From fd7c3c3767c38dbbf2f5c993c6fd42a71846e7e0 Mon Sep 17 00:00:00 2001 From: Mark-PK Tsai Date: Thu, 19 Oct 2023 09:35:22 +0100 Subject: [PATCH 1/5] ARM: 9327/1: vfp: Add missing VFP instructions to neon_support_hook Add the missing "Unconditional Advanced SIMD and floating-point instructions" in [1] to the VFP undef hook. This commit addresses the issue reported in [2], where executing the vudot instruction on a platform with FEAT_DotProd support resulted in an undefined instruction error. Link: https://developer.arm.com/documentation/ddi0597/2023-06/?lang=en [1] Link: https://lore.kernel.org/lkml/20230920083907.30479-1-mark-pk.tsai@mediatek.com/ [2] Signed-off-by: Mark-PK Tsai Tested-by: Xuewen Yan Signed-off-by: Russell King (Oracle) --- arch/arm/vfp/vfpmodule.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/arch/arm/vfp/vfpmodule.c b/arch/arm/vfp/vfpmodule.c index 7e8773a2d99d..b68efe643a12 100644 --- a/arch/arm/vfp/vfpmodule.c +++ b/arch/arm/vfp/vfpmodule.c @@ -800,6 +800,24 @@ static struct undef_hook neon_support_hook[] = {{ .cpsr_mask = PSR_T_BIT, .cpsr_val = PSR_T_BIT, .fn = vfp_support_entry, +}, { + .instr_mask = 0xff000800, + .instr_val = 0xfc000800, + .cpsr_mask = 0, + .cpsr_val = 0, + .fn = vfp_support_entry, +}, { + .instr_mask = 0xff000800, + .instr_val = 0xfd000800, + .cpsr_mask = 0, + .cpsr_val = 0, + .fn = vfp_support_entry, +}, { + .instr_mask = 0xff000800, + .instr_val = 0xfe000800, + .cpsr_mask = 0, + .cpsr_val = 0, + .fn = vfp_support_entry, }}; static struct undef_hook vfp_support_hook = { From f54e8634d1366926c807e2af6125b33cff555fa7 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Tue, 7 Nov 2023 01:36:03 +0100 Subject: [PATCH 2/5] ARM: 9330/1: davinci: also select PINCTRL kconfig warns when PINCTRL_SINGLE is selected but PINCTRL is not set, so also set PINCTRL for ARCH_DAVINCI. This prevents a kconfig/build warning: WARNING: unmet direct dependencies detected for PINCTRL_SINGLE Depends on [n]: PINCTRL [=n] && OF [=y] && HAS_IOMEM [=y] Selected by [y]: - ARCH_DAVINCI [=y] && ARCH_MULTI_V5 [=y] Closes: lore.kernel.org/r/202311070548.0f6XfBrh-lkp@intel.com Fixes: f962396ce292 ("ARM: davinci: support multiplatform build for ARM v5") Signed-off-by: Randy Dunlap Reported-by: kernel test robot Cc: Bartosz Golaszewski Cc: Arnd Bergmann Cc: linux-arm-kernel@lists.infradead.org Cc: patches@armlinux.org.uk Signed-off-by: Russell King (Oracle) --- arch/arm/mach-davinci/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm/mach-davinci/Kconfig b/arch/arm/mach-davinci/Kconfig index 4316e1370627..1c8bf56982e5 100644 --- a/arch/arm/mach-davinci/Kconfig +++ b/arch/arm/mach-davinci/Kconfig @@ -10,6 +10,7 @@ menuconfig ARCH_DAVINCI select PM_GENERIC_DOMAINS_OF if PM && OF select REGMAP_MMIO select RESET_CONTROLLER + select PINCTRL select PINCTRL_SINGLE if ARCH_DAVINCI From c16af1212479570454752671a170a1756e11fdfb Mon Sep 17 00:00:00 2001 From: Wang Kefeng Date: Thu, 19 Oct 2023 12:21:35 +0100 Subject: [PATCH 3/5] ARM: 9328/1: mm: try VMA lock-based page fault handling first Attempt VMA lock-based page fault handling first, and fall back to the existing mmap_lock-based handling if that fails, the ebizzy benchmark shows 25% improvement on qemu with 2 cpus. Signed-off-by: Kefeng Wang Signed-off-by: Russell King (Oracle) --- arch/arm/Kconfig | 1 + arch/arm/mm/fault.c | 30 ++++++++++++++++++++++++++++++ 2 files changed, 31 insertions(+) diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index f8567e95f98b..8f47d6762ea4 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -35,6 +35,7 @@ config ARM select ARCH_OPTIONAL_KERNEL_RWX_DEFAULT if CPU_V7 select ARCH_SUPPORTS_ATOMIC_RMW select ARCH_SUPPORTS_HUGETLBFS if ARM_LPAE + select ARCH_SUPPORTS_PER_VMA_LOCK select ARCH_USE_BUILTIN_BSWAP select ARCH_USE_CMPXCHG_LOCKREF select ARCH_USE_MEMTEST diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c index fef62e4a9edd..e96fb40b9cc3 100644 --- a/arch/arm/mm/fault.c +++ b/arch/arm/mm/fault.c @@ -278,6 +278,35 @@ do_page_fault(unsigned long addr, unsigned int fsr, struct pt_regs *regs) perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, addr); + if (!