kernel/linux.git/drivers/pci/endpoint, branch v5.6.17

PCI: endpoint: Fix for concurrent memory allocation in OB address region

2020-04-17T14:13:29+00:00

commit 04e046ca57ebed3943422dee10eec9e73aec081e upstream. pci-epc-mem uses a bitmap to manage the Endpoint outbound (OB) address region. This address region will be shared by multiple endpoint functions (in the case of multi function endpoint) and it has to be protected from concurrent access to avoid updating an inconsistent state. Use a mutex to protect bitmap updates to prevent the memory allocation API from returning incorrect addresses. Signed-off-by: Kishon Vijay Abraham I Signed-off-by: Lorenzo Pieralisi Cc: stable@vger.kernel.org # v4.14+ Signed-off-by: Greg Kroah-Hartman

Merge branch 'remotes/lorenzo/pci/endpoint'

2019-11-28T14:54:42+00:00

- Fix endpoint driver sign extension problem when shifting page number to phys_addr_t (Alan Mikhak) * remotes/lorenzo/pci/endpoint: PCI: endpoint: Cast the page number to phys_addr_t

PCI: endpoint: Cast the page number to phys_addr_t

2019-10-15T14:47:54+00:00

Modify pci_epc_mem_alloc_addr() to cast the variable 'pageno' from type 'int' to 'phys_addr_t' before shifting left. This cast is needed to avoid treating bit 31 of 'pageno' as the sign bit which would otherwise get sign-extended to produce a negative value. When added to the base address of PCI memory space, the negative value would produce an invalid physical address which falls before the start of the PCI memory space. Signed-off-by: Alan Mikhak Signed-off-by: Lorenzo Pieralisi Acked-by: Kishon Vijay Abraham I

PCI: Add PCI_STD_NUM_BARS for the number of standard BARs

2019-10-14T15:22:26+00:00

Code that iterates over all standard PCI BARs typically uses PCI_STD_RESOURCE_END. However, that requires the unusual test "i <= PCI_STD_RESOURCE_END" rather than something the typical "i < PCI_STD_NUM_BARS". Add a definition for PCI_STD_NUM_BARS and change loops to use the more idiomatic C style to help avoid fencepost errors. Link: https://lore.kernel.org/r/20190927234026.23342-1-efremov@linux.com Link: https://lore.kernel.org/r/20190927234308.23935-1-efremov@linux.com Link: https://lore.kernel.org/r/20190916204158.6889-3-efremov@linux.com Signed-off-by: Denis Efremov Signed-off-by: Bjorn Helgaas Acked-by: Sebastian Ott # arch/s390/ Acked-by: Bartlomiej Zolnierkiewicz # video/fbdev/ Acked-by: Gustavo Pimentel # pci/controller/dwc/ Acked-by: Jack Wang # scsi/pm8001/ Acked-by: Martin K. Petersen # scsi/pm8001/ Acked-by: Ulf Hansson # memstick/

PCI: endpoint: Clear BAR before freeing its space

2019-06-11T09:57:54+00:00

Associated pci_epf_bar structure is needed in pci_epc_clear_bar() to clear a BAR correctly but it is reset in pci_epf_free_space() (that is called first) which results in pci_epc_clear_bar() failure. Reorder the pci_epc_clear_bar()/pci_epf_free_space() calls execution to fix the issue. Signed-off-by: Alan Mikhak [lorenzo.pieralisi@arm.com: reworded the commit log] Signed-off-by: Lorenzo Pieralisi Acked-by: Kishon Vijay Abraham I

