From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 81B204E80C8; Thu, 3 Sep 2026 15:50:17 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788450618; cv=none; b=ITthGmaVGbFwliegvf4I3DuYay1v/lLgK2RdhjVykDSIckfVW60te7/LRacoN4LdC+HLZ3Gc096HLyTc0aH91uM2Xj6enCVW4JIr/z9ZQgWfFY+SCvysSmlEqaTdGQFK5hOubZrpdWGgV4ZTZrgh5scvMVmAsbKWMoftZMQjExo= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788450618; c=relaxed/simple; bh=lh2dT7ZewrgHg9g4KMVMPX90guRzzwDyD8C2CSWrPmk=; h=From:Date:Subject:MIME-Version:Content-Type:Message-Id:References: In-Reply-To:To:Cc; b=t35Dho0aRahsPUolMPyE0EP2kpfUuIrGmQ1/MbnS/NppDOEaouw25ZmmDQ+/I2GlwmWxuqpf5DJMf2g3Xdu2v4nD0eMwMERF2CHWTQyYI1IzCUTU4rmdd/wPh8vmF0m/Y0VMPOrrPsKOzE+GidbnXQsWM7mkGWEt4pwTZM8m9ZM= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=OYJ+Zl3t; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="OYJ+Zl3t" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 4180E1F00A3F; Thu, 3 Sep 2026 15:50:15 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1788450617; bh=Z0X9ZLAQ4mUEchhwpqleI41xzP5aRuvK6f6Sh6Yxzi4=; h=From:Date:Subject:References:In-Reply-To:To:Cc; b=OYJ+Zl3tMLn1MD1z6fiXKp4+i9tHpSYBMlJNN0tpYuVSfwQKTR5zVkw45bX7Q4PTw o6NmhmyOQC1T8A7OxYMHgBjsFPek/w5rxSk5f3hxPJ7wUkgyINX5RUuG5hZErfLTDK d8FnpW/qGCiLySjLSSd7a5lqoeN4/j150u5+rP8OLeGhNApjIC16fJT869KuufeTIf Xz+fwfT6vjd7WFXrTSmMhfp1+3YJtwirO/TcznnlXg4cFTacvgDB1ywtBQ022spaHK A5TX3vACapU+ds4GgORfT0T1iv6dfsmsWcn7E41bTU0lD7+wbcf2SY42YlJ4vMlOeH geH2yghWdDQmg== From: "Mike Rapoport (Microsoft)" Date: Thu, 03 Sep 2026 18:50:00 +0300 Subject: [PATCH 3/5] mm/execmem: make sure ROX cache always contains multiples of PMD_SIZE Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Message-Id: <20260903-execmem-rox-cache-pmd-v1-v1-3-11beb2a3d249@kernel.org> References: <20260903-execmem-rox-cache-pmd-v1-v1-0-11beb2a3d249@kernel.org> In-Reply-To: <20260903-execmem-rox-cache-pmd-v1-v1-0-11beb2a3d249@kernel.org> To: Andrew Morton , Benjamin Tissoires , Jiri Kosina , Uladzislau Rezki Cc: Luis Chamberlain , Mike Rapoport , linux-input@vger.kernel.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org X-Mailer: b4 0.17-dev The ROX cache relies on its chunks being PMD mapped. For a PMD mapped chunk, set_memory_rox() updates the direct map alias one PMD at a time and the large mappings there survive. When execmem refills the cache, it rounds up the requested allocation size to PMD_SIZE and tries to allocate that with vmalloc(VM_ALLOW_HUGE_VMAP). If that allocation fails, execmem falls back to vmalloc() of the original size. There are two issues with this approach: * If huge pages are not available, __vmalloc_node_range() silently falls back to base pages. The area execmem gets is virtually contiguous, but it is backed by 512 scattered base pages. Permission updates on such areas split large mappings in the direct map that contain those base pages, up to 512 PMD splits in the worst case. * execmem's own fallback adds a small base page mapped area to the cache. This adds the overhead of cache management to these allocations with no benefit of reducing fragmentation either in the vmalloc/modules address space or in the direct map. Worse, these areas are never freed from the cache, because execmem_cache_clean() releases only chunks that are a multiple of PMD_SIZE and aligned to PMD_SIZE, exactly to minimize the number of base page mappings. Add VM_REQUIRE_HUGE_VMAP option to vmalloc that fails if the allocation of huge pages fails or if such