- Kernel - mailweb.openeuler.org

[PATCH OLK-5.10] erofs: Fix erofs_bmap error
by Zizhi Wo 13 Nov '25

13 Nov '25

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/ID6FYU -------------------------------- Currently, there is an issue with the bmap of erofs. When formatted as the chunk type, the bmap returns not the block granularity but the chunk granularity, which is inaccurate. Unify it with the bmap of other file systems, return the block granularity. Calculate the offset within the chunk to solve this problem. Fixes: 9da681e017a3 ("staging: erofs: support bmap") Signed-off-by: Zizhi Wo <wozizhi(a)huawei.com> --- fs/erofs/data.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/erofs/data.c b/fs/erofs/data.c index e228a2d52aaf..99c5da8e6338 100644 --- a/fs/erofs/data.c +++ b/fs/erofs/data.c @@ -458,6 +458,7 @@ static sector_t erofs_bmap(struct address_space *mapping, sector_t block) struct erofs_map_blocks map = { .m_la = blknr_to_addr(block), }; + erofs_off_t offset = map.m_la; if (EROFS_I(inode)->datalayout == EROFS_INODE_FLAT_INLINE) { erofs_blk_t blks = i_size_read(inode) >> LOG_BLOCK_SIZE; @@ -467,7 +468,7 @@ static sector_t erofs_bmap(struct address_space *mapping, sector_t block) } if (!erofs_map_blocks(inode, &map)) - return erofs_blknr(map.m_pa); + return erofs_blknr(map.m_pa + offset - map.m_la); return 0; } -- 2.39.2

2 1

[PATCH OLK-6.6 00/15] mm: add huge pfnmap support for remap_pfn_range()
by Yin Tirui 13 Nov '25

13 Nov '25

Overview ======== This patch series adds huge page support for remap_pfn_range(), automatically creating huge mappings when prerequisites are satisfied (size, alignment, architecture support, etc.) and falling back to normal page mappings otherwise. This work builds on Peter Xu's previous efforts on huge pfnmap support [0]. TODO ==== - Add PUD-level huge page support. Currently, only PMD-level huge pages are supported. - Consider the logic related to vmap_page_range and extract reusable common code. Tests Done ========== - Cross-build tests. - Performance tests with custom device driver implementing mmap() with remap_pfn_range(): - lat_mem_rd benchmark modified to use mmap(device_fd) instead of malloc() shows around 40% improvement in memory access latency with huge page support compared to normal page mappings. numactl -C 0 lat_mem_rd -t 4096M (stride=64) Memory Size (MB) Without Huge Mapping With Huge Mapping Improvement ---------------- ----------------- -------------- ----------- 64.00 148.858 ns 100.780 ns 32.3% 128.00 164.745 ns 103.537 ns 37.2% 256.00 169.907 ns 103.179 ns 39.3% 512.00 171.285 ns 103.072 ns 39.8% 1024.00 173.054 ns 103.055 ns 40.4% 2048.00 172.820 ns 103.091 ns 40.3% 4096.00 172.877 ns 103.115 ns 40.4% - Custom memory copy operations on mmap(device_fd) show around 18% performance improvement with huge page support compared to normal page mappings. numactl -C 0 memcpy_test (memory copy performance test) Memory Size (MB) Without Huge Mapping With Huge Mapping Improvement ---------------- ----------------- -------------- ----------- 1024.00 95.76 ms 77.91 ms 18.6% 2048.00 190.87 ms 155.64 ms 18.5% 4096.00 380.84 ms 311.45 ms 18.2% [0] https://lore.kernel.org/all/20240826204353.2228736-2-peterx@redhat.com/T/#u David Hildenbrand (1): mm/huge_memory: check pmd_special() only after pmd_present() Peter Xu (11): mm: introduce ARCH_SUPPORTS_HUGE_PFNMAP and special bits to pmd/pud mm: drop is_huge_zero_pud() mm: mark special bits for huge pfn mappings when inject mm: allow THP orders for PFNMAPs mm/gup: detect huge pfnmap entries in gup-fast mm/pagewalk: check pfnmap for folio_walk_start() mm/fork: accept huge pfnmap entries mm: always define pxx_pgprot() mm/x86: support large pfn mappings mm/arm64: support large pfn mappings arm64: mm: Drop dead code for pud special bit handling Yin Tirui (3): pgtable: add pte_clrhuge() implementation for arm64 mm: introduce remap_pfn_range_try_pmd() for PMD-level hugepage mapping arm64/x86: configs: Enable huge pfnmap support in openeuler defconfig arch/arm64/Kconfig | 1 + arch/arm64/configs/openeuler_defconfig | 2 + arch/arm64/include/asm/pgtable.h | 39 ++++++ arch/powerpc/include/asm/book3s/64/pgtable.h | 2 +- arch/powerpc/include/asm/pgtable.h | 1 + arch/s390/include/asm/pgtable.h | 1 + arch/sparc/include/asm/pgtable_64.h | 1 + arch/x86/Kconfig | 1 + arch/x86/configs/openeuler_defconfig | 2 + arch/x86/include/asm/pgtable.h | 81 +++++++----- fs/dax.c | 2 +- include/linux/huge_mm.h | 16 +-- include/linux/mm.h | 26 ++++ include/linux/pgtable.h | 20 ++- mm/Kconfig | 13 ++ mm/gup.c | 6 + mm/huge_memory.c | 93 ++++++++++---- mm/memory.c | 126 +++++++++++++++---- 18 files changed, 337 insertions(+), 96 deletions(-) -- 2.43.0

