Re: [PATCH] mm/mmap: Properly unaccount memory on mas_preallocate() failure

[Liam Howlett <liam.howlett@oracle.com>]

* Alistair Popple <apopple@nvidia.com> [221201 23:54]:
> security_vm_enough_memory_mm() accounts memory via a call to
> vm_acct_memory(). Therefore any subsequent failures should unaccount
> for this memory prior to returning the error.
> 
> Signed-off-by: Alistair Popple <apopple@nvidia.com>
> Fixes: 28c5609fb236 ("mm/mmap: preallocate maple nodes for brk vma expansion")

Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>

> 
> ---
> 
> Pretty trivial fixup that I noticed was needed while working on the
> code.
> ---
>  mm/mmap.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 74a84eb33b90..3e50a571c3c4 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -2954,7 +2954,7 @@ static int do_brk_flags(struct ma_state *mas, struct vm_area_struct *vma,
>  	    ((vma->vm_flags & ~VM_SOFTDIRTY) == flags)) {
>  		mas_set_range(mas, vma->vm_start, addr + len - 1);
>  		if (mas_preallocate(mas, vma, GFP_KERNEL))
> -			return -ENOMEM;
> +			goto unacct_fail;
>  
>  		vma_adjust_trans_huge(vma, vma->vm_start, addr + len, 0);
>  		if (vma->anon_vma) {
> @@ -2976,7 +2976,7 @@ static int do_brk_flags(struct ma_state *mas, struct vm_area_struct *vma,
>  	/* create a vma struct for an anonymous mapping */
>  	vma = vm_area_alloc(mm);
>  	if (!vma)
> -		goto vma_alloc_fail;
> +		goto unacct_fail;
>  
>  	vma_set_anonymous(vma);
>  	vma->vm_start = addr;
> @@ -3001,7 +3001,7 @@ static int do_brk_flags(struct ma_state *mas, struct vm_area_struct *vma,
>  
>  mas_store_fail:
>  	vm_area_free(vma);
> -vma_alloc_fail:
> +unacct_fail:
>  	vm_unacct_memory(len >> PAGE_SHIFT);
>  	return -ENOMEM;
>  }
> -- 
> 2.35.1
>