kernel/linux.git/mm/init-mm.c, branch v6.18.22

mm: replace vm_lock and detached flag with a reference count

2025-03-17T05:06:20+00:00

rw_semaphore is a sizable structure of 40 bytes and consumes considerable space for each vm_area_struct. However vma_lock has two important specifics which can be used to replace rw_semaphore with a simpler structure: 1. Readers never wait. They try to take the vma_lock and fall back to mmap_lock if that fails. 2. Only one writer at a time will ever try to write-lock a vma_lock because writers first take mmap_lock in write mode. Because of these requirements, full rw_semaphore functionality is not needed and we can replace rw_semaphore and the vma->detached flag with a refcount (vm_refcnt). When vma is in detached state, vm_refcnt is 0 and only a call to vma_mark_attached() can take it out of this state. Note that unlike before, now we enforce both vma_mark_attached() and vma_mark_detached() to be done only after vma has been write-locked. vma_mark_attached() changes vm_refcnt to 1 to indicate that it has been attached to the vma tree. When a reader takes read lock, it increments vm_refcnt, unless the top usable bit of vm_refcnt (0x40000000) is set, indicating presence of a writer. When writer takes write lock, it sets the top usable bit to indicate its presence. If there are readers, writer will wait using newly introduced mm->vma_writer_wait. Since all writers take mmap_lock in write mode first, there can be only one writer at a time. The last reader to release the lock will signal the writer to wake up. refcount might overflow if there are many competing readers, in which case read-locking will fail. Readers are expected to handle such failures. In summary: 1. all readers increment the vm_refcnt; 2. writer sets top usable (writer) bit of vm_refcnt; 3. readers cannot increment the vm_refcnt if the writer bit is set; 4. in the presence of readers, writer must wait for the vm_refcnt to drop to 1 (plus the VMA_LOCK_OFFSET writer bit), indicating an attached vma with no readers; 5. vm_refcnt overflow is handled by the readers. While this vm_lock replacement does not yet result in a smaller vm_area_struct (it stays at 256 bytes due to cacheline alignment), it allows for further size optimization by structure member regrouping to bring the size of vm_area_struct below 192 bytes. [surenb@google.com: fix a crash due to vma_end_read() that should have been removed] Link: https://lkml.kernel.org/r/20250220200208.323769-1-surenb@google.com Link: https://lkml.kernel.org/r/20250213224655.1680278-13-surenb@google.com Signed-off-by: Suren Baghdasaryan Suggested-by: Peter Zijlstra Suggested-by: Matthew Wilcox Tested-by: Shivank Garg Link: https://lkml.kernel.org/r/5e19ec93-8307-47c2-bb13-3ddf7150624e@amd.com Reviewed-by: Vlastimil Babka Cc: Christian Brauner Cc: David Hildenbrand Cc: David Howells Cc: Davidlohr Bueso Cc: Hugh Dickins Cc: Jann Horn Cc: Johannes Weiner Cc: Jonathan Corbet Cc: Klara Modin Cc: Liam R. Howlett Cc: Lokesh Gidra Cc: Lorenzo Stoakes Cc: Mateusz Guzik Cc: Mel Gorman Cc: Michal Hocko Cc: Minchan Kim Cc: Oleg Nesterov Cc: Pasha Tatashin Cc: "Paul E . McKenney" Cc: Peter Xu Cc: Shakeel Butt Cc: Sourav Panda Cc: Wei Yang Cc: Will Deacon Cc: Heiko Carstens Cc: Stephen Rothwell Signed-off-by: Andrew Morton

mm: convert mm_lock_seq to a proper seqcount

2024-12-02T11:01:38+00:00

Convert mm_lock_seq to be seqcount_t and change all mmap_write_lock variants to increment it, in-line with the usual seqcount usage pattern. This lets us check whether the mmap_lock is write-locked by checking mm_lock_seq.sequence counter (odd=locked, even=unlocked). This will be used when implementing mmap_lock speculation functions. As a result vm_lock_seq is also change to be unsigned to match the type of mm_lock_seq.sequence. Suggested-by: Peter Zijlstra Signed-off-by: Suren Baghdasaryan Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Liam R. Howlett Link: https://lkml.kernel.org/r/20241122174416.1367052-2-surenb@google.com