2 16

[PATCH OLK-6.6] KVM: SVM: Skip fastpath emulation on VM-Exit if next RIP isn't valid
by Pan Taixi 13 Nov '25

13 Nov '25

From: Sean Christopherson <seanjc(a)google.com> stable inclusion from stable-v6.6.113 commit cd3efb93677c4b0cf76348882fb429165fee33fd category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/ID3WHO CVE: CVE-2025-40038 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 0910dd7c9ad45a2605c45fd2bf3d1bcac087687c ] Skip the WRMSR and HLT fastpaths in SVM's VM-Exit handler if the next RIP isn't valid, e.g. because KVM is running with nrips=false. SVM must decode and emulate to skip the instruction if the CPU doesn't provide the next RIP, and getting the instruction bytes to decode requires reading guest memory. Reading guest memory through the emulator can fault, i.e. can sleep, which is disallowed since the fastpath handlers run with IRQs disabled. BUG: sleeping function called from invalid context at ./include/linux/uaccess.h:106 in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 32611, name: qemu preempt_count: 1, expected: 0 INFO: lockdep is turned off. irq event stamp: 30580 hardirqs last enabled at (30579): [<ffffffffc08b2527>] vcpu_run+0x1787/0x1db0 [kvm] hardirqs last disabled at (30580): [<ffffffffb4f62e32>] __schedule+0x1e2/0xed0 softirqs last enabled at (30570): [<ffffffffb4247a64>] fpu_swap_kvm_fpstate+0x44/0x210 softirqs last disabled at (30568): [<ffffffffb4247a64>] fpu_swap_kvm_fpstate+0x44/0x210 CPU: 298 UID: 0 PID: 32611 Comm: qemu Tainted: G U 6.16.0-smp--e6c618b51cfe-sleep #782 NONE Tainted: [U]=USER Hardware name: Google Astoria-Turin/astoria, BIOS 0.20241223.2-0 01/17/2025 Call Trace: <TASK> dump_stack_lvl+0x7d/0xb0 __might_resched+0x271/0x290 __might_fault+0x28/0x80 kvm_vcpu_read_guest_page+0x8d/0xc0 [kvm] kvm_fetch_guest_virt+0x92/0xc0 [kvm] __do_insn_fetch_bytes+0xf3/0x1e0 [kvm] x86_decode_insn+0xd1/0x1010 [kvm] x86_emulate_instruction+0x105/0x810 [kvm] __svm_skip_emulated_instruction+0xc4/0x140 [kvm_amd] handle_fastpath_invd+0xc4/0x1a0 [kvm] vcpu_run+0x11a1/0x1db0 [kvm] kvm_arch_vcpu_ioctl_run+0x5cc/0x730 [kvm] kvm_vcpu_ioctl+0x578/0x6a0 [kvm] __se_sys_ioctl+0x6d/0xb0 do_syscall_64+0x8a/0x2c0 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x7f479d57a94b </TASK> Note, this is essentially a reapply of commit 5c30e8101e8d ("KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"), but with different justification (KVM now grabs SRCU when skipping the instruction for other reasons). Fixes: b439eb8ab578 ("Revert "KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"") Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/20250805190526.1453366-2-seanjc@google.com Signed-off-by: Sean Christopherson <seanjc(a)google.com> [ adapted switch-based MSR/HLT fastpath to if-based MSR-only check ] Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Conflicts: arch/x86/kvm/svm/svm.c [Context conflicts due to stable commit ca3cc405a3bc ("KVM: x86: Move handling of is_guest_mode() into fastpath exit handlers") not merged. Also this makes nrip check comes after is_guest_mode check, which is fine as both checks return the same exit code.] Signed-off-by: Pan Taixi <pantaixi1(a)huawei.com> --- arch/x86/kvm/svm/svm.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c index 8692b7892d0f..7bf68a753a97 100644 --- a/arch/x86/kvm/svm/svm.c +++ b/arch/x86/kvm/svm/svm.c @@ -4206,8 +4206,17 @@ static int svm_vcpu_pre_run(struct kvm_vcpu *vcpu) static fastpath_t svm_exit_handlers_fastpath(struct kvm_vcpu *vcpu) { - if (to_svm(vcpu)->vmcb->control.exit_code == SVM_EXIT_MSR && - to_svm(vcpu)->vmcb->control.exit_info_1) + struct vcpu_svm *svm = to_svm(vcpu); + + /* + * Next RIP must be provided as IRQs are disabled, and accessing guest + * memory to decode the instruction might fault, i.e. might sleep. + */ + if (!nrips || !svm->vmcb->control.next_rip) + return EXIT_FASTPATH_NONE; + + if (svm->vmcb->control.exit_code == SVM_EXIT_MSR && + svm->vmcb->control.exit_info_1) return handle_fastpath_set_msr_irqoff(vcpu); return EXIT_FASTPATH_NONE; -- 2.34.1

