re: 9.1: boot-time delay? [WORKAROUND FOUND]

[matthew green <mrg%eterna.com.au@localhost>]

Manuel Bouyer writes:
> On Tue, May 25, 2021 at 10:46:04PM -0000, Michael van Elst wrote:
> > bouyer%antioche.eu.org@localhost (Manuel Bouyer) writes:
> > 
> > >Another issue could be mstohz() called with a delay too short;
> > >mstohz() will round it up to 1 tick.
> > 
> > 
> > #  define mstohz(ms) ((unsigned int)((ms + 0ul) * hz / 1000ul))
> > 
> > If mstohz() would round up to full ticks, it could actually avoid
> > some pitfalls. But it doesn't.
>
> indeed. But in this case the problem will show up with smaller hz, not
> larger. So I think we can rule out this case.

it's hztoms() that is the problem here.

#  define hztoms(t) ((unsigned int)(((t) + 0ul) * 1000ul / hz))

ah... this is the new one.  the old one was:

#define hztoms(t) \
     (__predict_false((t) >= 0x20000) ? \
         ((t +0u) / hz) * 1000u : \
         ((t +0u) * 1000u) / hz)

looks like christos fixed it in 2019.

revision 1.615
date: 2019-09-28 08:10:58 -0700;  author: christos;  state: Exp;  lines: +25 -12;  commitid: QCMKDk7BwqyE4NEB;
For 32 bit the mstohz and hztoms functions evaluate their parameter multiple
times. This is inefficient for cases like:
    unsigned ms = hztoms(MIN(timeout, mstohz(INT_MAX)));
Make them inline functions; also provide the 64 bit versions for them here
so all the LP64 machines can use them (before only amd64 and sparc64
specialized mstohz).
Make them both return unsigned int.

which explains why the change i made in mid-2019 that seemed to no
longer be necessary was no longer necessary!


.mrg.