kernel/drivers/pci/endpoint, branch linux-6.9.y

PCI: endpoint: Drop only_64bit on reserved BARs

2024-02-16T15:01:10Z

The definition of a reserved BAR is that EPF drivers should not touch them. The definition of only_64bit is that the EPF driver must configure this BAR as 64-bit. (An EPF driver is not allowed to choose if this BAR should be configured as 32-bit or 64-bit.) Thus, it does not make sense to put only_64bit of a BAR that EPF drivers are not allow to touch. Drop the only_64bit property from hardware descriptions that are of type reserved BAR. Signed-off-by: Niklas Cassel Reviewed-by: Kishon Vijay Abraham I Reviewed-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/20240216134524.1142149-3-cassel@kernel.org Signed-off-by: Manivannan Sadhasivam

PCI: endpoint: Clean up hardware description for BARs

2024-02-16T15:00:46Z

The hardware description for BARs is scattered in many different variables in pci_epc_features. Some of these things are mutually exclusive, so it can create confusion over which variable that has precedence over another. Improve the situation by creating a struct pci_epc_bar_desc, and a new enum pci_epc_bar_type, and convert the endpoint controller drivers to use this more well defined format. Additionally, some endpoint controller drivers mark the BAR succeeding a "64-bit only BAR" as reserved, while some do not. By definition, a 64-bit BAR uses the succeeding BAR for the upper 32-bits, so an EPF driver cannot use a BAR succeeding a 64-bit BAR. Ensure that all endpoint controller drivers are uniform, and actually describe a reserved BAR as reserved. Signed-off-by: Niklas Cassel Reviewed-by: Kishon Vijay Abraham I Reviewed-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/20240216134524.1142149-2-cassel@kernel.org Signed-off-by: Manivannan Sadhasivam

PCI: epf-mhi: Add support for SA8775P SoC

2024-02-16T10:49:14Z

Add support for Qualcomm Snapdragon SA8775P SoC to the EPF driver. SA8775P is currently reusing the PID 0x0306 (the default one hardcoded in the config space header) as the unique PID is not yet allocated. But the host side stack works fine with the default PID. It will get updated once the PID is finalized. Also, it has no fixed PCI class as of now, so it is being advertised as "PCI_CLASS_OTHERS". Signed-off-by: Mrinmay Sarkar Reviewed-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/1701432377-16899-5-git-send-email-quic_msarkar@quicinc.com Signed-off-by: Manivannan Sadhasivam

PCI: epf-mhi: Add "pci_epf_mhi_" prefix to the function names

2024-02-16T10:48:13Z

Without the prefix, the function name would appear as "/sys/kernel/config/functions/{sdx55/sm8450}". This will be a problem if multiple functions are supported for this endpoint device. So let's add the "pci_epf_mhi_" prefix to identify _this_ function uniquely. Even though it is an ABI breakage, this driver is not used anywhere outside Qcom and myself to my knowledge. So it safe to change the function name. Signed-off-by: Manivannan Sadhasivam Signed-off-by: Mrinmay Sarkar Link: https://lore.kernel.org/r/1701432377-16899-4-git-send-email-quic_msarkar@quicinc.com

PCI: epf-vntb: Return actual error code during pci_vntb_probe() failure

2024-02-16T10:44:01Z

If dma_set_mask_and_coherent() in pci_vntb_probe() fails, return the actual error code instead of -EINVAL. Signed-off-by: Yang Yingliang Reviewed-by: Manivannan Sadhasivam Reviewed-by: Dave Jiang Reviewed-by: Ilpo Järvinen Link: https://lore.kernel.org/r/20231201033057.1399131-2-yangyingliang@huaweicloud.com [mani: reworded commit message and subject] Signed-off-by: Manivannan Sadhasivam

NTB: fix possible name leak in ntb_register_device()

2024-02-16T10:43:38Z

If device_register() fails in ntb_register_device(), the device name allocated by dev_set_name() should be freed. As per the comment in device_register(), callers should use put_device() to give up the reference in the error path. So fix this by calling put_device() in the error path so that the name can be freed in kobject_cleanup(). As a result of this, put_device() in the error path of ntb_register_device() is removed and the actual error is returned. Fixes: a1bd3baeb2f1 ("NTB: Add NTB hardware abstraction layer") Signed-off-by: Yang Yingliang Reviewed-by: Ilpo Järvinen Reviewed-by: Manivannan Sadhasivam Reviewed-by: Dave Jiang Link: https://lore.kernel.org/r/20231201033057.1399131-1-yangyingliang@huaweicloud.com [mani: reworded commit message] Signed-off-by: Manivannan Sadhasivam

PCI: endpoint: pci-epf-vntb: Remove superfluous checks for pci_epf_alloc_space() API

2024-02-09T17:10:23Z

Now that the checks are performed by the pci_epf_alloc_space() API, let's remove the superfluous checks in this driver. Signed-off-by: Niklas Cassel Reviewed-by: Frank Li Reviewed-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/20240207213922.1796533-5-cassel@kernel.org [mani: reworded the commit message] Signed-off-by: Manivannan Sadhasivam

PCI: endpoint: pci-epf-test: Remove superfluous checks for pci_epf_alloc_space() API

2024-02-09T17:10:22Z

Now that the checks are performed by the pci_epf_alloc_space() API, let's remove the superfluous checks in this driver. Signed-off-by: Niklas Cassel Reviewed-by: Manivannan Sadhasivam Reviewed-by: Frank Li Link: https://lore.kernel.org/r/20240207213922.1796533-4-cassel@kernel.org [mani: reworded commit message] Signed-off-by: Manivannan Sadhasivam

PCI: endpoint: Improve pci_epf_alloc_space() API

2024-02-09T17:09:46Z

pci_epf_alloc_space() already performs checks on the requested BAR size, and will allocate and set epf_bar->size to a size higher than the requested BAR size if some constraint deems it necessary. However, there are additional checks done in the function drivers like pci-epf-test.c, other than the existing checks in this API. And similar checks are proposed to other endpoint function drivers, see: https://lore.kernel.org/linux-pci/20240108151015.2030469-1-Frank.Li@nxp.com Having these checks scattered over different locations in multiple EPF drivers is not maintainable and makes the code hard to follow. Since pci_epf_alloc_space() already performs roundups and some checks, let's move the additional checks from pci-epf-test.c to pci_epf_alloc_space(). This makes the API more robust and also offloads the checks from the function drivers. Signed-off-by: Niklas Cassel Reviewed-by: Manivannan Sadhasivam Reviewed-by: Frank Li Link: https://lore.kernel.org/r/20240207213922.1796533-3-cassel@kernel.org [mani: reworded commit message and fixed uninitialized 'dev' pointer issue] Signed-off-by: Manivannan Sadhasivam

PCI: endpoint: Refactor pci_epf_alloc_space() API

2024-02-09T08:50:19Z

Refactor pci_epf_alloc_space() API to accept "epc_features" as a parameter. This is a preparatory work to make the API more robust. Reviewed-by: Frank Li Reviewed-by: Manivannan Sadhasivam Signed-off-by: Niklas Cassel Link: https://lore.kernel.org/r/20240207213922.1796533-2-cassel@kernel.org [mani: reworded commit message] Signed-off-by: Manivannan Sadhasivam