2 1

[openeuler:OLK-6.6 3163/3163] mm/gmem.c:190:26: error: no member named 'mems_allowed' in 'struct task_struct'
by kernel test robot 13 Nov '25

13 Nov '25

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: d07bcf91fa24bfac8f6c62724829741c738bb85d commit: 380709c322aaa306b5ac3c278a0700043f7d33e2 [3163/3163] gmem: prepare config, boot parameter, vma flag and basic GMEM related structs config: arm64-randconfig-002-20251113 (https://download.01.org/0day-ci/archive/20251113/202511131248.lRKFr4Bu-lkp@…) compiler: clang version 22.0.0git (https://github.com/llvm/llvm-project 996639d6ebb86ff15a8c99b67f1c2e2117636ae7) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251113/202511131248.lRKFr4Bu-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202511131248.lRKFr4Bu-lkp@intel.com/ All errors (new ones prefixed by >>): >> mm/gmem.c:190:26: error: no member named 'mems_allowed' in 'struct task_struct' 190 | node_set(nid, current->mems_allowed); | ~~~~~~~ ^ include/linux/nodemask.h:128:50: note: expanded from macro 'node_set' 128 | #define node_set(node, dst) __node_set((node), &(dst)) | ^~~ 1 error generated. Kconfig warnings: (for reference only) WARNING: unmet direct dependencies detected for HALTPOLL_CPUIDLE Depends on [n]: CPU_IDLE [=n] && ARCH_CPUIDLE_HALTPOLL [=y] && ARCH_HAS_OPTIMIZED_POLL [=y] Selected by [y]: - ARM64 [=y] vim +190 mm/gmem.c 144 145 int gm_as_attach(struct gm_as *as, struct gm_dev *dev, 146 bool activate, struct gm_context **out_ctx) 147 { 148 struct gm_context *ctx; 149 int nid; 150 151 ctx = kmem_cache_alloc(gm_ctx_cache, GFP_KERNEL); 152 if (!ctx) 153 return -ENOMEM; 154 155 ctx->as = as; 156 ctx->dev = dev; 157 158 INIT_LIST_HEAD(&ctx->gm_dev_link); 159 INIT_LIST_HEAD(&ctx->gm_as_link); 160 161 if (!list_empty(&as->gm_ctx_list)) { 162 struct list_head *old_node; 163 struct gm_context *old_ctx; 164 165 old_node = as->gm_ctx_list.prev; 166 list_del_init(old_node); 167 old_ctx = list_entry(old_node, struct gm_context, gm_as_link); 168 kfree(old_ctx); 169 } 170 171 list_add_tail(&dev->gm_ctx_list, &ctx->gm_dev_link); 172 list_add_tail(&ctx->gm_as_link, &as->gm_ctx_list); 173 174 if (activate) { 175 /* 176 * Here we should really have a callback function to perform the context switch 177 * for the hardware. E.g. in x86 this function is effectively 178 * flushing the CR3 value. Currently we do not care time-sliced context switch, 179 * unless someone wants to support it. 180 */ 181 dev->current_ctx = ctx; 182 } 183 *out_ctx = ctx; 184 185 /* 186 * gm_as_attach will be used to attach device to process address space. 187 * Handle this case and add hnodes registered by device to process mems_allowed. 188 */ 189 for_each_node_mask(nid, dev->registered_hnodes) > 190 node_set(nid, current->mems_allowed); -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[openeuler:OLK-6.6 3163/3163] drivers/ub/ubfi/irq.c:18:26: error: 'struct dev_msi_info' has no member named 'domain'
by kernel test robot 13 Nov '25

