On Thu, 8 Oct 2020 at 11:21, Maxim Uvarov <maxim.uva...@linaro.org> wrote:
>
> On Thu, 8 Oct 2020 at 05:43, Shashi Mallela <shashi.mall...@linaro.org> wrote:
> > +static void sbsa_gwdt_update_timer(SBSA_GWDTState *s)
> > +{
> > + uint64_t timeout = 0;
> > +
> > + if (s->enabled) {
> > + /*
> > + * Extract the upper 16 bits from woru & 32 bits from worl
> > + * registers to construct the 48 bit offset value
> > + */
> > + timeout = s->woru & SBSA_GWDT_WOR_MASK;
> > + timeout <<= 32;
> > + timeout |= s->worl;
> > + timeout = muldiv64(timeout, NANOSECONDS_PER_SECOND,
> > SBSA_TIMER_FREQ);
>
> static inline uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c)
> {
> return (__int128_t)a * b / c;
> }
>
> #define NANOSECONDS_PER_SECOND 1000000000LL
>
> Interesting why gcc does not warn on 64bit signed to 32bit unsigned
> truncation here. Looks like it's too smart to understand
> that value fits in 32 bits.
What truncation? 1000000000 in decimal is 0x3B9ACA00 in hex:
the number fits in an 32 bit integer without truncation.
(ns = muldiv64(ticks, NANOSECONDS_PER_SECOND, frequency) is a pretty
common pattern in our timer devices for converting a number of
ticks to a duration in nanoseconds, as is the reverse
conversion of ticks = muldiv64(ns, NANOSECONDS_PER_SECOND, frequency).)
thanks
-- PMM
