blaster4385/linux-IllusionX/drivers/of/address.c, branch v6.12.10

of: Fix refcount leakage for OF node returned by __of_get_dma_parent()

2024-12-27T13:02:19Z

commit 5d009e024056ded20c5bb1583146b833b23bbd5a upstream. __of_get_dma_parent() returns OF device node @args.np, but the node's refcount is increased twice, by both of_parse_phandle_with_args() and of_node_get(), so causes refcount leakage for the node. Fix by directly returning the node got by of_parse_phandle_with_args(). Fixes: f83a6e5dea6c ("of: address: Add support for the parent DMA bus") Cc: stable@vger.kernel.org Signed-off-by: Zijun Hu Link: https://lore.kernel.org/r/20241206-of_core_fix-v1-4-dc28ed56bec3@quicinc.com Signed-off-by: Rob Herring (Arm) Signed-off-by: Greg Kroah-Hartman

of: address: Preserve the flags portion on 1:1 dma-ranges mapping

2024-12-27T13:02:18Z

commit 7f05e20b989ac33c9c0f8c2028ec0a566493548f upstream. A missing or empty dma-ranges in a DT node implies a 1:1 mapping for dma translations. In this specific case, the current behaviour is to zero out the entire specifier so that the translation could be carried on as an offset from zero. This includes address specifier that has flags (e.g. PCI ranges). Once the flags portion has been zeroed, the translation chain is broken since the mapping functions will check the upcoming address specifier against mismatching flags, always failing the 1:1 mapping and its entire purpose of always succeeding. Set to zero only the address portion while passing the flags through. Fixes: dbbdee94734b ("of/address: Merge all of the bus translation code") Cc: stable@vger.kernel.org Signed-off-by: Andrea della Porta Tested-by: Herve Codina Link: https://lore.kernel.org/r/e51ae57874e58a9b349c35e2e877425ebc075d7a.1732441813.git.andrea.porta@suse.com Signed-off-by: Rob Herring (Arm) Signed-off-by: Greg Kroah-Hartman

of: address: Unify resource bounds overflow checking

2024-09-13T19:00:52Z

The members "start" and "end" of struct resource are of type "resource_size_t" which can be 32bit wide. Values read from OF however are always 64bit wide. Refactor the diff overflow checks into a helper function. Also extend the checks to validate each calculation step. Signed-off-by: Thomas Weißschuh Link: https://lore.kernel.org/r/20240906-of-address-overflow-v1-1-19567aaa61da@linutronix.de [robh: Fix to not return error on 0 sized resource] Signed-off-by: Rob Herring (Arm)

of: address: Report error on resource bounds overflow

2024-09-05T13:19:47Z

The members "start" and "end" of struct resource are of type "resource_size_t" which can be 32bit wide. Values read from OF however are always 64bit wide. Avoid silently truncating the value and instead return an error value. This can happen on real systems when the DT was created for a PAE-enabled kernel and a non-PAE kernel is actually running. For example with an arm defconfig and "qemu-system-arm -M virt". Link: https://bugs.launchpad.net/qemu/+bug/1790975 Signed-off-by: Thomas Weißschuh Tested-by: Nam Cao Reviewed-by: Nam Cao Link: https://lore.kernel.org/r/20240905-of-resource-overflow-v1-1-0cd8bb92cc1f@linutronix.de Cc: stable@vger.kernel.org Signed-off-by: Rob Herring (Arm)

of: Use scope based of_node_put() cleanups

2024-04-15T13:40:40Z

Use the relatively new scope based of_node_put() cleanup to simplify function exit handling. Doing so reduces the chances of forgetting an of_node_put() and simplifies error paths by avoiding the need for goto statements. Reviewed-by: Saravana Kannan Reviewed-by: Jonathan Cameron Link: https://lore.kernel.org/r/20240409-dt-cleanup-free-v2-3-5b419a4af38d@kernel.org Signed-off-by: Rob Herring

dma-mapping: don't store redundant offsets

2023-12-15T11:32:42Z

A bus_dma_region necessarily stores both CPU and DMA base addresses for a range, so there's no need to also store the difference between them. Signed-off-by: Robin Murphy Acked-by: Rob Herring Signed-off-by: Christoph Hellwig