(flags & FAULT_FLAG_USER)) + goto lock_mmap; + + vma = lock_vma_under_rcu(mm, addr); + if (!vma) + goto lock_mmap; + + if (!(vma->vm_flags & vm_flags)) { + vma_end_read(vma); + goto lock_mmap; + } + fault = handle_mm_fault(vma, addr, flags | FAULT_FLAG_VMA_LOCK, regs); + if (!(fault & (VM_FAULT_RETRY | VM_FAULT_COMPLETED))) + vma_end_read(vma); + + if (!(fault & VM_FAULT_RETRY)) { + count_vm_vma_lock_event(VMA_LOCK_SUCCESS); + goto done; + } + count_vm_vma_lock_event(VMA_LOCK_RETRY); + + /* Quick path to respond to signals */ + if (fault_signal_pending(fault, regs)) { + if (!user_mode(regs)) + goto no_context; + return 0; + } +lock_mmap: + retry: vma = lock_mm_and_find_vma(mm, addr, regs); if (unlikely(!vma)) { @@ -316,6 +345,7 @@ retry: } mmap_read_unlock(mm); +done: /* * Handle the "normal" case first - VM_FAULT_MAJOR From 89320c9785e8429155f8dfa44a183b509866e852 Mon Sep 17 00:00:00 2001 From: Mark-PK Tsai Date: Thu, 19 Oct 2023 14:38:16 +0100 Subject: [PATCH 4/5] ARM: 9329/1: kasan: Use memblock_alloc_try_nid_raw for shadow page kasan_pte_populate fill KASAN_SHADOW_INIT in the newly allocated shadow page, so it's unnecessary to use memblock_alloc_try_nid, which always zero the new allocated memory. Use memblock_alloc_try_nid_raw instead of memblock_alloc_try_nid like arm64 does which can make kasan init faster. Signed-off-by: Mark-PK Tsai Signed-off-by: Russell King (Oracle) --- arch/arm/mm/kasan_init.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/arch/arm/mm/kasan_init.c b/arch/arm/mm/kasan_init.c index 24d71b5db62d..111d4f703136 100644 --- a/arch/arm/mm/kasan_init.c +++ b/arch/arm/mm/kasan_init.c @@ -28,6 +28,12 @@ static pgd_t tmp_pgd_table[PTRS_PER_PGD] __initdata __aligned(PGD_SIZE); pmd_t tmp_pmd_table[PTRS_PER_PMD] __page_aligned_bss; +static __init void *kasan_alloc_block_raw(size_t size) +{ + return memblock_alloc_try_nid_raw(size, size, __pa(MAX_DMA_ADDRESS), + MEMBLOCK_ALLOC_NOLEAKTRACE, NUMA_NO_NODE); +} + static __init void *kasan_alloc_block(size_t size) { return memblock_alloc_try_nid(size, size, __pa(MAX_DMA_ADDRESS), @@ -50,7 +56,7 @@ static void __init kasan_pte_populate(pmd_t *pmdp, unsigned long addr, if (!pte_none(READ_ONCE(*ptep))) continue; - p = kasan_alloc_block(PAGE_SIZE); + p = kasan_alloc_block_raw(PAGE_SIZE); if (!p) { panic("%s failed to allocate shadow page for address 0x%lx\n", __func__, addr); From c17d8847c3bef9e4fe4aef34edecc29cee3cd06f Mon Sep 17 00:00:00 2001 From: Chen Haonan Date: Fri, 15 Dec 2023 11:04:40 +0100 Subject: [PATCH 5/5] ARM: 9331/1: ARM/dma-mapping: replace kzalloc() and vzalloc() with kvzalloc() using kvzalloc() simplifies the code by avoiding the use of different memory allocation functions for different situations, making the code more uniform and readable. Signed-off-by: Chen Haonan Signed-off-by: Russell King (Oracle) --- arch/arm/mm/dma-mapping.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index 5409225b4abc..d688eac6dbc1 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -859,10 +859,7 @@ static struct page **__iommu_alloc_buffer(struct device *dev, size_t size, int i = 0; int order_idx = 0; - if (array_size <= PAGE_SIZE) - pages = kzalloc(array_size, GFP_KERNEL); - else - pages = vzalloc(array_size); + pages = kvzalloc(array_size, GFP_KERNEL); if (!pages) return NULL;