Changes in Kconfig

d5cae2261b86913e602452ce4a07e6aefc0f603b - dma-contiguous: simplify numa cma area handling

Mon, 25 May 2026 03:51:11 +0200

dma-contiguous: simplify numa cma area handlingCurrently, there are 2 kernel cmdline ways to setup numa cma area:"cma_pernuma=" and "numa_cma=", and there are 2 cma arrays as well,while they have no difference technically. Robin suggested to cleanupthe code and only use one array [1], as "the apparent intent thatusers only want one _or_ the other".Simplify the code by only using one array to save the numa cma area.And in rare case that a user really setup the 2 cmdline parametersat the same time, let the per-node specific size setting 'numa_cma='take priority over the global numa cma setting.Link[1]: https://lore.kernel.org/lkml/43c5301c-fe6a-41e4-9482-ccfc7b62f2a7@arm.com/Suggested-by: Robin Murphy Signed-off-by: Feng Tang Signed-off-by: Marek Szyprowski Link: https://lore.kernel.org/r/20260525015111.6267-1-feng.tang@linux.alibaba.com List of files: /linux/kernel/dma/Kconfig

7e6ace2535d032c908e4d8747d9a7952617c001a - dma-contiguous: add kconfig option to setup numa cma area if not configured explicitly

Tue, 12 May 2026 10:55:09 +0200

dma-contiguous: add kconfig option to setup numa cma area if not configured explicitlyThere was a report on a multi-numa-nodes arm64 server that when IOMMUis disabled, the dma_alloc_coherent() function always returns memoryfrom node 0 even for devices attaching to other nodes, while they canget local dma memory when IOMMU is on with the same API.The reason is, when IOMMU is disabled, the dma_alloc_coherent() willgo the direct way and call dma_alloc_contiguous(). The system doesn'thave any explicit cma setting (like per-numa cma), and only has adefault 64MB cma reserved area (on node 0), where kernel will tryfirst to allocate memory from.Robin Murphy suggested to setup pernuma cma or disable cma, which didsolve the issue. While there is still concern that for customerswhich don't have much kernel knowledge, they could still suffer fromthis silently as some architectures enable cma area by default (notan issue for X86 though, which set CONFIG_CMA_SIZE_MBYTES to 0 bydefault) for most Linux distributions.One thought is to follow the current cma reserving policy for platformwith 'CONFIG_DMA_NUMA_CMA=y', that if the numa cma (either the 'numa cma'or 'cma pernuma' method) is not explicitly configured, and the platformreally has multiple NUMA nodes, set it up according to size of default'dma_contiguous_default_area'. This way, the default behavior ofplatform with one NUMA node is kept unchanged (say embedded/smalldevices don't need to allocate extra memory), while the general dmalocality is improved.Add a new bool kernel config CONFIG_CMA_SIZE_PERNUMA to control whetherto enable it. Even when the config is enabled, user can still disableit by kernel-cmdline setting like "numa_cma=0:0" or "cma_pernuma=0".Reported-by: Changrong Chen Suggested-by: Ying Huang Suggested-by: Robin Murphy Signed-off-by: Feng Tang Link: https://lore.kernel.org/r/20260512085509.83002-1-feng.tang@linux.alibaba.comLink: https://lore.kernel.org/all/20260520222742.GA1607511@ax162/[mszyprow: squashed changes from both links, added __initdata attribute to the numa_cma_configured variable]Signed-off-by: Marek Szyprowski List of files: /linux/kernel/dma/Kconfig

d7eafe655b741dfc241d5b920f6d2cea45b568d9 - dma-mapping: Separate DMA sync issuing and completion waiting