of: address: Consolidate bus .map() functions

2023-10-27T21:37:42Z

The bus .map() functions vary only by checking the flag cells values and skipping over any flag cells to read the addresses. Otherwise they all do the same reading 'ranges' address and size and returning the address's offset if it is within the 'ranges' entry. Refactor all the .map() functions to pass in the flag cell size so that each bus can check the bus specific flags and then call a common function to do everything else. Acked-by: Herve Codina Link: https://lore.kernel.org/r/20231026135358.3564307-3-robh@kernel.org Signed-off-by: Rob Herring

of: address: Store number of bus flag cells rather than bool

2023-10-27T21:37:42Z

It is more useful to know how many flags cells a bus has rather than whether a bus has flags or not as ultimately the number of cells is the information used. Replace 'has_flags' boolean with 'flag_cells' count. Acked-by: Herve Codina Link: https://lore.kernel.org/r/20231026135358.3564307-2-robh@kernel.org Signed-off-by: Rob Herring

of: address: Remove duplicated functions

2023-10-26T14:07:14Z

The recently added of_bus_default_flags_translate() performs the exact same operation as of_bus_pci_translate() and of_bus_isa_translate(). Avoid duplicated code replacing both of_bus_pci_translate() and of_bus_isa_translate() with of_bus_default_flags_translate(). Signed-off-by: Herve Codina Link: https://lore.kernel.org/r/20231017110221.189299-3-herve.codina@bootlin.com Signed-off-by: Rob Herring

of: address: Fix address translation when address-size is greater than 2

2023-10-26T14:05:50Z

With the recent addition of of_pci_prop_ranges() in commit 407d1a51921e ("PCI: Create device tree node for bridge"), the ranges property can have a 3 cells child address, a 3 cells parent address and a 2 cells child size. A range item property for a PCI device is filled as follow: 0 0 <-- Child --> <-- Parent (PCI definition) --> <- BAR size (64bit) --> This allow to translate BAR addresses from the DT. For instance: pci@0,0 { #address-cells = <0x03>; #size-cells = <0x02>; device_type = "pci"; compatible = "pci11ab,100", "pciclass,060400", "pciclass,0604"; ranges = <0x82000000 0x00 0xe8000000 0x82000000 0x00 0xe8000000 0x00 0x4400000>; ... dev@0,0 { #address-cells = <0x03>; #size-cells = <0x02>; compatible = "pci1055,9660", "pciclass,020000", "pciclass,0200"; /* Translations for BAR0 to BAR5 */ ranges = <0x00 0x00 0x00 0x82010000 0x00 0xe8000000 0x00 0x2000000 0x01 0x00 0x00 0x82010000 0x00 0xea000000 0x00 0x1000000 0x02 0x00 0x00 0x82010000 0x00 0xeb000000 0x00 0x800000 0x03 0x00 0x00 0x82010000 0x00 0xeb800000 0x00 0x800000 0x04 0x00 0x00 0x82010000 0x00 0xec000000 0x00 0x20000 0x05 0x00 0x00 0x82010000 0x00 0xec020000 0x00 0x2000>; ... pci-ep-bus@0 { #address-cells = <0x01>; #size-cells = <0x01>; compatible = "simple-bus"; /* Translate 0xe2000000 to BAR0 and 0xe0000000 to BAR1 */ ranges = <0xe2000000 0x00 0x00 0x00 0x2000000 0xe0000000 0x01 0x00 0x00 0x1000000>; ... }; }; }; During the translation process, the "default-flags" map() function is used to select the matching item in the ranges table and determine the address offset from this matching item. This map() function simply calls of_read_number() and when address-size is greater than 2, the map() function skips the extra high address part (ie part over 64bit). This lead to a wrong matching item and a wrong offset computation. Also during the translation itself, the extra high part related to the parent address is not present in the translated address. Fix the "default-flags" map() and translate() in order to take into account the child extra high address part in map() and the parent extra high address part in translate() and so having a correct address translation for ranges patterns such as the one given in the example above. Signed-off-by: Herve Codina Link: https://lore.kernel.org/r/20231017110221.189299-2-herve.codina@bootlin.com Signed-off-by: Rob Herring