Re: [PATCH v3 2/7] drm: Add drm_timeout_rel_to_jiffies()

[Jani Nikula <jani.nikula@linux.intel.com>]

On Tue, 15 Sep 2026, Maíra Canal <mcanal@igalia.com> wrote:
> drm_timeout_abs_to_jiffies() covers the drivers whose wait UAPI takes an
> absolute deadline, but there is no equivalent for the drivers that
> express a wait as a duration. Drivers such as i915 and v3d convert the
> value themselves.
>
> Converting a nanosecond duration to jiffies needs some care.
> nsecs_to_jiffies() returns unsigned long, so on 32-bit a large
> userspace-supplied timeout overflows its range and is silently truncated.
>
> i915 already handles both cases in a local helper, which v3d has a copy
> of. Add the same conversion to the core, so that it is available to any
> driver and both copies can be dropped.
>
> Signed-off-by: Maíra Canal <mcanal@igalia.com>
> ---
>  drivers/gpu/drm/drm_timeout.c | 45 +++++++++++++++++++++++++++++++++++++++++++
>  include/drm/drm_timeout.h     |  1 +
>  2 files changed, 46 insertions(+)
>
> diff --git a/drivers/gpu/drm/drm_timeout.c b/drivers/gpu/drm/drm_timeout.c
> index 35e10293e0a1..ee545475429c 100644
> --- a/drivers/gpu/drm/drm_timeout.c
> +++ b/drivers/gpu/drm/drm_timeout.c
> @@ -9,6 +9,7 @@
>  #include <linux/export.h>
>  #include <linux/jiffies.h>
>  #include <linux/ktime.h>
> +#include <linux/math64.h>
>  #include <linux/sched.h>
>  
>  #include <drm/drm_timeout.h>
> @@ -48,3 +49,47 @@ signed long drm_timeout_abs_to_jiffies(int64_t timeout_nsec)
>  	return timeout_jiffies64 + 1;
>  }
>  EXPORT_SYMBOL(drm_timeout_abs_to_jiffies);
> +
> +/**
> + * drm_timeout_rel_to_jiffies - calculate jiffies timeout from relative value
> + * @timeout_nsec: relative timeout in ns, 0 for poll
> + *
> + * Calculate the timeout in jiffies from a relative timeout in ns, for drivers
> + * whose UAPI expresses a wait as a duration rather than as a deadline.
> + *
> + * The result is clamped to MAX_JIFFY_OFFSET. That keeps it positive once it is
> + * converted to the signed long taken by dma_fence_wait_timeout() and friends,
> + * which matters on 32-bit, and keeps it distinct from MAX_SCHEDULE_TIMEOUT so
> + * that a finite wait is never understood as an infinite one.
> + *
> + * It's strongly discouraged to use relative timeouts in uAPIs, as they do not
> + * survive a restarted ioctl. A signal-interrupted ioctl is re-entered with the
> + * same arguments, so the duration starts counting from zero again. New uAPIs
> + * should take an absolute deadline and use drm_timeout_abs_to_jiffies().
> + *
> + * Returns:
> + * 0 if @timeout_nsec is 0. Otherwise the equivalent number of jiffies and
> + * clamped to MAX_JIFFY_OFFSET.
> + */
> +unsigned long drm_timeout_rel_to_jiffies(u64 timeout_nsec)
> +{
> +	u64 secs;
> +	u32 rem;
> +
> +	/* Make 0 timeout means poll, as for the absolute variant. */
> +	if (timeout_nsec == 0)
> +		return 0;
> +
> +	/*
> +	 * As nsecs_to_jiffies64() does not guard against overflow, split
> +	 * the timeout into whole seconds and nanoseconds. This way
> +	 * nsecs_to_jiffies64() is always handed a value below a second.
> +	 */
> +	secs = div_u64_rem(timeout_nsec, NSEC_PER_SEC, &rem);
> +	if (secs >= MAX_JIFFY_OFFSET / HZ)
> +		return MAX_JIFFY_OFFSET;
> +
> +	return min_t(u64, MAX_JIFFY_OFFSET,
> +		     secs_to_jiffies(secs) + nsecs_to_jiffies64(rem) + 1);
> +}
> +EXPORT_SYMBOL(drm_timeout_rel_to_jiffies);
> diff --git a/include/drm/drm_timeout.h b/include/drm/drm_timeout.h
> index cd9621c52062..6ee222a3e97c 100644
> --- a/include/drm/drm_timeout.h
> +++ b/include/drm/drm_timeout.h
> @@ -12,5 +12,6 @@
>  #include <linux/types.h>
>  
>  signed long drm_timeout_abs_to_jiffies(s64 timeout_nsec);
> +unsigned long drm_timeout_rel_to_jiffies(u64 timeout_nsec);

The signed/unsigned differences here bug me a little. Maybe explain why
in the commit message if there's a rationale?

I note that drm_timeout_rel_to_jiffies() return value is assigned to
unsigned long variables half the time. I guess its kernel-doc could say
it never returns negative values?

BR,
Jani.




>  
>  #endif

-- 
Jani Nikula, Intel