[PATCH v2 1/5] fs: use raw_copy_from_user() to copy mount() data

Dmitry Vyukov dvyukov at google.com
Mon Nov 22 23:50:55 PST 2021


On Tue, 23 Nov 2021 at 06:17, Peter Collingbourne <pcc at google.com> wrote:
>
> With uaccess logging the contract is that the kernel must not report
> accessing more data than necessary, as this can lead to false positive
> reports in downstream consumers. This generally works out of the box
> when instrumenting copy_{from,to}_user(), but with the data argument
> to mount() we use copy_from_user() to copy PAGE_SIZE bytes (or as
> much as we can, if the PAGE_SIZE sized access failed) and figure out
> later how much we actually need.
>
> To prevent this from leading to a false positive report, use
> raw_copy_from_user(), which will prevent the access from being logged.
> Recall that it is valid for the kernel to report accessing less
> data than it actually accessed, as uaccess logging is a best-effort
> mechanism for reporting uaccesses.
>
> Link: https://linux-review.googlesource.com/id/I5629b92a725c817acd9a861288338dd605cafee6
> Signed-off-by: Peter Collingbourne <pcc at google.com>
> ---
>  fs/namespace.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/fs/namespace.c b/fs/namespace.c
> index 659a8f39c61a..695b30e391f0 100644
> --- a/fs/namespace.c
> +++ b/fs/namespace.c
> @@ -3197,7 +3197,12 @@ static void *copy_mount_options(const void __user * data)
>         if (!copy)
>                 return ERR_PTR(-ENOMEM);
>
> -       left = copy_from_user(copy, data, PAGE_SIZE);
> +       /*
> +        * Use raw_copy_from_user to avoid reporting overly large accesses in
> +        * the uaccess buffer, as this can lead to false positive reports in
> +        * downstream consumers.
> +        */
> +       left = raw_copy_from_user(copy, data, PAGE_SIZE);

This will skip KASAN/etc checks as well, right? I guess it is fine b/c
this affects just this place and the code looks safe (famous last
words :)) and we can refine it in future.
But I wonder about false positives under KMSAN. However, we probably
can add an explicit KMSAN annotation to mark it as initialised.
Alex?

>         /*
>          * Not all architectures have an exact copy_from_user(). Resort to
> --
> 2.34.0.rc2.393.gf8c9666880-goog
>



More information about the linux-arm-kernel mailing list