13 Nov '25

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: fbfb361fe93de1a562df2ace860f64c4067abfdc commit: 028a4161a77d2e15d331a13599a6f06cdd7490b6 [3163/3163] ub:ubfi: register ubc usi domain config: x86_64-buildonly-randconfig-004-20251113 (https://download.01.org/0day-ci/archive/20251113/202511130841.AMzPRIEd-lkp@…) compiler: gcc-14 (Debian 14.2.0-19) 14.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251113/202511130841.AMzPRIEd-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202511130841.AMzPRIEd-lkp@intel.com/ All errors (new ones prefixed by >>): drivers/ub/ubfi/irq.c: In function 'ub_update_msi_domain': >> drivers/ub/ubfi/irq.c:18:26: error: 'struct dev_msi_info' has no member named 'domain' 18 | domain = dev->msi.domain; | ^ vim +18 drivers/ub/ubfi/irq.c 11 12 int ub_update_msi_domain(struct device *dev, 13 enum irq_domain_bus_token bus_token) 14 { 15 struct fwnode_handle *fwnode; 16 struct irq_domain *domain; 17 > 18 domain = dev->msi.domain; 19 if (!domain) { 20 dev_err(dev, "find base irq domain failed!\n"); 21 return -ENODEV; 22 } 23 24 fwnode = domain->fwnode; 25 if (!fwnode) { 26 dev_err(dev, "find fwnode failed!\n"); 27 return -ENODEV; 28 } 29 30 domain = irq_find_matching_fwnode(fwnode, bus_token); 31 if (!domain) { 32 dev_err(dev, "find irq domain failed!\n"); 33 return -ENODEV; 34 } 35 36 /* Update msi domain with new bus_token */ 37 dev_set_msi_domain(dev, domain); 38 39 return 0; 40 } 41 EXPORT_SYMBOL_GPL(ub_update_msi_domain); 42 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[openeuler:OLK-6.6 3135/3135] mm/gmem_phys.c:245:19: warning: no previous prototype for function 'gm_evict_page_locked'
by kernel test robot 13 Nov '25