Sat, 28 Feb 2026 23:13:16 +0100

dma-mapping: Separate DMA sync issuing and completion waitingCurrently, arch_sync_dma_for_cpu and arch_sync_dma_for_devicealways wait for the completion of each DMA buffer. That is,issuing the DMA sync and waiting for completion is done in asingle API call.For scatter-gather lists with multiple entries, this meansissuing and waiting is repeated for each entry, which can hurtperformance. Architectures like ARM64 may be able to issue allDMA sync operations for all entries first and then wait forcompletion together.To address this, arch_sync_dma_for_* now batches DMA operationsand performs a flush afterward. On ARM64, the flush is implementedwith a dsb instruction in arch_sync_dma_flush(). On otherarchitectures, arch_sync_dma_flush() is currently a nop.Cc: Leon Romanovsky Cc: Catalin Marinas Cc: Will Deacon Cc: Marek Szyprowski Cc: Robin Murphy Cc: Ada Couprie Diaz Cc: Ard Biesheuvel Cc: Marc Zyngier Cc: Anshuman Khandual Cc: Ryan Roberts Cc: Suren Baghdasaryan Cc: Joerg Roedel Cc: Stefano Stabellini Cc: Oleksandr Tyshchenko Cc: Tangquan Zheng Reviewed-by: Juergen Gross # drivers/xen/swiotlb-xen.cTested-by: Xueyuan Chen Signed-off-by: Barry Song Reviewed-by: Leon Romanovsky Signed-off-by: Marek Szyprowski Link: https://lore.kernel.org/r/20260228221316.59934-1-21cnbao@gmail.com List of files: /linux/kernel/dma/Kconfig

8a840ab0567ff2b7d382694ba24a58a893d2c7af - dma-mapping: Remove dma_mark_clean (again)

Tue, 06 Jan 2026 20:27:53 +0100

dma-mapping: Remove dma_mark_clean (again)With IA-64 now gone, there are no users of the dma_mark_clean hook,so we can retire it for good.Signed-off-by: Robin Murphy Signed-off-by: Marek Szyprowski Link: https://lore.kernel.org/r/c004927f01962726ff1dcf94d1b4efff84db805a.1767727673.git.robin.murphy@arm.com List of files: /linux/kernel/dma/Kconfig

150745b49aca4dec8057e8908d5ce5383e036a4f - dma-debug: remove DMA_API_DEBUG_SG

Tue, 15 Oct 2024 10:29:20 +0200

dma-debug: remove DMA_API_DEBUG_SGThe scatterlist validity checks are pretty simple and cheap, perform themunconditionally.Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

de6c85bf918ea52d5c680f0d130b37ee2ff152d6 - dma-mapping: clearly mark DMA ops as an architecture feature

Wed, 28 Aug 2024 08:02:47 +0200

dma-mapping: clearly mark DMA ops as an architecture featureDMA ops are a helper for architectures and not for drivers to overridethe DMA implementation.Unfortunately driver authors keep ignoring this. Make the fact moreclear by renaming the symbol to ARCH_HAS_DMA_OPS and having the two driversoverriding their dma_ops depend on that. These drivers should probably bemarked broken, but we can give them a bit of a grace period for that.Signed-off-by: Christoph Hellwig Reviewed-by: Thomas Gleixner Acked-by: Sakari Ailus # for IPU6Acked-by: Robin Murphy List of files: /linux/kernel/dma/Kconfig

b5c58b2fdc427e7958412ecb2de2804a1f7c1572 - dma-mapping: direct calls for dma-iommu

Wed, 24 Jul 2024 20:04:49 +0200

dma-mapping: direct calls for dma-iommuDirectly call into dma-iommu just like we have been doing for dma-directfor a while. This avoids the indirect call overhead for IOMMU ops andremoves the need to have DMA ops entirely for many common configurations.Signed-off-by: Leon Romanovsky Signed-off-by: Leon Romanovsky Acked-by: Greg Kroah-Hartman Acked-by: Robin Murphy Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

fe7514b149e0a8a6f3031d286e52d40163b0b11a - dma: compile-out DMA sync op calls when not used

