The patch titled Subject: mm: abstract get_arg_page() stack expansion and mmap read lock has been added to the -mm mm-unstable branch. Its filename is mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock.patch This patch will later appear in the mm-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx> Subject: mm: abstract get_arg_page() stack expansion and mmap read lock Date: Tue, 3 Dec 2024 18:05:10 +0000 Right now fs/exec.c invokes expand_downwards(), an otherwise internal implementation detail of the VMA logic in order to ensure that an arg page can be obtained by get_user_pages_remote(). In order to be able to move the stack expansion logic into mm/vma.c in order to make it available to userland testing we need to find an alternative approach here. We do so by providing the mmap_read_lock_maybe_expand() function which also helpfully documents what get_arg_page() is doing here and adds an additional check against VM_GROWSDOWN to make explicit that the stack expansion logic is only invoked when the VMA is indeed a downward-growing stack. This allows expand_downwards() to become a static function. Importantly, the VMA referenced by mmap_read_maybe_expand() must NOT be currently user-visible in any way, that is place within an rmap or VMA tree. It must be a newly allocated VMA. This is the case when exec invokes this function. Link: https://lkml.kernel.org/r/5295d1c70c58e6aa63d14be68d4e1de9fa1c8e6d.1733248985.git.lorenzo.stoakes@xxxxxxxxxx Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx> Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx> Cc: Christian Brauner <brauner@xxxxxxxxxx> Cc: Eric W. Biederman <ebiederm@xxxxxxxxxxxx> Cc: Jan Kara <jack@xxxxxxx> Cc: Jann Horn <jannh@xxxxxxxxxx> Cc: Kees Cook <kees@xxxxxxxxxx> Cc: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx> Cc: Vlastimil Babka <vbabka@xxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- fs/exec.c | 14 ++--------- include/linux/mm.h | 5 +-- mm/mmap.c | 54 ++++++++++++++++++++++++++++++++++++++++++- 3 files changed, 58 insertions(+), 15 deletions(-) --- a/fs/exec.c~mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock +++ a/fs/exec.c @@ -205,18 +205,10 @@ static struct page *get_arg_page(struct /* * Avoid relying on expanding the stack down in GUP (which * does not work for STACK_GROWSUP anyway), and just do it - * by hand ahead of time. + * ahead of time. */ - if (write && pos < vma->vm_start) { - mmap_write_lock(mm); - ret = expand_downwards(vma, pos); - if (unlikely(ret < 0)) { - mmap_write_unlock(mm); - return NULL; - } - mmap_write_downgrade(mm); - } else - mmap_read_lock(mm); + if (!mmap_read_lock_maybe_expand(mm, vma, pos, write)) + return NULL; /* * We are doing an exec(). 'current' is the process --- a/include/linux/mm.h~mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock +++ a/include/linux/mm.h @@ -3313,6 +3313,8 @@ extern int __vm_enough_memory(struct mm_ extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); extern void exit_mmap(struct mm_struct *); int relocate_vma_down(struct vm_area_struct *vma, unsigned long shift); +bool mmap_read_lock_maybe_expand(struct mm_struct *mm, struct vm_area_struct *vma, + unsigned long addr, bool write); static inline int check_data_rlimit(unsigned long rlim, unsigned long new, @@ -3426,9 +3428,6 @@ extern unsigned long stack_guard_gap; int expand_stack_locked(struct vm_area_struct *vma, unsigned long address); struct vm_area_struct *expand_stack(struct mm_struct * mm, unsigned long addr); -/* CONFIG_STACK_GROWSUP still needs to grow downwards at some places */ -int expand_downwards(struct vm_area_struct *vma, unsigned long address); - /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, --- a/mm/mmap.c~mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock +++ a/mm/mmap.c @@ -1009,7 +1009,7 @@ static int expand_upwards(struct vm_area * vma is the first one with address < vma->vm_start. Have to extend vma. * mmap_lock held for writing. */ -int expand_downwards(struct vm_area_struct *vma, unsigned long address) +static int expand_downwards(struct vm_area_struct *vma, unsigned long address) { struct mm_struct *mm = vma->vm_mm; struct vm_area_struct *prev; @@ -1940,3 +1940,55 @@ int relocate_vma_down(struct vm_area_str /* Shrink the vma to just the new range */ return vma_shrink(&vmi, vma, new_start, new_end, vma->vm_pgoff); } + +#ifdef CONFIG_MMU +/* + * Obtain a read lock on mm->mmap_lock, if the specified address is below the + * start of the VMA, the intent is to perform a write, and it is a + * downward-growing stack, then attempt to expand the stack to contain it. + * + * This function is intended only for obtaining an argument page from an ELF + * image, and is almost certainly NOT what you want to use for any other + * purpose. + * + * IMPORTANT - VMA fields are accessed without an mmap lock being held, so the + * VMA referenced must not be linked in any user-visible tree, i.e. it must be a + * new VMA being mapped. + * + * The function assumes that addr is either contained within the VMA or below + * it, and makes no attempt to validate this value beyond that. + * + * Returns true if the read lock was obtained and a stack was perhaps expanded, + * false if the stack expansion failed. + * + * On stack expansion the function temporarily acquires an mmap write lock + * before downgrading it. + */ +bool mmap_read_lock_maybe_expand(struct mm_struct *mm, + struct vm_area_struct *new_vma, + unsigned long addr, bool write) +{ + if (!write || addr >= new_vma->vm_start) { + mmap_read_lock(mm); + return true; + } + + if (!(new_vma->vm_flags & VM_GROWSDOWN)) + return false; + + mmap_write_lock(mm); + if (expand_downwards(new_vma, addr)) { + mmap_write_unlock(mm); + return false; + } + + mmap_write_downgrade(mm); + return true; +} +#else +bool mmap_read_lock_maybe_expand(struct mm_struct *mm, struct vm_area_struct *vma, + unsigned long addr, bool write) +{ + return false; +} +#endif _ Patches currently in -mm which might be from lorenzo.stoakes@xxxxxxxxxx are mm-reinstate-ability-to-map-write-sealed-memfd-mappings-read-only.patch selftests-memfd-add-test-for-mapping-write-sealed-memfd-read-only.patch mm-correct-typo-in-mmap_state-macro.patch docs-mm-add-vma-locks-documentation.patch mm-vma-move-brk-internals-to-mm-vmac.patch mm-vma-move-unmapped_area-internals-to-mm-vmac.patch mm-abstract-get_arg_page-stack-expansion-and-mmap-read-lock.patch mm-vma-move-stack-expansion-logic-to-mm-vmac.patch mm-vma-move-__vm_munmap-to-mm-vmac.patch