On Tue, 15 Sep 2026, Maíra Canal <[email protected]> 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 <[email protected]> > --- > 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
