Re: [PATCH] time64: Avoid undefined behaviour in timespec64_add()

[Arnd Bergmann]

On Mon, Feb 25, 2019 at 10:07 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> On Mon, Feb 25, 2019 at 10:01 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> > On Mon, Feb 25, 2019 at 5:53 AM Deepa Dinamani <deepa.kernel@xxxxxxxxx> wrote:
> > > On Sun, Feb 24, 2019 at 7:13 PM Hongbo Yao <yaohongbo@xxxxxxxxxx> wrote:

>
> diff --git a/drivers/net/ethernet/cadence/macb_ptp.c
> b/drivers/net/ethernet/cadence/macb_ptp.c
> index a6dc47edc4cf..0d5ebde29c0d 100644
> --- a/drivers/net/ethernet/cadence/macb_ptp.c
> +++ b/drivers/net/ethernet/cadence/macb_ptp.c
> @@ -160,7 +160,7 @@ static int gem_ptp_adjfine(struct ptp_clock_info
> *ptp, long scaled_ppm)
>  static int gem_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
>  {
>         struct macb *bp = container_of(ptp, struct macb, ptp_clock_info);
> -       struct timespec64 now, then = ns_to_timespec64(delta);
> +       struct timespec64 now;
>         u32 adj, sign = 0;
>
>         if (delta < 0) {
> @@ -170,7 +170,7 @@ static int gem_ptp_adjtime(struct ptp_clock_info
> *ptp, s64 delta)
>
>         if (delta > TSU_NSEC_MAX_VAL) {
>                 gem_tsu_get_time(&bp->ptp_clock_info, &now);
> -               now = timespec64_add(now, then);
> +               now = timespec64_add(now, ns_to_timespec64(delta));
>
>                 gem_tsu_set_time(&bp->ptp_clock_info,
>                                  (const struct timespec64 *)&now);

This part of of course completely bogus, I had first changed it
to timespec64_add_ns() for testing, and then back. Please ignore.

> diff --git a/drivers/net/ethernet/intel/i40e/i40e_ptp.c
> b/drivers/net/ethernet/intel/i40e/i40e_ptp.c
> index 5fb4353c742b..4efcba0252a4 100644
> --- a/drivers/net/ethernet/intel/i40e/i40e_ptp.c
> +++ b/drivers/net/ethernet/intel/i40e/i40e_ptp.c
> @@ -151,7 +151,7 @@ static int i40e_ptp_adjtime(struct ptp_clock_info
> *ptp, s64 delta)
>         mutex_lock(&pf->tmreg_lock);
>
>         i40e_ptp_read(pf, &now, NULL);
> -       timespec64_add_ns(&now, delta);
> +       timespec64_add_ns(&now, ns_to_timespec64(delta));
>         i40e_ptp_write(pf, (const struct timespec64 *)&now);
>
>         mutex_unlock(&pf->tmreg_lock);
>

I think this is still needed, I've added this to my y2038 backlog
and will send a patch for it eventually.

       Arnd