mm: Deprecate pasid field

2023-12-12T09:11:32+00:00

Drop the pasid field, as all the information needed for sva domain management has been moved to the newly added iommu_mm field. Reviewed-by: Lu Baolu Reviewed-by: Vasant Hegde Reviewed-by: Jason Gunthorpe Signed-off-by: Tina Zhang Signed-off-by: Jason Gunthorpe Link: https://lore.kernel.org/r/20231027000525.1278806-7-tina.zhang@intel.com Signed-off-by: Joerg Roedel

iommu: Change kconfig around IOMMU_SVA

2023-12-12T09:11:27+00:00

Linus suggested that the kconfig here is confusing: https://lore.kernel.org/all/CAHk-=wgUiAtiszwseM1p2fCJ+sC4XWQ+YN4TanFhUgvUqjr9Xw@mail.gmail.com/ Let's break it into three kconfigs controlling distinct things: - CONFIG_IOMMU_MM_DATA controls if the mm_struct has the additional fields for the IOMMU. Currently only PASID, but later patches store a struct iommu_mm_data * - CONFIG_ARCH_HAS_CPU_PASID controls if the arch needs the scheduling bit for keeping track of the ENQCMD instruction. x86 will select this if IOMMU_SVA is enabled - IOMMU_SVA controls if the IOMMU core compiles in the SVA support code for iommu driver use and the IOMMU exported API This way ARM will not enable CONFIG_ARCH_HAS_CPU_PASID Signed-off-by: Jason Gunthorpe Link: https://lore.kernel.org/r/20231027000525.1278806-2-tina.zhang@intel.com Signed-off-by: Joerg Roedel

mm: move dummy_vm_ops out of a header

2023-08-21T20:37:46+00:00

Otherwise the kernel ends up with multiple copies: $ nm vmlinux | grep dummy_vm_ops ffffffff81e4ea00 d dummy_vm_ops.2 ffffffff81e11760 d dummy_vm_ops.254 ffffffff81e406e0 d dummy_vm_ops.4 ffffffff81e3c780 d dummy_vm_ops.7 While here prefix it with vma_. Link: https://lkml.kernel.org/r/20230806231611.1395735-1-mjguzik@gmail.com Signed-off-by: Mateusz Guzik Cc: Matthew Wilcox Signed-off-by: Andrew Morton

Merge tag 'iommu-updates-v6.4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

2023-04-30T20:00:38+00:00

Pull iommu updates from Joerg Roedel: - Convert to platform remove callback returning void - Extend changing default domain to normal group - Intel VT-d updates: - Remove VT-d virtual command interface and IOASID - Allow the VT-d driver to support non-PRI IOPF - Remove PASID supervisor request support - Various small and misc cleanups - ARM SMMU updates: - Device-tree binding updates: * Allow Qualcomm GPU SMMUs to accept relevant clock properties * Document Qualcomm 8550 SoC as implementing an MMU-500 * Favour new "qcom,smmu-500" binding for Adreno SMMUs - Fix S2CR quirk detection on non-architectural Qualcomm SMMU implementations - Acknowledge SMMUv3 PRI queue overflow when consuming events - Document (in a comment) why ATS is disabled for bypass streams - AMD IOMMU updates: - 5-level page-table support - NUMA awareness for memory allocations - Unisoc driver: Support for reattaching an existing domain - Rockchip driver: Add missing set_platform_dma_ops callback - Mediatek driver: Adjust the dma-ranges - Various other small fixes and cleanups * tag 'iommu-updates-v6.4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (82 commits) iommu: Remove iommu_group_get_by_id() iommu: Make iommu_release_device() static iommu/vt-d: Remove BUG_ON in dmar_insert_dev_scope() iommu/vt-d: Remove a useless BUG_ON(dev->is_virtfn) iommu/vt-d: Remove BUG_ON in map/unmap() iommu/vt-d: Remove BUG_ON when domain->pgd is NULL iommu/vt-d: Remove BUG_ON in handling iotlb cache invalidation iommu/vt-d: Remove BUG_ON on checking valid pfn range iommu/vt-d: Make size of operands same in bitwise operations iommu/vt-d: Remove PASID supervisor request support iommu/vt-d: Use non-privileged mode for all PASIDs iommu/vt-d: Remove extern from function prototypes iommu/vt-d: Do not use GFP_ATOMIC when not needed iommu/vt-d: Remove unnecessary checks in iopf disabling path iommu/vt-d: Move PRI handling to IOPF feature path iommu/vt-d: Move pfsid and ats_qdep calculation to device probe path iommu/vt-d: Move iopf code from SVA to IOPF enabling path iommu/vt-d: Allow SVA with device-specific IOPF dmaengine: idxd: Add enable/disable device IOPF feature arm64: dts: mt8186: Add dma-ranges for the parent "soc" node ...