Tue, 07 May 2024 13:20:20 +0200

dma: compile-out DMA sync op calls when not usedSome platforms do have DMA, but DMA there is always direct and coherent.Currently, even on such platforms DMA sync operations are compiled andcalled.Add a new hidden Kconfig symbol, DMA_NEED_SYNC, and set it only wheneither sync operations are needed or there is DMA ops or swiotlbor DMA debug is enabled. Compile global dma_sync_*() and dma_need_sync()only when it's set, otherwise provide empty inline stubs.The change allows for future optimizations of DMA sync calls dependingon runtime conditions.Signed-off-by: Alexander Lobakin Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

2c8ed1b960fb97c82ede5afc974329bfdb457f5f - dma-direct: add a CONFIG_ARCH_HAS_DMA_ALLOC symbol

Thu, 05 Oct 2023 09:05:36 +0200

dma-direct: add a CONFIG_ARCH_HAS_DMA_ALLOC symbolInstead of using arch_dma_alloc if none of the generic coherentallocators are used, require the architectures to explicitly opt intoproviding it. This will used to deal with the case of m68knommu andcoldfire where we can't do any coherent allocations whatsoever, andalso makes it clear that arch_dma_alloc is a last resort.Signed-off-by: Christoph Hellwig Reviewed-by: Robin Murphy Reviewed-by: Greg Ungerer Tested-by: Greg Ungerer List of files: /linux/kernel/dma/Kconfig

da323d4640704001f2287f729124e1cd9d5684d0 - dma-direct: add dependencies to CONFIG_DMA_GLOBAL_POOL

Thu, 05 Oct 2023 09:06:55 +0200

dma-direct: add dependencies to CONFIG_DMA_GLOBAL_POOLCONFIG_DMA_GLOBAL_POOL can't be combined with other DMA coherentallocators. Add dependencies to Kconfig to document this, and makekconfig complain about unmet dependencies if someone tries.Signed-off-by: Christoph Hellwig Reviewed-by: Robin Murphy Reviewed-by: Lad Prabhakar Reviewed-by: Greg Ungerer Tested-by: Greg Ungerer List of files: /linux/kernel/dma/Kconfig

2dcdf8c18d5c1835571bfa40f40ac134c8a1f0f5 - dma-contiguous: fix the Kconfig entry for CONFIG_DMA_NUMA_CMA

Wed, 30 Aug 2023 11:25:25 +0200

dma-contiguous: fix the Kconfig entry for CONFIG_DMA_NUMA_CMAIt makes no sense to expose CONFIG_DMA_NUMA_CMA if CONFIG_NUMA is notenabled, and random config options shouldn't be default unless thereis a good reason. Replace the default NUMA with a depends on to fix bothissues.Signed-off-by: Christoph Hellwig Reviewed-by: Robin Murphy List of files: /linux/kernel/dma/Kconfig

62708b2ba4055cad43a95754e939566b56dde5b6 - swiotlb: add a flag whether SWIOTLB is allowed to grow

Tue, 01 Aug 2023 08:24:00 +0200

swiotlb: add a flag whether SWIOTLB is allowed to growAdd a config option (CONFIG_SWIOTLB_DYNAMIC) to enable or disable dynamicallocation of additional bounce buffers.If this option is set, mark the default SWIOTLB as able to grow andrestricted DMA pools as unable.However, if the address of the default memory pool is explicitly queried,make the default SWIOTLB also unable to grow. This is currently used to setup PCI BAR movable regions on some Octeon MIPS boards which may not be ableto use a SWIOTLB pool elsewhere in physical memory. See octeon_pci_setup()for more details.If a remap function is specified, it must be also called on any dynamicallyallocated pools, but there are some issues:- The remap function may block, so it should not be called from an atomic context.- There is no corresponding unremap() function if the memory pool is freed.- The only in-tree implementation (xen_swiotlb_fixup) requires that the number of slots in the memory pool is a multiple of SWIOTLB_SEGSIZE.Keep it simple for now and disable growing the SWIOTLB if a remap functionwas specified.Signed-off-by: Petr Tesarik Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

bf29bfaa54901a4bdee2a18cd10eb951a884a5f9 - dma-contiguous: support numa CMA for specified node

