kernel/linux.git/mm/hmm.c, branch v5.16

mm/hmm: bypass devmap pte when all pfn requested flags are fulfilled

2021-09-09T01:45:52+00:00

Previously, we noticed the one rpma example was failed[1] since commit 36f30e486dce ("IB/core: Improve ODP to use hmm_range_fault()"), where it will use ODP feature to do RDMA WRITE between fsdax files. After digging into the code, we found hmm_vma_handle_pte() will still return EFAULT even though all the its requesting flags has been fulfilled. That's because a DAX page will be marked as (_PAGE_SPECIAL | PAGE_DEVMAP) by pte_mkdevmap(). Link: https://github.com/pmem/rpma/issues/1142 [1] Link: https://lkml.kernel.org/r/20210830094232.203029-1-lizhijian@cn.fujitsu.com Fixes: 405506274922 ("mm/hmm: add missing call to hmm_pte_need_fault in HMM_PFN_SPECIAL handling") Signed-off-by: Li Zhijian Reviewed-by: Christoph Hellwig Reviewed-by: Jason Gunthorpe Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: device exclusive memory access

2021-07-01T18:06:03+00:00

Some devices require exclusive write access to shared virtual memory (SVM) ranges to perform atomic operations on that memory. This requires CPU page tables to be updated to deny access whilst atomic operations are occurring. In order to do this introduce a new swap entry type (SWP_DEVICE_EXCLUSIVE). When a SVM range needs to be marked for exclusive access by a device all page table mappings for the particular range are replaced with device exclusive swap entries. This causes any CPU access to the page to result in a fault. Faults are resovled by replacing the faulting entry with the original mapping. This results in MMU notifiers being called which a driver uses to update access permissions such as revoking atomic access. After notifiers have been called the device will no longer have exclusive access to the region. Walking of the page tables to find the target pages is handled by get_user_pages() rather than a direct page table walk. A direct page table walk similar to what migrate_vma_collect()/unmap() does could also have been utilised. However this resulted in more code similar in functionality to what get_user_pages() provides as page faulting is required to make the PTEs present and to break COW. [dan.carpenter@oracle.com: fix signedness bug in make_device_exclusive_range()] Link: https://lkml.kernel.org/r/YNIz5NVnZ5GiZ3u1@mwanda Link: https://lkml.kernel.org/r/20210616105937.23201-8-apopple@nvidia.com Signed-off-by: Alistair Popple Signed-off-by: Dan Carpenter Reviewed-by: Christoph Hellwig Cc: Ben Skeggs Cc: Hugh Dickins Cc: Jason Gunthorpe Cc: John Hubbard Cc: "Matthew Wilcox (Oracle)" Cc: Peter Xu Cc: Ralph Campbell Cc: Shakeel Butt Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm/swapops: rework swap entry manipulation code

2021-07-01T18:06:03+00:00

Both migration and device private pages use special swap entries that are manipluated by a range of inline functions. The arguments to these are somewhat inconsistent so rework them to remove flag type arguments and to make the arguments similar for both read and write entry creation. Link: https://lkml.kernel.org/r/20210616105937.23201-3-apopple@nvidia.com Signed-off-by: Alistair Popple Reviewed-by: Christoph Hellwig Reviewed-by: Jason Gunthorpe Reviewed-by: Ralph Campbell Cc: Ben Skeggs Cc: Hugh Dickins Cc: John Hubbard Cc: "Matthew Wilcox (Oracle)" Cc: Peter Xu Cc: Shakeel Butt Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: remove special swap entry functions