13 Nov '25

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: fbfb361fe93de1a562df2ace860f64c4067abfdc commit: 8f4f66a9d851f366942e8ee319292e41ad2baa95 [3135/3135] gmem: support swapping heterogeneous memory config: arm64-randconfig-001-20251112 (https://download.01.org/0day-ci/archive/20251113/202511130834.0qWsUngh-lkp@…) compiler: clang version 22.0.0git (https://github.com/llvm/llvm-project 996639d6ebb86ff15a8c99b67f1c2e2117636ae7) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251113/202511130834.0qWsUngh-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202511130834.0qWsUngh-lkp@intel.com/ All warnings (new ones prefixed by >>): In file included from <built-in>:3: In file included from include/linux/compiler_types.h:150: include/linux/compiler-clang.h:33:9: warning: '__SANITIZE_ADDRESS__' macro redefined [-Wmacro-redefined] 33 | #define __SANITIZE_ADDRESS__ | ^ <built-in>:368:9: note: previous definition is here 368 | #define __SANITIZE_ADDRESS__ 1 | ^ >> mm/gmem_phys.c:245:19: warning: no previous prototype for function 'gm_evict_page_locked' [-Wmissing-prototypes] 245 | enum gm_evict_ret gm_evict_page_locked(struct gm_page *gm_page) | ^ mm/gmem_phys.c:245:1: note: declare 'static' if the function is not intended to be used outside of this translation unit 245 | enum gm_evict_ret gm_evict_page_locked(struct gm_page *gm_page) | ^ | static >> mm/gmem_phys.c:345:19: warning: no previous prototype for function 'gm_evict_page' [-Wmissing-prototypes] 345 | enum gm_evict_ret gm_evict_page(struct gm_page *gm_page) | ^ mm/gmem_phys.c:345:1: note: declare 'static' if the function is not intended to be used outside of this translation unit 345 | enum gm_evict_ret gm_evict_page(struct gm_page *gm_page) | ^ | static 3 warnings generated. vim +/gm_evict_page_locked +245 mm/gmem_phys.c 244 > 245 enum gm_evict_ret gm_evict_page_locked(struct gm_page *gm_page) 246 { 247 struct gm_dev *gm_dev; 248 struct gm_mapping *gm_mapping; 249 struct vm_area_struct *vma; 250 struct mm_struct *mm; 251 struct page *page; 252 struct device *dma_dev; 253 unsigned long va; 254 struct folio *folio = NULL; 255 struct gm_fault_t gmf = { 256 .size = HPAGE_SIZE, 257 .copy = true 258 }; 259 enum gm_evict_ret ret = GM_EVICT_SUCCESS; 260 enum gm_ret gm_ret; 261 262 gm_dev = get_gm_dev(gm_page->hnid); 263 if (!gm_dev) 264 return GM_EVICT_DEVERR; 265 266 spin_lock(&gm_page->rmap_lock); 267 if (!gm_page->mm) { 268 /* Evicting gm_page conflicts with unmap.*/ 269 ret = GM_EVICT_UNMAP; 270 goto rmap_unlock; 271 } 272 273 mm = gm_page->mm; 274 va = gm_page->va; 275 vma = find_vma(mm, va); 276 if (!vma || !vma->vm_obj) { 277 gmem_err("%s: cannot find vma or vma->vm_obj is null for va %lx", __func__, va); 278 ret = GM_EVICT_UNMAP; 279 goto rmap_unlock; 280 } 281 282 gm_mapping = vm_object_lookup(vma->vm_obj, va); 283 if (!gm_mapping) { 284 gmem_err("%s: no gm_mapping for va %lx", __func__, va); 285 ret = GM_EVICT_UNMAP; 286 goto rmap_unlock; 287 } 288 289 spin_unlock(&gm_page->rmap_lock); 290 291 mutex_lock(&gm_mapping->lock); 292 if (!gm_mapping_device(gm_mapping)) { 293 /* Evicting gm_page conflicts with unmap.*/ 294 ret = GM_EVICT_UNMAP; 295 goto gm_mapping_unlock; 296 } 297 298 if (gm_mapping->gm_page != gm_page) { 299 /* gm_mapping maps to another gm_page. */ 300 ret = GM_EVICT_UNMAP; 301 goto gm_mapping_unlock; 302 } 303 304 folio = vma_alloc_folio(GFP_TRANSHUGE, HPAGE_PMD_ORDER, vma, va, true); 305 if (!folio) { 306 gmem_err("%s: allocate host page failed.", __func__); 307 ret = GM_EVICT_FALLBACK; 308 goto gm_mapping_unlock; 309 } 310 page = &folio->page; 311 312 gmf.mm = mm; 313 gmf.va = va; 314 gmf.dev = gm_dev; 315 gmf.pfn = gm_page->dev_pfn; 316 dma_dev = gm_dev->dma_dev; 317 gmf.dma_addr = dma_map_page(dma_dev, page, 0, HPAGE_SIZE, DMA_BIDIRECTIONAL); 318 if (dma_mapping_error(dma_dev, gmf.dma_addr)) { 319 gmem_err("%s: dma map failed.", __func__); 320 ret = GM_EVICT_FALLBACK; 321 goto gm_mapping_unlock; 322 } 323 324 gm_ret = gm_dev->mmu->peer_unmap(&gmf); 325 if (gm_ret != GM_RET_SUCCESS) { 326 gmem_err("%s: peer_unmap failed.", __func__); 327 ret = GM_EVICT_DEVERR; 328 goto dma_unmap; 329 } 330 331 gm_mapping_flags_set(gm_mapping, GM_MAPPING_CPU); 332 gm_page_remove_rmap(gm_page); 333 gm_mapping->page = page; 334 put_gm_page(gm_page); 335 dma_unmap: 336 dma_unmap_page(dma_dev, gmf.dma_addr, HPAGE_SIZE, DMA_BIDIRECTIONAL); 337 gm_mapping_unlock: 338 mutex_unlock(&gm_mapping->lock); 339 return ret; 340 rmap_unlock: 341 spin_unlock(&gm_page->rmap_lock); 342 return ret; 343 } 344 > 345 enum gm_evict_ret gm_evict_page(struct gm_page *gm_page) 346 { 347 struct mm_struct *mm = gm_page->mm; 348 enum gm_evict_ret ret; 349 350 mmap_read_lock(mm); 351 ret = gm_evict_page_locked(gm_page); 352 mmap_read_unlock(mm); 353 return ret; 354 } 355 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[PATCH OLK-5.10 0/4] Backport kunpeng920G to OLK-5.10
by Lin Yujun 13 Nov '25