Wed, 12 Jul 2023 09:47:58 +0200

dma-contiguous: support numa CMA for specified nodeThe kernel parameter 'cma_pernuma=' only supports reserving the samesize of CMA area for each node. We need to reserve different sizes ofCMA area for specified nodes if these devices belong to different nodes.Adding another kernel parameter 'numa_cma=' to reserve CMA area forthe specified node. If we want to use one of these parameters, we need toenable DMA_NUMA_CMA.At the same time, print the node id in cma_declare_contiguous_nid() ifCONFIG_NUMA is enabled.Signed-off-by: Yajun Deng Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

22e4a348f87c59df2c02f1efb7ba9a56b622c7b8 - dma-contiguous: support per-numa CMA for all architectures

Fri, 12 May 2023 11:42:10 +0200

dma-contiguous: support per-numa CMA for all architecturesIn the commit b7176c261cdb ("dma-contiguous: provide the ability toreserve per-numa CMA"), Barry adds DMA_PERNUMA_CMA for ARM64.But this feature is architecture independent, so support per-numa CMAfor all architectures, and enable it by default if NUMA.Signed-off-by: Yajun Deng Tested-by: Yicong Yang Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

1e6d5dea34325df8dc204575cd0726cd5f2b864f - Merge tag 'dma-mapping-6.5-2023-06-28' of git://git.infradead.org/users/hch/dma-mapping

Fri, 30 Jun 2023 06:12:20 +0200

Merge tag 'dma-mapping-6.5-2023-06-28' of git://git.infradead.org/users/hch/dma-mappingPull dma-mapping updates from Christoph Hellwig: - swiotlb cleanups (Petr Tesarik) - use kvmalloc_array (gaoxu) - a small step towards removing is_swiotlb_active (Christoph Hellwig) - fix a Kconfig typo Sui Jingfeng)* tag 'dma-mapping-6.5-2023-06-28' of git://git.infradead.org/users/hch/dma-mapping: drm/nouveau: stop using is_swiotlb_active swiotlb: use the atomic counter of total used slabs if available swiotlb: remove unused field "used" from struct io_tlb_mem dma-remap: use kvmalloc_array/kvfree for larger dma memory remap dma-mapping: fix a Kconfig typo List of files: /linux/kernel/dma/Kconfig

370645f41e6e2fdd2fb6f6982530b04612c9793c - dma-mapping: force bouncing if the kmalloc() size is not cache-line-aligned

Mon, 12 Jun 2023 17:31:58 +0200

dma-mapping: force bouncing if the kmalloc() size is not cache-line-alignedFor direct DMA, if the size is small enough to have originated from akmalloc() cache below ARCH_DMA_MINALIGN, check its alignment againstdma_get_cache_alignment() and bounce if necessary. For larger sizes, itis the responsibility of the DMA API caller to ensure proper alignment.At this point, the kmalloc() caches are properly aligned but this willchange in a subsequent patch.Architectures can opt in by selecting DMA_BOUNCE_UNALIGNED_KMALLOC.Link: https://lkml.kernel.org/r/20230612153201.554742-15-catalin.marinas@arm.comSigned-off-by: Catalin Marinas Reviewed-by: Christoph Hellwig Reviewed-by: Robin Murphy Tested-by: Isaac J. Manjarres Cc: Alasdair Kergon Cc: Ard Biesheuvel Cc: Arnd Bergmann Cc: Daniel Vetter Cc: Greg Kroah-Hartman Cc: Herbert Xu Cc: Jerry Snitselaar Cc: Joerg Roedel Cc: Jonathan Cameron Cc: Jonathan Cameron Cc: Lars-Peter Clausen Cc: Logan Gunthorpe Cc: Marc Zyngier Cc: Mark Brown Cc: Mike Snitzer Cc: "Rafael J. Wysocki" Cc: Saravana Kannan Cc: Vlastimil Babka Cc: Will Deacon Signed-off-by: Andrew Morton List of files: /linux/kernel/dma/Kconfig

af2880ec44021d32cc72a5aa7c5d7d7beaa722d3 - scatterlist: add dedicated config for DMA flags

