On Tue, 17 Nov 2015 18:26:38 +0100 Piotr Kwapulinski <kwapulinski.piotr@xxxxxxxxx> wrote:
> The following flag comparison in mmap_region is not fully correct:
>
> if (!(vm_flags & MAP_FIXED))
>
> The vm_flags should not be compared with MAP_FIXED (0x10). It is a bit
> confusing. This condition is almost always true since VM_MAYREAD (0x10)
> flag is almost always set by default. This patch removes this condition.
>
> ...
>
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1547,13 +1547,6 @@ unsigned long mmap_region(struct file *file, unsigned long addr,
> if (!may_expand_vm(mm, len >> PAGE_SHIFT)) {
> unsigned long nr_pages;
>
> - /*
> - * MAP_FIXED may remove pages of mappings that intersects with
> - * requested mapping. Account for the pages it would unmap.
> - */
> - if (!(vm_flags & MAP_FIXED))
> - return -ENOMEM;
> -
> nr_pages = count_vma_pages_range(mm, addr, addr + len);
>
> if (!may_expand_vm(mm, (len >> PAGE_SHIFT) - nr_pages))
That looks simpler.
However the changelog doesn't describe the end-user visible effects of
the bug, as changelogs should always do. Presumably this is causing
incorrect ENOMEM reporting due to RLIMIT_AS being exceeded, but this
isn't very specific.
So can you please fill in the details here? Such info is needed when
deciding which kernel version(s) need the fix.
Thanks.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>