[CRIU] [PATCHv6 22/36] x86/vdso: Add offsets page in vvar

[Thomas Gleixner]

On Thu, 15 Aug 2019, Dmitry Safonov wrote:
> ---
>  arch/Kconfig                          |  5 +++
>  arch/x86/Kconfig                      |  1 +
>  arch/x86/entry/vdso/vdso-layout.lds.S |  9 ++++-
>  arch/x86/entry/vdso/vdso2c.c          |  3 ++
>  arch/x86/entry/vdso/vma.c             | 12 +++++++
>  arch/x86/include/asm/vdso.h           |  1 +
>  init/Kconfig                          |  1 +
>  lib/vdso/gettimeofday.c               | 47 +++++++++++++++++++++++++++

This needs to be split into the generic lib/vdso part and then x86 making
use of it.

> +#ifdef CONFIG_TIME_NS

This should be COMPILE_WITH_TIME_NS and not CONFIG_TIME_NS

> +extern u8 timens_page
> +	__attribute__((visibility("hidden")));
> +
> +notrace static __always_inline void clk_to_ns(clockid_t clk, struct __kernel_timespec *ts)

This needs notrace because?

> +{
> +	struct timens_offsets *timens = (struct timens_offsets *) &timens_page;
> +	struct timespec64 *offset64;
> +
> +	switch (clk) {
> +	case CLOCK_MONOTONIC:
> +	case CLOCK_MONOTONIC_COARSE:
> +	case CLOCK_MONOTONIC_RAW:
> +		offset64 = &timens->monotonic;
> +		break;
> +	case CLOCK_BOOTTIME:
> +		offset64 = &timens->boottime;
> +		break;
> +	default:
> +		return;
> +	}
> +
> +	/*
> +	 * The kernel allows to set a negative offset only if the current clock
> +	 * value in a namespace is positive, so the result tv_sec can't be
> +	 * negative here.
> +	 */
> +	ts->tv_nsec += offset64->tv_nsec;
> +	ts->tv_sec += offset64->tv_sec;
> +	if (ts->tv_nsec >= NSEC_PER_SEC) {
> +		ts->tv_nsec -= NSEC_PER_SEC;
> +		ts->tv_sec++;
> +	}
> +	if (ts->tv_nsec < 0) {
> +		ts->tv_nsec += NSEC_PER_SEC;
> +		ts->tv_sec--;
> +	}

That's broken for 32bit user space on 64bit hosts. On LE due to
misalignment and on BE because 32bit will read always 0.

> +}
> +#else
> +notrace static __always_inline void clk_to_ns(clockid_t clk, struct __kernel_timespec *ts) {}
> +#endif
> +
>  static int do_hres(const struct vdso_data *vd, clockid_t clk,
>  		   struct __kernel_timespec *ts)
>  {
> @@ -65,6 +108,8 @@ static int do_hres(const struct vdso_data *vd, clockid_t clk,
>  	ts->tv_sec = sec + __iter_div_u64_rem(ns, NSEC_PER_SEC, &ns);
>  	ts->tv_nsec = ns;
>  
> +	clk_to_ns(clk, ts);
> +
>  	return 0;
>  }
>  
> @@ -79,6 +124,8 @@ static void do_coarse(const struct vdso_data *vd, clockid_t clk,
>  		ts->tv_sec = vdso_ts->sec;
>  		ts->tv_nsec = vdso_ts->nsec;
>  	} while (unlikely(vdso_read_retry(vd, seq)));
> +
> +	clk_to_ns(clk, ts);
>  }
>  
>  static __maybe_unused int
> -- 
> 2.22.0
> 
>