Mon, 12 Jun 2023 17:31:56 +0200

scatterlist: add dedicated config for DMA flagsThe DMA flags field will be useful for users beyond PCI P2P, so upgrade toits own dedicated config option.[catalin.marinas@arm.com: use #ifdef CONFIG_NEED_SG_DMA_FLAGS in scatterlist.h][catalin.marinas@arm.com: update PCI_P2PDMA dma_flags comment in scatterlist.h]Link: https://lkml.kernel.org/r/20230612153201.554742-13-catalin.marinas@arm.comSigned-off-by: Robin Murphy Signed-off-by: Catalin Marinas Reviewed-by: Christoph Hellwig Tested-by: Isaac J. Manjarres Cc: Alasdair Kergon Cc: Ard Biesheuvel Cc: Arnd Bergmann Cc: Daniel Vetter Cc: Greg Kroah-Hartman Cc: Herbert Xu Cc: Jerry Snitselaar Cc: Joerg Roedel Cc: Jonathan Cameron Cc: Jonathan Cameron Cc: Lars-Peter Clausen Cc: Logan Gunthorpe Cc: Marc Zyngier Cc: Mark Brown Cc: Mike Snitzer Cc: "Rafael J. Wysocki" Cc: Saravana Kannan Cc: Vlastimil Babka Cc: Will Deacon Signed-off-by: Andrew Morton List of files: /linux/kernel/dma/Kconfig

4d3af20eaf3fcd481a797738bb46634e37f4a1cc - dma-mapping: fix a Kconfig typo

Sun, 04 Jun 2023 15:50:54 +0200

dma-mapping: fix a Kconfig typo'thing' -> 'think'Signed-off-by: Sui Jingfeng Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

1d3f56b295302fdb4ac9caf6ce09f5ae7d2e651a - dma-mapping: provide CONFIG_ARCH_DMA_DEFAULT_COHERENT

Sat, 01 Apr 2023 11:15:30 +0200

dma-mapping: provide CONFIG_ARCH_DMA_DEFAULT_COHERENTProvide a kconfig option to allow arches to manipulate defaultvalue of dma_default_coherent in Kconfig.Signed-off-by: Jiaxun Yang Signed-off-by: Christoph Hellwig List of files: /linux/kernel/dma/Kconfig

f5ff79fddf0efecca538046b5cc20fb3ded2ec4f - dma-mapping: remove CONFIG_DMA_REMAP

Sat, 26 Feb 2022 16:40:21 +0100

dma-mapping: remove CONFIG_DMA_REMAPCONFIG_DMA_REMAP is used to build a few helpers around the corevmalloc code, and to use them in case there is a highmem page indma-direct, and to make dma coherent allocations be able to usenon-contiguous pages allocations for DMA allocations in the dma-iommulayer.Right now it needs to be explicitly selected by architectures, andis only done so by architectures that require remapping to dealwith devices that are not DMA coherent. Make it unconditional forbuilds with CONFIG_MMU as it is very little extra code, but makesit much more likely that large DMA allocations succeed on x86.This fixes hot plugging a NVMe thunderbolt SSD for me, which triesto allocate a 1MB buffer that is otherwise hard to obtain due tomemory fragmentation on a heavily used laptop.Signed-off-by: Christoph Hellwig Reviewed-by: Robin Murphy List of files: /linux/kernel/dma/Kconfig