->msi_data, PAGE_SIZE, DMA_FROM_DEVICE);
> > - if (pp->msi_page)
> > - __free_page(pp->msi_page);
> > + dma_free_coherent(dev, PAGE_SIZE, pp->msi_page, pp->msi_data);
> > + pp->msi_data = 0;
> > + pp->msi_page = NULL;
> > }
> > }
> >
> > @@ -375,22 +375,17 @@ static int dw_pcie_msi_host_init(struct dw_pcie_rp *pp)
> > dw_chained_msi_isr, pp);
> > }
> >
> > - ret = dma_set_mask(dev, DMA_BIT_MASK(32));
> > + ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
> > if (ret)
> > dev_warn(dev, "Failed to set DMA mask to 32-bit. Devices with only 32-bit MSI support may not work properly\n");
> >
> > - pp->msi_page = alloc_page(GFP_DMA32);
> > - pp->msi_data = dma_map_page(dev, pp->msi_page, 0,
> > - PAGE_SIZE, DMA_FROM_DEVICE);
> > - ret = dma_mapping_error(dev, pp->msi_data);
> > - if (ret) {
> > - dev_err(pci->dev, "Failed to map MSI data\n");
> > - __free_page(pp->msi_page);
> > - pp->msi_page = NULL;
> > + pp->msi_page = dma_alloc_coherent(dev, PAGE_SIZE, &pp->msi_data,
> > + GFP_KERNEL);
>
> You can use the managed version, dmam_alloc_coherent(), and avoid the
> freeing yourself. Also with that, I think you don't need 'msi_page'?
>
> Also, no need to alloc a whole page. A u32 or u64? should be fine. The
> write never makes it to memory, so doesn't really matter.
>
> > + if (!pp->msi_page) {
> > + dev_err(dev, "Failed to alloc and map MSI data\n");
> > pp->msi_data = 0;
> > dw_pcie_free_msi(pp);
> > -
> > - return ret;
> > + return -ENOMEM;
> > }
> >
> > return 0;
> > --
> > 2.37.1.559.g78731f0fdb-goog
> >

Thanks for the review Rob! I've updated the series to address your concerns.

Regards,
Will