13 Nov '25

Cong Wang (1): iommu: avoid taking iova_rbtree_lock twice Xiang Chen (3): iommu/arm-smmu-v3: Add a check to avoid invalid iotlb sync dma-mapping: benchmark: Add support for multi-pages map/unmap iommu/iova: Put free_iova_mem() outside of spinlock iova_rbtree_lock drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 3 +++ drivers/iommu/iova.c | 24 ++++++++++++------- kernel/dma/map_benchmark.c | 21 ++++++++++------ .../testing/selftests/dma/dma_map_benchmark.c | 20 ++++++++++++---- 4 files changed, 49 insertions(+), 19 deletions(-) -- 2.34.1

2 5

[PATCH OLK-6.6 0/3] wakeup kswapd during node_reclaim
by Wupeng Ma 13 Nov '25

13 Nov '25

wakeup kswapd during node_reclaim and add new bits for zone recalim to enable this kind of swap. Wupeng Ma (3): arm64: support node_reclaim_distance adjust for arch mm: vmscan: wakeup kswapd during node_reclaim mm: node_reclaim: add wakeup kswapd mode Documentation/admin-guide/sysctl/vm.rst | 1 + arch/arm64/mm/init.c | 20 ++++++++++++++++++++ include/linux/swap.h | 3 ++- include/uapi/linux/mempolicy.h | 1 + mm/internal.h | 5 +++-- mm/page_alloc.c | 3 ++- mm/vmscan.c | 7 ++++++- 7 files changed, 35 insertions(+), 5 deletions(-) -- 2.43.0

2 4

[PATCH OLK-6.6 00/14] mm: add huge pfnmap support for remap_pfn_range()
by Yin Tirui 12 Nov '25