PCI: endpoint: Skip odd BAR when skipping 64bit BAR

2019-06-11T09:55:36+00:00

Always skip odd BAR when skipping 64bit BARs in pci_epf_test_set_bar() and pci_epf_test_alloc_space() otherwise pci_epf_test_set_bar() will call pci_epc_set_bar() on an odd loop index when skipping reserved 64bit BAR. Moreover, pci_epf_test_alloc_space() will call pci_epf_alloc_space() on bind for an odd loop index when BAR is 64bit but leaks on subsequent unbind by not calling pci_epf_free_space(). Signed-off-by: Alan Mikhak Signed-off-by: Lorenzo Pieralisi Acked-by: Kishon Vijay Abraham I Reviewed-by: Paul Walmsley

PCI: endpoint: Allocate enough space for fixed size BAR

2019-06-11T09:55:22+00:00

PCI endpoint test function code should honor the .bar_fixed_size parameter from underlying endpoint controller drivers or results may be unexpected. In pci_epf_test_alloc_space(), check if BAR being used for test register space is a fixed size BAR. If so, allocate the required fixed size. Signed-off-by: Alan Mikhak Signed-off-by: Lorenzo Pieralisi Acked-by: Kishon Vijay Abraham I

PCI: endpoint: Set endpoint controller pointer to NULL

2019-06-11T09:55:12+00:00

Set endpoint controller pointer to NULL in pci_epc_remove_epf() to avoid -EBUSY on subsequent call to pci_epc_add_epf(). Add a check for NULL endpoint function pointer. Signed-off-by: Alan Mikhak Signed-off-by: Lorenzo Pieralisi Acked-by: Kishon Vijay Abraham I

Merge branch 'remotes/lorenzo/pci/misc'

2019-05-13T23:34:44+00:00

- Exit pcitest with error code when test fails (Jean-Jacques Hiblot) - Fix leaked of_node references in dra7xx, uniphier, layerscape, rockchip, aardvark, iproc, mediatek, rpadlpar (Wen Yang) - Fix pcitest "help" option parsing (Kishon Vijay Abraham I) - Fix Makefile bug that inadvertently removes pcitest.sh (Kishon Vijay Abraham I) - Check for alloc_workqueue() failure in endpoint test driver (Kangjie Lu) * remotes/lorenzo/pci/misc: PCI: endpoint: Fix a potential NULL pointer dereference tools: PCI: Handle pcitest.sh independently from pcitest tools: PCI: Add 'h' in optstring of getopt() PCI: mediatek: Fix a leaked reference by adding missing of_node_put() PCI: iproc: Fix a leaked reference by adding missing of_node_put() PCI: aardvark: Fix a leaked reference by adding missing of_node_put() PCI: rockchip: Fix a leaked reference by adding missing of_node_put() PCI: dwc: layerscape: Fix a leaked reference by adding missing of_node_put() PCI: uniphier: Fix a leaked reference by adding missing of_node_put() PCI: dwc: pci-dra7xx: Fix a leaked reference by adding missing of_node_put() tools: PCI: Exit with error code when test fails

PCI: endpoint: Add support to specify alignment for buffers allocated to BARs

2019-04-15T12:24:02+00:00

The address that is allocated using pci_epf_alloc_space() is directly written to the target address of the Inbound Address Translation unit (ie the HW component implementing inbound address decoding) on endpoint controllers. Designware IP [1] has a configuration parameter (CX_ATU_MIN_REGION_SIZE [2]) which has 64KB as default value and the lower 16 bits of the Base, Limit and Target registers of the Inbound ATU are fixed to zero. If the programmed memory address is not aligned to 64 KB boundary this causes memory corruption. Modify pci_epf_alloc_space() API to take alignment size as argument in order to allocate buffers to be mapped to BARs with an alignment that suits the platform where they are used. Add an 'align' parameter to epc_features which can be used by platform drivers to specify the BAR allocation alignment requirements and use this while invoking pci_epf_alloc_space(). [1] "I/O and MEM Match Modes" section in DesignWare Cores PCI Express Controller Databook version 4.90a [2] http://www.ti.com/lit/ug/spruid7c/spruid7c.pdf Signed-off-by: Kishon Vijay Abraham I Signed-off-by: Lorenzo Pieralisi