linux-dev/arch/arm/mach-footbridge, branch master

Merge branch 'arm-multiplatform-cleanup' of https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc into arm/soc

2022-09-15T20:20:59Z

Now that everything except StrongARM is unified under CONFIG_ARCH_MULTIPLATFORM, the option is rather meaningless in its current form. Rework the Kconfig logic to make this useful again, similar to the way that RISC-V has CONFIG_NONPORTABLE (with the opposite polarity), this now controls the visibility of options that get in the way of building generic kernels, while allowing custom kernels. One side-effect is that 'randconfig' builds now rarely hit strongarm machines, rather than testing them three quarters of the time. * 'arm-multiplatform-cleanup' of https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: ARM: make ARCH_MULTIPLATFORM user-visible ARM: fix XIP_KERNEL dependencies ARM: Kconfig: clean up platform selection ARM: simplify machdirs/platdirs handling ARM: remove obsolete Makefile.boot infrastructure

ARM: footbridge: remove custom DMA address handling

2022-09-15T13:59:16Z

Footbridge is the last Arm platform that has its own __virt_to_bus()/__bus_to_virt()/phys_to_dma()/dma_to_phys() abstraction, but this is just a simple offset now. For PCI devices, the offset that is programmed into the PCI bridge must also be set in each device using dma_direct_set_offset(). As Arm does not have a pcibios_bus_add_device() helper yet, just use a bus notifier for this. For the ISA DMA, drivers now pass a non-translated physical address into set_dma_addr(), so they have to be converted back with the corresponding isa_bus_to_virt() function and then into the correct bus address with the offset using the isa_dma_dev. Tested-by: Marc Zyngier Signed-off-by: Arnd Bergmann

ARM: footbridge: move isa-dma support into footbridge

2022-09-09T15:14:34Z

The dma-isa.c was shared between footbridge and shark a long time ago, but as shark was removed, it can be made footbridge specific again. The fb_dma bits in turn are not used at all and can be removed. All the ISA related files are now built into the platform regardless of CONFIG_ISA, as they just refer to on-chip devices rather than actual ISA cards. Reviewed-by: Christoph Hellwig Tested-by: Marc Zyngier Signed-off-by: Arnd Bergmann

ARM: footbridge: remove leftover from personal-server

2022-09-09T15:14:28Z

The machine was removed a while ago, and the checks are now useless. Reviewed-by: Christoph Hellwig Tested-by: Marc Zyngier Signed-off-by: Arnd Bergmann

ARM: footbridge: remove addin mode

2022-09-09T15:14:16Z

This does not appear to have been used in many years, we can kill off some of the uglier code. Among other things, it avoids a randconfig issue when both modes are disabled: arch/arm/mach-footbridge/common.c:149:24: error: 'ebsa285_host_io_desc' defined but not used [-Werror=unused-variable] 149 | static struct map_desc ebsa285_host_io_desc[] __initdata = { | ^~~~~~~~~~~~~~~~~~~~ arch/arm/mach-footbridge/common.c:136:24: error: 'fb_common_io_desc' defined but not used [-Werror=unused-variable] 136 | static struct map_desc fb_common_io_desc[] __initdata = { | ^~~~~~~~~~~~~~~~~ The recently added phys_to_dma() functions are now trivial and could probably be removed again as a follow-up, if anyone knows how. Cc: Christoph Hellwig Tested-by: Marc Zyngier Signed-off-by: Arnd Bergmann

ARM: Kconfig: clean up platform selection

2022-08-18T14:44:59Z

The top-level platform selection is mostly meaningless these days after almost everything is sorted below the CONFIG_ARCH_MULTIPLATFORM, with the only exception being the 20+ year old StrongARM based machines. Make this more consistent by removing the entire choice statement and moving the StrongARM specific options into regular platform specific Kconfig files. The three platforms (footbridge, rpc and sa1100) are still mutually exclusive and cannot coexist with other ARMv4/v5 machines, but since there are only three of them and we will not add more, this can be expressed using Kconfig 'depends on' statements. Signed-off-by: Arnd Bergmann