12 Nov '25

Overview ======== This patch series adds huge page support for remap_pfn_range(), automatically creating huge mappings when prerequisites are satisfied (size, alignment, architecture support, etc.) and falling back to normal page mappings otherwise. This work builds on Peter Xu's previous efforts on huge pfnmap support [0]. TODO ==== - Add PUD-level huge page support. Currently, only PMD-level huge pages are supported. - Consider the logic related to vmap_page_range and extract reusable common code. Tests Done ========== - Cross-build tests. - Performance tests with custom device driver implementing mmap() with remap_pfn_range(): - lat_mem_rd benchmark modified to use mmap(device_fd) instead of malloc() shows around 40% improvement in memory access latency with huge page support compared to normal page mappings. numactl -C 0 lat_mem_rd -t 4096M (stride=64) Memory Size (MB) Without Huge Mapping With Huge Mapping Improvement ---------------- ----------------- -------------- ----------- 64.00 148.858 ns 100.780 ns 32.3% 128.00 164.745 ns 103.537 ns 37.2% 256.00 169.907 ns 103.179 ns 39.3% 512.00 171.285 ns 103.072 ns 39.8% 1024.00 173.054 ns 103.055 ns 40.4% 2048.00 172.820 ns 103.091 ns 40.3% 4096.00 172.877 ns 103.115 ns 40.4% - Custom memory copy operations on mmap(device_fd) show around 18% performance improvement with huge page support compared to normal page mappings. numactl -C 0 memcpy_test (memory copy performance test) Memory Size (MB) Without Huge Mapping With Huge Mapping Improvement ---------------- ----------------- -------------- ----------- 1024.00 95.76 ms 77.91 ms 18.6% 2048.00 190.87 ms 155.64 ms 18.5% 4096.00 380.84 ms 311.45 ms 18.2% [0] https://lore.kernel.org/all/20240826204353.2228736-2-peterx@redhat.com/T/#u David Hildenbrand (1): mm/huge_memory: check pmd_special() only after pmd_present() Peter Xu (11): mm: introduce ARCH_SUPPORTS_HUGE_PFNMAP and special bits to pmd/pud mm: drop is_huge_zero_pud() mm: mark special bits for huge pfn mappings when inject mm: allow THP orders for PFNMAPs mm/gup: detect huge pfnmap entries in gup-fast mm/pagewalk: check pfnmap for folio_walk_start() mm/fork: accept huge pfnmap entries mm: always define pxx_pgprot() mm/x86: support large pfn mappings mm/arm64: support large pfn mappings arm64: mm: Drop dead code for pud special bit handling Yin Tirui (2): pgtable: add pte_clrhuge() implementation for arm64 mm: introduce remap_pfn_range_try_pmd() for PMD-level hugepage mapping arch/arm64/Kconfig | 1 + arch/arm64/include/asm/pgtable.h | 39 ++++++ arch/powerpc/include/asm/book3s/64/pgtable.h | 2 +- arch/powerpc/include/asm/pgtable.h | 1 + arch/s390/include/asm/pgtable.h | 1 + arch/sparc/include/asm/pgtable_64.h | 1 + arch/x86/Kconfig | 1 + arch/x86/include/asm/pgtable.h | 81 +++++++----- fs/dax.c | 2 +- include/linux/huge_mm.h | 16 +-- include/linux/mm.h | 26 ++++ include/linux/pgtable.h | 20 ++- mm/Kconfig | 13 ++ mm/gup.c | 6 + mm/huge_memory.c | 93 ++++++++++---- mm/memory.c | 126 +++++++++++++++---- 16 files changed, 333 insertions(+), 96 deletions(-) -- 2.43.0

