alloc_tag-fix-module-allocation-tags-populated-area-calculation.patch removed from -mm tree
The quilt patch titled
Subject: alloc_tag: fix module allocation tags populated area calculation
has been removed from the -mm tree. Its filename was
alloc_tag-fix-module-allocation-tags-populated-area-calculation.patch
This patch was dropped because it was merged into the mm-hotfixes-stable branch of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
------------------------------------------------------
From: Suren Baghdasaryan <[email protected]>
Subject: alloc_tag: fix module allocation tags populated area calculation
Date: Fri, 29 Nov 2024 16:14:22 -0800
vm_module_tags_populate() calculation of the populated area assumes that area starts at a page boundary and therefore when new pages are allocation, the end of the area is page-aligned as well. If the start of the area is not page-aligned then allocating a page and incrementing the end of the area by PAGE_SIZE leads to an area at the end but within the area boundary which is not populated. Accessing this are will lead to a kernel panic. Fix the calculation by down-aligning the start of the area and using that as the location allocated pages are mapped to.
---
lib/alloc_tag.c | 34 +++++++++++++++++++++++++++++----- 1 file changed, 29 insertions(+), 5 deletions(-)
lib/alloc_tag.c renamed from lib/alloc_tag.c~alloc_tag-fix-module-allocation-tags-populated-area-calculation +29 -5
@@ -408,28 +408,52 @@ repeat:static int vm_module_tags_populate (void ) { -unsigned long phys_size = vm_module_tags->nr_pages <<PAGE_SHIFT ; +unsigned long phys_end =ALIGN_DOWN (module_tags.start_addr,PAGE_SIZE ) + + (vm_module_tags->nr_pages <<PAGE_SHIFT ); +unsigned long new_end = module_tags.start_addr + module_tags.size; -if (phys_size < module_tags.size) { +if (phys_end < new_end) {struct page **next_page = vm_module_tags->pages + vm_module_tags->nr_pages; -unsigned long addr = module_tags.start_addr + phys_size; +unsigned long old_shadow_end =ALIGN (phys_end,MODULE_ALIGN ); +unsigned long new_shadow_end =ALIGN (new_end,MODULE_ALIGN );unsigned long more_pages;unsigned long nr; - more_pages =ALIGN (module_tags.size - phys_size,PAGE_SIZE ) >>PAGE_SHIFT ; + more_pages =ALIGN (new_end - phys_end,PAGE_SIZE ) >>PAGE_SHIFT ; nr =alloc_pages_bulk_array_node (GFP_KERNEL | __GFP_NOWARN,NUMA_NO_NODE , more_pages, next_page);if (nr < more_pages || -vmap_pages_range (addr, addr + (nr <<PAGE_SHIFT ),PAGE_KERNEL , +vmap_pages_range (phys_end, phys_end + (nr <<PAGE_SHIFT ),PAGE_KERNEL , next_page,PAGE_SHIFT ) <0 ) {/* Clean up and error out */ for (int i =0 ; i < nr; i++)__free_page (next_page[i]);return -ENOMEM ; } + vm_module_tags->nr_pages += nr; + +/* +* Kasan allocates 1 byte of shadow for every 8 bytes of data. +* When kasan_alloc_module_shadow allocates shadow memory, +* its unit of allocation is a page. +* Therefore, here we need to align to MODULE_ALIGN. +*/ +if (old_shadow_end < new_shadow_end) +kasan_alloc_module_shadow ((void *)old_shadow_end, + new_shadow_end - old_shadow_end, +GFP_KERNEL ); } +/* +* Mark the pages as accessible, now that they are mapped. +* With hardware tag-based KASAN, marking is skipped for +* non-VM_ALLOC mappings, see __kasan_unpoison_vmalloc(). +*/ +kasan_unpoison_vmalloc ((void *)module_tags.start_addr, + new_end - module_tags.start_addr, +KASAN_VMALLOC_PROT_NORMAL ); +return 0 ; }
_
Patches currently in -mm which might be from [email protected] are
seqlock-add-raw_seqcount_try_begin.patch mm-convert-mm_lock_seq-to-a-proper-seqcount.patch mm-introduce-mmap_lock_speculate_try_beginretry.patch