ARM: remove obsolete Makefile.boot infrastructure

2022-08-18T14:44:59Z

There are a number of old Makefile.boot files that remain from the multiplatform conversion, and three that are still in use. These provide the "ZRELADDR", "PARAMS_PHYS" and "INITRD_PHYS" values that are platform specific. It turns out that we can generally just derive this from information that is available elsewhere: - ZRELADDR is normally detected at runtime with the CONFIG_AUTO_ZRELADDR flag, but also needed to be passed to for 'make uImage'. In a multiplatform kernel, one always has to pass this as the $(LOADADDR) variable, but in the StrongARM kernels we can derive it from the sum of $(CONFIG_PHYS_OFFSET) and $(TEXT_OFFSET) that are already known. - PARAMS_PHYS and INITRD_PHYS are only used for bootpImage, which in turn is only used for the pre-ATAGS 'param_struct' based boot interface on StrongARM based machines with old boot loaders. They can both be derived from CONFIG_PHYS_OFFSET and a machine specific offset for the initrd, so all of the logic for these can be part of arch/arm/boot/bootp/Makefile. Signed-off-by: Arnd Bergmann

Merge tag 'dma-mapping-5.20-2022-08-06' of git://git.infradead.org/users/hch/dma-mapping

2022-08-06T17:56:45Z

Pull dma-mapping updates from Christoph Hellwig: - convert arm32 to the common dma-direct code (Arnd Bergmann, Robin Murphy, Christoph Hellwig) - restructure the PCIe peer to peer mapping support (Logan Gunthorpe) - allow the IOMMU code to communicate an optional DMA mapping length and use that in scsi and libata (John Garry) - split the global swiotlb lock (Tianyu Lan) - various fixes and cleanup (Chao Gao, Dan Carpenter, Dongli Zhang, Lukas Bulwahn, Robin Murphy) * tag 'dma-mapping-5.20-2022-08-06' of git://git.infradead.org/users/hch/dma-mapping: (45 commits) swiotlb: fix passing local variable to debugfs_create_ulong() dma-mapping: reformat comment to suppress htmldoc warning PCI/P2PDMA: Remove pci_p2pdma_[un]map_sg() RDMA/rw: drop pci_p2pdma_[un]map_sg() RDMA/core: introduce ib_dma_pci_p2p_dma_supported() nvme-pci: convert to using dma_map_sgtable() nvme-pci: check DMA ops when indicating support for PCI P2PDMA iommu/dma: support PCI P2PDMA pages in dma-iommu map_sg iommu: Explicitly skip bus address marked segments in __iommu_map_sg() dma-mapping: add flags to dma_map_ops to indicate PCI P2PDMA support dma-direct: support PCI P2PDMA pages in dma-direct map_sg dma-mapping: allow EREMOTEIO return code for P2PDMA transfers PCI/P2PDMA: Introduce helpers for dma_map_sg implementations PCI/P2PDMA: Attempt to set map_type if it has not been set lib/scatterlist: add flag for indicating P2PDMA segments in an SGL swiotlb: clean up some coding style and minor issues dma-mapping: update comment after dmabounce removal scsi: sd: Add a comment about limiting max_sectors to shost optimal limit ata: libata-scsi: cap ata_device->max_sectors according to shost->max_sectors scsi: scsi_transport_sas: cap shost opt_sectors according to DMA optimal limit ...

ARM: footbridge: mark cats board for removal

2022-07-22T11:11:21Z

There are three remaining footbridge boards, as the CO285 and the HP personal server got removed already over the years. Russell still uses his ebsa285, while both Linus and Marc have a NetWinder that they use for testing. Nobody so far replied that they are using cats, so it goes on the long list of machines to be removed in early 2023 if it stays like this. Cc: Russell King Cc: Linus Walleij Cc: Marc Zyngier Signed-off-by: Arnd Bergmann

ARM/dma-mapping: use the generic versions of dma_to_phys/phys_to_dma by default

2022-07-07T16:18:57Z

Only the footbridge platforms provide their own DMA address translation helpers, so switch to the generic version for all other platforms, and consolidate the footbridge implementation to remove two levels of indirection. Signed-off-by: Christoph Hellwig Reviewed-by: Arnd Bergmann Tested-by: Marc Zyngier