an allocation is not possible because huge page allocations in vmalloc were disabled or the architecture does not support them. Use this option when populating the ROX cache. If vmalloc(VM_REQUIRE_HUGE_VMAP) fails or vmalloc of huge pages is unavailable, handle the memory allocation outside the ROX cache with plain vmalloc(). Since the fallback allocation has to return ROX memory, add an execmem_alloc_rox() helper and use it for both populating the ROX cache and dealing with a fallback allocation in a ROX execmem_range. With that, the cache only ever contains PMD aligned chunks sized as a multiple of PMD_SIZE, and the PMD checks in execmem_cache_clean() become a VM_WARN_ON_ONCE() to ensure that the PMD mapping invariant does not change. Assisted-by: copilot:claude-opus-5 Signed-off-by: Mike Rapoport (Microsoft) --- include/linux/vmalloc.h | 1 + mm/execmem.c | 76 ++++++++++++++++++++++++++++++------------------- mm/vmalloc.c | 15 +++++++++- 3 files changed, 62 insertions(+), 30 deletions(-) diff --git a/include/linux/vmalloc.h b/include/linux/vmalloc.h index aed121d729b01..6e555e31e6225 100644 --- a/include/linux/vmalloc.h +++ b/include/linux/vmalloc.h @@ -38,6 +38,7 @@ struct iov_iter; /* in uio.h */ #define VM_DEFER_KMEMLEAK 0 #endif #define VM_SPARSE 0x00001000 /* sparse vm_area. not all pages are present. */ +#define VM_REQUIRE_HUGE_VMAP 0x00002000 /* huge page mapping or nothing */ /* bits [20..32] reserved for arch specific ioremap internals */ diff --git a/mm/execmem.c b/mm/execmem.c index 00dd6324cae01..77653b7f163dc 100644 --- a/mm/execmem.c +++ b/mm/execmem.c @@ -51,7 +51,8 @@ static void *execmem_vmalloc(struct execmem_range *range, size_t size, } if (!p) { - pr_warn_ratelimited("unable to allocate memory\n"); + if (!(vm_flags & VM_REQUIRE_HUGE_VMAP)) + pr_warn_ratelimited("unable to allocate memory\n"); return NULL; } @@ -146,9 +147,10 @@ static void execmem_cache_clean(struct work_struct *work) struct vm_struct *vm = find_vm_area(area); size_t size = mas_range_len(&mas); - if (vm && get_vm_area_size(vm) == size && - IS_ALIGNED(size, PMD_SIZE) && - IS_ALIGNED(mas.index, PMD_SIZE)) { + if (vm && get_vm_area_size(vm) == size) { + VM_WARN_ON_ONCE(!IS_ALIGNED(mas.index, PMD_SIZE) || + !IS_ALIGNED(size, PMD_SIZE)); + /* * Preallocate to ensure mas_store does not fail * If there is no memory for the tree update, bail out, @@ -264,38 +266,41 @@ static void *__execmem_cache_alloc(struct execmem_range *range, size_t size) return execmem_cache_alloc_locked(range, size); } -static void *execmem_cache_populate_alloc(struct execmem_range *range, size_t size) +static void *execmem_vmalloc_rox(struct execmem_range *range, size_t size, + unsigned long vm_flags) { - unsigned long vm_flags = VM_ALLOW_HUGE_VMAP; - struct mutex *mutex = &execmem_cache.mutex; - struct vm_struct *vm; - size_t alloc_size; - int err = -ENOMEM; - void *p; - - alloc_size = round_up(size, PMD_SIZE); - p = execmem_vmalloc(range, alloc_size, PAGE_KERNEL, vm_flags); - if (!p) { - alloc_size = size; - p = execmem_vmalloc(range, alloc_size, PAGE_KERNEL, vm_flags); - } + void *p = execmem_vmalloc(range, size, PAGE_KERNEL, vm_flags); + int err; if (!p) return NULL; - vm = find_vm_area(p); - if (!vm) - goto err_free_mem; - /* fill memory with instructions that will trap */ - execmem_fill_trapping_insns(p, alloc_size); - + execmem_fill_trapping_insns(p, size); set_vm_flush_reset_perms(p); - - err = set_memory_rox((unsigned long)p, vm->nr_pages); + err = set_memory_rox((unsigned long)p, size >> PAGE_SHIFT); if (err) goto err_free_mem; + return p; + +err_free_mem: + vfree(p); + return NULL; +} + +static void *execmem_cache_populate_alloc(struct execmem_range *range, size_t size) +{ + unsigned long vm_flags = VM_REQUIRE_HUGE_VMAP; + size_t alloc_size = round_up(size, PMD_SIZE); + struct mutex *mutex = &execmem_cache.mutex; + int err; + void *p; + + p = execmem_vmalloc_rox(range, alloc_size, vm_flags); + if (!p) + return NULL; + /* * New memory blocks must be allocated and added to the cache * as an atomic operation, otherwise they may be consumed @@ -317,6 +322,11 @@ static void *execmem_cache_populate_alloc(struct execmem_range *range, size_t si return NULL; } +static void *execmem_alloc_rox(struct execmem_range *range, size_t size) +{ + return execmem_vmalloc_rox(range, size, 0); +} + static void *execmem_cache_alloc(struct execmem_range *range, size_t size) { void *p; @@ -444,6 +454,11 @@ static void *execmem_cache_alloc(struct execmem_range *range, size_t size) return NULL; } +static void *execmem_alloc_rox(struct execmem_range *range, size_t size) +{ + return NULL; +} + static bool execmem_cache_free(void *ptr) { return false; @@ -453,17 +468,20 @@ static bool execmem_cache_free(void *ptr) void *execmem_alloc(enum execmem_type type, size_t size) { struct execmem_range *range = &execmem_info->ranges[type]; - bool use_cache = range->flags & EXECMEM_ROX_CACHE; + bool use_rox_cache = range->flags & EXECMEM_ROX_CACHE; unsigned long vm_flags = VM_FLUSH_RESET_PERMS; pgprot_t pgprot = range->pgprot; void *p = NULL; size = PAGE_ALIGN(size); - if (use_cache) + if (use_rox_cache) { p = execmem_cache_alloc(range, size); - else + if (!p) + p = execmem_alloc_rox(range, size); + } else { p = execmem_vmalloc(range, size, pgprot, vm_flags); + } return kasan_reset_tag(p); } diff --git a/mm/vmalloc.c b/mm/vmalloc.c index 6ed6c160abed7..4df5c25786c6d 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -4027,6 +4027,12 @@ static gfp_t vmalloc_fix_flags(gfp_t flags) * %__GFP_SKIP_KASAN can be used to skip unpoisoning of mapped pages * (when prot=%PAGE_KERNEL). * + * %VM_ALLOW_HUGE_VMAP allocates huge pages when possible and falls back to + * base pages if huge page allocation fails. + * + * %VM_REQUIRE_HUGE_VMAP implies %VM_ALLOW_HUGE_VMAP and fails instead of + * silently falling back to base pages. + * * Can not be called from interrupt nor NMI contexts. * Return: the address of the area or %NULL on failure */ @@ -4052,6 +4058,10 @@ void *__vmalloc_node_range_noprof(unsigned long size, unsigned long align, return NULL; } + /* VM_REQUIRE_HUGE_VMAP implies VM_ALLOW_HUGE_VMAP */ + if (vm_flags & VM_REQUIRE_HUGE_VMAP) + vm_flags |= VM_ALLOW_HUGE_VMAP; + if (vmap_allow_huge && (vm_flags & VM_ALLOW_HUGE_VMAP)) { /* * Try huge pages. Only try for PAGE_KERNEL allocations, @@ -4068,6 +4078,9 @@ void *__vmalloc_node_range_noprof(unsigned long size, unsigned long align, align = max(original_align, 1UL << shift); } + if ((vm_flags & VM_REQUIRE_HUGE_VMAP) && shift == PAGE_SHIFT) + return NULL; + again: area = __get_vm_area_node(size, align, shift, VM_ALLOC | VM_UNINITIALIZED | vm_flags, start, end, node, @@ -4142,7 +4155,7 @@ void *__vmalloc_node_range_noprof(unsigned long size, unsigned long align, return area->addr; fail: - if (shift > PAGE_SHIFT) { + if (shift > PAGE_SHIFT && !(vm_flags & VM_REQUIRE_HUGE_VMAP)) { shift = PAGE_SHIFT; align = original_align; goto again; -- 2.53.0