[PATCH] mm/mremap: Fix vma iterator initialization

From: Liam R. Howlett
Date: Thu Jan 26 2023 - 16:20:38 EST


The vma iterator location is incorrect when there is a failure in the
move_vma() function which alters the address being modified.

Delay the initialization of the vma iterator until the address is
stable.

Reported-by: Sanan Hasanov <sanan.hasanov@xxxxxxxxxxxxxxx>
Link: https://lore.kernel.org/linux-mm/IA1PR07MB98306BC0F55667A760EABE91ABCE9@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
Signed-off-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>
---
mm/mremap.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

Andrew,

This can be placed next to ("mm: change munmap splitting order and
move_vma()") and squashed before sending upstream.

diff --git a/mm/mremap.c b/mm/mremap.c
index 1bc81afd90de..6c7f49ab7d19 100644
--- a/mm/mremap.c
+++ b/mm/mremap.c
@@ -585,7 +585,7 @@ static unsigned long move_vma(struct vm_area_struct *vma,
unsigned long hiwater_vm;
int err = 0;
bool need_rmap_locks;
- VMA_ITERATOR(vmi, mm, old_addr);
+ struct vma_iterator vmi;

/*
* We'd prefer to avoid failure later on in do_munmap:
@@ -701,6 +701,7 @@ static unsigned long move_vma(struct vm_area_struct *vma,
return new_addr;
}

+ vma_iter_init(&vmi, mm, old_addr);
if (do_vmi_munmap(&vmi, mm, old_addr, old_len, uf_unmap, false) < 0) {
/* OOM: unable to split vma, just get accounts right */
if (vm_flags & VM_ACCOUNT && !(flags & MREMAP_DONTUNMAP))
--
2.39.0