mm: add per-VMA lock and helper functions to control it

2023-04-06T03:02:57+00:00

Introduce per-VMA locking. The lock implementation relies on a per-vma and per-mm sequence counters to note exclusive locking: - read lock - (implemented by vma_start_read) requires the vma (vm_lock_seq) and mm (mm_lock_seq) sequence counters to differ. If they match then there must be a vma exclusive lock held somewhere. - read unlock - (implemented by vma_end_read) is a trivial vma->lock unlock. - write lock - (vma_start_write) requires the mmap_lock to be held exclusively and the current mm counter is assigned to the vma counter. This will allow multiple vmas to be locked under a single mmap_lock write lock (e.g. during vma merging). The vma counter is modified under exclusive vma lock. - write unlock - (vma_end_write_all) is a batch release of all vma locks held. It doesn't pair with a specific vma_start_write! It is done before exclusive mmap_lock is released by incrementing mm sequence counter (mm_lock_seq). - write downgrade - if the mmap_lock is downgraded to the read lock, all vma write locks are released as well (effectivelly same as write unlock). Link: https://lkml.kernel.org/r/20230227173632.3292573-13-surenb@google.com Signed-off-by: Suren Baghdasaryan Signed-off-by: Andrew Morton

iommu/ioasid: Rename INVALID_IOASID

2023-03-31T08:03:27+00:00

INVALID_IOASID and IOMMU_PASID_INVALID are duplicated. Rename INVALID_IOASID and consolidate since we are moving away from IOASID infrastructure. Reviewed-by: Dave Jiang Reviewed-by: Jason Gunthorpe Reviewed-by: Kevin Tian Reviewed-by: Lu Baolu Signed-off-by: Jacob Pan Link: https://lore.kernel.org/r/20230322200803.869130-7-jacob.jun.pan@linux.intel.com Signed-off-by: Joerg Roedel

mm: remove rb tree.

2022-09-27T02:46:16+00:00

Remove the RB tree and start using the maple tree for vm_area_struct tracking. Drop validate_mm() calls in expand_upwards() and expand_downwards() as the lock is not held. Link: https://lkml.kernel.org/r/20220906194824.2110408-18-Liam.Howlett@oracle.com Signed-off-by: Liam R. Howlett Tested-by: Yu Zhao Cc: Catalin Marinas Cc: David Hildenbrand Cc: David Howells Cc: Davidlohr Bueso Cc: "Matthew Wilcox (Oracle)" Cc: SeongJae Park Cc: Sven Schnelle Cc: Vlastimil Babka Cc: Will Deacon Signed-off-by: Andrew Morton

mm: start tracking VMAs with maple tree

2022-09-27T02:46:14+00:00

Start tracking the VMAs with the new maple tree structure in parallel with the rb_tree. Add debug and trace events for maple tree operations and duplicate the rb_tree that is created on forks into the maple tree. The maple tree is added to the mm_struct including the mm_init struct, added support in required mm/mmap functions, added tracking in kernel/fork for process forking, and used to find the unmapped_area and checked against what the rbtree finds. This also moves the mmap_lock() in exit_mmap() since the oom reaper call does walk the VMAs. Otherwise lockdep will be unhappy if oom happens. When splitting a vma fails due to allocations of the maple tree nodes, the error path in __split_vma() calls new->vm_ops->close(new). The page accounting for hugetlb is actually in the close() operation, so it accounts for the removal of 1/2 of the VMA which was not adjusted. This results in a negative exit value. To avoid the negative charge, set vm_start = vm_end and vm_pgoff = 0. There is also a potential accounting issue in special mappings from insert_vm_struct() failing to allocate, so reverse the charge there in the failure scenario. Link: https://lkml.kernel.org/r/20220906194824.2110408-9-Liam.Howlett@oracle.com Signed-off-by: Liam R. Howlett Signed-off-by: Matthew Wilcox (Oracle) Tested-by: Yu Zhao Cc: Catalin Marinas Cc: David Hildenbrand Cc: David Howells Cc: Davidlohr Bueso Cc: SeongJae Park Cc: Sven Schnelle Cc: Vlastimil Babka Cc: Will Deacon Signed-off-by: Andrew Morton