2021-07-01T18:06:03+00:00

Patch series "Add support for SVM atomics in Nouveau", v11. Introduction ============ Some devices have features such as atomic PTE bits that can be used to implement atomic access to system memory. To support atomic operations to a shared virtual memory page such a device needs access to that page which is exclusive of the CPU. This series introduces a mechanism to temporarily unmap pages granting exclusive access to a device. These changes are required to support OpenCL atomic operations in Nouveau to shared virtual memory (SVM) regions allocated with the CL_MEM_SVM_ATOMICS clSVMAlloc flag. A more complete description of the OpenCL SVM feature is available at https://www.khronos.org/registry/OpenCL/specs/3.0-unified/html/ OpenCL_API.html#_shared_virtual_memory . Implementation ============== Exclusive device access is implemented by adding a new swap entry type (SWAP_DEVICE_EXCLUSIVE) which is similar to a migration entry. The main difference is that on fault the original entry is immediately restored by the fault handler instead of waiting. Restoring the entry triggers calls to MMU notifers which allows a device driver to revoke the atomic access permission from the GPU prior to the CPU finalising the entry. Patches ======= Patches 1 & 2 refactor existing migration and device private entry functions. Patches 3 & 4 rework try_to_unmap_one() by splitting out unrelated functionality into separate functions - try_to_migrate_one() and try_to_munlock_one(). Patch 5 renames some existing code but does not introduce functionality. Patch 6 is a small clean-up to swap entry handling in copy_pte_range(). Patch 7 contains the bulk of the implementation for device exclusive memory. Patch 8 contains some additions to the HMM selftests to ensure everything works as expected. Patch 9 is a cleanup for the Nouveau SVM implementation. Patch 10 contains the implementation of atomic access for the Nouveau driver. Testing ======= This has been tested with upstream Mesa 21.1.0 and a simple OpenCL program which checks that GPU atomic accesses to system memory are atomic. Without this series the test fails as there is no way of write-protecting the page mapping which results in the device clobbering CPU writes. For reference the test is available at https://ozlabs.org/~apopple/opencl_svm_atomics/ Further testing has been performed by adding support for testing exclusive access to the hmm-tests kselftests. This patch (of 10): Remove multiple similar inline functions for dealing with different types of special swap entries. Both migration and device private swap entries use the swap offset to store a pfn. Instead of multiple inline functions to obtain a struct page for each swap entry type use a common function pfn_swap_entry_to_page(). Also open-code the various entry_to_pfn() functions as this results is shorter code that is easier to understand. Link: https://lkml.kernel.org/r/20210616105937.23201-1-apopple@nvidia.com Link: https://lkml.kernel.org/r/20210616105937.23201-2-apopple@nvidia.com Signed-off-by: Alistair Popple Reviewed-by: Ralph Campbell Reviewed-by: Christoph Hellwig Cc: "Matthew Wilcox (Oracle)" Cc: Hugh Dickins Cc: Peter Xu Cc: Shakeel Butt Cc: Ben Skeggs Cc: Jason Gunthorpe Cc: John Hubbard Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: do page fault accounting in handle_mm_fault

2020-08-12T17:58:02+00:00

Patch series "mm: Page fault accounting cleanups", v5. This is v5 of the pf accounting cleanup series. It originates from Gerald Schaefer's report on an issue a week ago regarding to incorrect page fault accountings for retried page fault after commit 4064b9827063 ("mm: allow VM_FAULT_RETRY for multiple times"): https://lore.kernel.org/lkml/20200610174811.44b94525@thinkpad/ What this series did: - Correct page fault accounting: we do accounting for a page fault (no matter whether it's from #PF handling, or gup, or anything else) only with the one that completed the fault. For example, page fault retries should not be counted in page fault counters. Same to the perf events. - Unify definition of PERF_COUNT_SW_PAGE_FAULTS: currently this perf event is used in an adhoc way across different archs. Case (1): for many archs it's done at the entry of a page fault handler, so that it will also cover e.g. errornous faults. Case (2): for some other archs, it is only accounted when the page fault is resolved successfully. Case (3): there're still quite some archs that have not enabled this perf event. Since this series will touch merely all the archs, we unify this perf event to always follow case (1), which is the one that makes most sense. And since we moved the accounting into handle_mm_fault, the other two MAJ/MIN perf events are well taken care of naturally. - Unify definition of "major faults": the definition of "major fault" is slightly changed when used in accounting (not VM_FAULT_MAJOR). More information in patch 1. - Always account the page fault onto the one that triggered the page fault. This does not matter much for #PF handlings, but mostly for gup. More information on this in patch 25. Patchset layout: Patch 1: Introduced the accounting in handle_mm_fault(), not enabled. Patch 2-23: Enable the new accounting for arch #PF handlers one by one. Patch 24: Enable the new accounting for the rest outliers (gup, iommu, etc.) Patch 25: Cleanup GUP task_struct pointer since it's not needed any more This patch (of 25): This is a preparation patch to move page fault accountings into the general code in handle_mm_fault(). This includes both the per task flt_maj/flt_min counters, and the major/minor page fault perf events. To do this, the pt_regs pointer is passed into handle_mm_fault(). PERF_COUNT_SW_PAGE_FAULTS should still be kept in per-arch page fault handlers. So far, all the pt_regs pointer that passed into handle_mm_fault() is NULL, which means this patch should have no intented functional change. Suggested-by: Linus Torvalds Signed-off-by: Peter Xu Signed-off-by: Andrew Morton Cc: Albert Ou Cc: Alexander Gordeev Cc: Andy Lutomirski Cc: Benjamin Herrenschmidt Cc: Borislav Petkov Cc: Brian Cain Cc: Catalin Marinas Cc: Christian Borntraeger Cc: Chris Zankel Cc: Dave Hansen Cc: David S. Miller Cc: Geert Uytterhoeven Cc: Gerald Schaefer Cc: Greentime Hu Cc: Guo Ren Cc: Heiko Carstens Cc: Helge Deller Cc: H. Peter Anvin Cc: Ingo Molnar Cc: Ivan Kokshaysky Cc: James E.J. Bottomley Cc: John Hubbard Cc: Jonas Bonn Cc: Ley Foon Tan Cc: "Luck, Tony" Cc: Matt Turner Cc: Max Filippov Cc: Michael Ellerman Cc: Michal Simek Cc: Nick Hu Cc: Palmer Dabbelt Cc: Paul Mackerras Cc: Paul Walmsley Cc: Pekka Enberg Cc: Peter Zijlstra Cc: Richard Henderson Cc: Rich Felker Cc: Russell King Cc: Stafford Horne Cc: Stefan Kristiansson Cc: Thomas Bogendoerfer Cc: Thomas Gleixner Cc: Vasily Gorbik Cc: Vincent Chen Cc: Vineet Gupta Cc: Will Deacon Cc: Yoshinori Sato Link: http://lkml.kernel.org/r/20200707225021.200906-1-peterx@redhat.com Link: http://lkml.kernel.org/r/20200707225021.200906-2-peterx@redhat.com Signed-off-by: Linus Torvalds