2 15

[PATCH OLK-6.6 v7 00/10] Add xcall 2.0 support
by Jinjie Ruan 12 Nov '25

12 Nov '25

Add xcall 2.0 support, which is compatible with xcall1.0, and can be used to dynamic instruction replacement for hardware xcall. Changes in v7: - Fix alternative bug in sync ventry. - Add "xcall" limit for xcall2.0 and hardware xcall. Changes in v6: - Add "xcall=debug" debug mode to use the "fast syscall" for every system call. - Use alternative to reduce the baseline performance noise floor when not set "xcall" on the kernel cmdline - Move binary_path out from xcall struct, and add to xcall_comm struct. - Fix1. insert_xcall_locked() check both "name" and "binary" to avoid repeated attach same name of xcall. - Fix2. Check if it is an executable file for the binary path. - Fix3. Check if it is an absolute path for the binary path. - Fix4. Remove the extra free_xcall_comm(info) in default branch in proc_xcall_command(). - Fix5. Add zero syscall hijack check for xcall_prog_register(). - Use __arm64_sys_ni_syscall() instead of define inv_xcall(). - Replace the -ENOSYS return value with -EINVAL, to avoid the warning. Changes in v5: - Fix build problem. - el0t_64_sync_ventry jump to kernel_ventry to do VMAP_STACK. - Some cleanups. - Merge the three refactor patch. Changes in v4: - Reuse uprobe_write_opcode() instead of use __replace_page() and copy_to_page(). - Handle error for uprobe_write_opcode(). - Support xcall1.0 before 920G. - Change xcall1.0 bitmap to byte array to save check insns. - Use percpu variable to save check insns for xcall1.0. - A little cleanup. Changes in v3: - Add validity checks during xcall registration to prevent out-of-bounds access and null pointer dereferences. - Extract __replace_page() and copy_to_page() from uprobe, to avoid code duplication. - Replace with new compatible refactored code. - Remove unnecessary member, such as old_name. Changes in v2: - Remove duplicates or unnecessary ~300 LOC, such as link_slot, init_task. - Fix kabi. - Some cleanups. Jinjie Ruan (2): xcall2.0: Support xcall1.0 for hardware xcall openeuler_defconfig: Enable CONFIG_DYNAMIC_XCALL by default Liao Chang (6): xcall2.0: Add userspace proc interface xcall2.0: Add xcall module register interface xcall2.0: Add xcall_area xcall2.0: Hajack syscall with dynamic instruciton replace xcall: Rework the early exception vector of XCALL and SYNC xcall2.0: Add a basic testcase Yipeng Zou (1): xcall: Add debug mode to force fast syscall for all threads Yuntao Liu (1): xcall2.0: Intruduce xcall2.0 prefetch kernel module arch/arm64/Kconfig.turbo | 13 + arch/arm64/configs/openeuler_defconfig | 1 + arch/arm64/include/asm/exception.h | 4 +- arch/arm64/include/asm/mmu_context.h | 7 - arch/arm64/include/asm/xcall.h | 154 ++++---- arch/arm64/kernel/cpufeature.c | 79 +++-- arch/arm64/kernel/entry-common.c | 28 +- arch/arm64/kernel/entry.S | 86 +---- arch/arm64/kernel/probes/uprobes.c | 6 + arch/arm64/kernel/process.c | 5 + arch/arm64/kernel/syscall.c | 14 + arch/arm64/kernel/xcall/Makefile | 3 +- arch/arm64/kernel/xcall/core.c | 390 +++++++++++++++++++++ arch/arm64/kernel/xcall/entry.S | 139 ++++++-- arch/arm64/kernel/xcall/proc.c | 226 ++++++++++++ arch/arm64/kernel/xcall/xcall.c | 66 ++-- arch/arm64/kvm/sys_regs.c | 1 + drivers/staging/Kconfig | 2 + drivers/staging/Makefile | 1 + drivers/staging/xcall/Kconfig | 19 + drivers/staging/xcall/Makefile | 1 + drivers/staging/xcall/dynamic_xcall_test.c | 90 +++++ drivers/staging/xcall/prefetch.c | 265 ++++++++++++++ fs/proc/proc_xcall.c | 144 ++------ include/linux/mm_types.h | 4 + include/linux/xcall.h | 51 +++ kernel/events/uprobes.c | 23 ++ kernel/fork.c | 3 + mm/mmap.c | 14 +- 29 files changed, 1465 insertions(+), 374 deletions(-) create mode 100644 arch/arm64/kernel/xcall/core.c create mode 100644 arch/arm64/kernel/xcall/proc.c create mode 100644 drivers/staging/xcall/Kconfig create mode 100644 drivers/staging/xcall/Makefile create mode 100644 drivers/staging/xcall/dynamic_xcall_test.c create mode 100644 drivers/staging/xcall/prefetch.c create mode 100644 include/linux/xcall.h -- 2.34.1

2 11