Re: mount on tmpfs failing to parse context option

[Laura Abbott]

On 10/7/19 9:26 PM, Al Viro wrote:
> On Mon, Oct 07, 2019 at 05:50:31PM -0700, Hugh Dickins wrote:
>
> [sorry for being MIA - had been sick through the last week, just digging
> myself from under piles of mail; my apologies]
>
> > (tmpfs, very tiresomely, supports a NUMA "mpol" mount option which can
> > have commas in it e.g "mpol=bind:0,2": which makes all its comma parsing
> > awkward.  I assume that where the new mount API commits bend over to
> > accommodate that peculiarity, they end up mishandling the comma in
> > the context string above.)
>
> 	Dumber than that, I'm afraid.  mpol is the reason for having
> ->parse_monolithic() in the first place, all right, but the problem is
> simply the lack of security_sb_eat_lsm_opts() call in it.
>
> 	Could you check if the following fixes that one?
>
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 0f7fd4a85db6..8dcc8d04cbaf 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -3482,6 +3482,12 @@ static int shmem_parse_options(struct fs_context *fc, void *data)
>   {
>   	char *options = data;
>   
> +	if (options) {
> +		int err = security_sb_eat_lsm_opts(options, &fc->security);
> +		if (err)
> +			return err;
> +	}
> +
>   	while (options != NULL) {
>   		char *this_char = options;
>   		for (;;) {

Yes the reporter says that works.

Thanks,
Laura