mm/hmm.c: delete duplicated word

2020-08-12T17:57:58+00:00

Drop the repeated word "pages". Signed-off-by: Randy Dunlap Signed-off-by: Andrew Morton Reviewed-by: Andrew Morton Reviewed-by: Zi Yan Link: http://lkml.kernel.org/r/20200801173822.14973-4-rdunlap@infradead.org Signed-off-by: Linus Torvalds

mm/hmm: provide the page mapping order in hmm_range_fault()

2020-07-10T19:24:28+00:00

hmm_range_fault() returns an array of page frame numbers and flags for how the pages are mapped in the requested process' page tables. The PFN can be used to get the struct page with hmm_pfn_to_page() and the page size order can be determined with compound_order(page). However, if the page is larger than order 0 (PAGE_SIZE), there is no indication that a compound page is mapped by the CPU using a larger page size. Without this information, the caller can't safely use a large device PTE to map the compound page because the CPU might be using smaller PTEs with different read/write permissions. Add a new function hmm_pfn_to_map_order() to return the mapping size order so that callers know the pages are being mapped with consistent permissions and a large device page table mapping can be used if one is available. This will allow devices to optimize mapping the page into HW by avoiding or batching work for huge pages. For instance the dma_map can be done with a high order directly. Link: https://lore.kernel.org/r/20200701225352.9649-3-rcampbell@nvidia.com Signed-off-by: Ralph Campbell Signed-off-by: Jason Gunthorpe

mmap locking API: add mmap_assert_locked() and mmap_assert_write_locked()

2020-06-09T16:39:14+00:00

Add new APIs to assert that mmap_sem is held. Using this instead of rwsem_is_locked and lockdep_assert_held[_write] makes the assertions more tolerant of future changes to the lock type. Signed-off-by: Michel Lespinasse Signed-off-by: Andrew Morton Reviewed-by: Vlastimil Babka Reviewed-by: Daniel Jordan Cc: Davidlohr Bueso Cc: David Rientjes Cc: Hugh Dickins Cc: Jason Gunthorpe Cc: Jerome Glisse Cc: John Hubbard Cc: Laurent Dufour Cc: Liam Howlett Cc: Matthew Wilcox Cc: Peter Zijlstra Cc: Ying Han Link: http://lkml.kernel.org/r/20200520052908.204642-10-walken@google.com Signed-off-by: Linus Torvalds

mm/hmm: remove the customizable pfn format from hmm_range_fault

2020-05-11T13:47:29+00:00

Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling Tested-by: Ralph Campbell Signed-off-by: Christoph Hellwig Reviewed-by: Christoph Hellwig Signed-off-by: Jason Gunthorpe

mm/hmm: remove HMM_PFN_SPECIAL

2020-05-11T13:47:29+00:00

This is just an alias for HMM_PFN_ERROR, nothing cares that the error was because of a special page vs any other error case. Link: https://lore.kernel.org/r/4-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling Reviewed-by: Christoph Hellwig Reviewed-by: John Hubbard Signed-off-by: Jason Gunthorpe