ExternalVendorCode/genode
1
0
Fork 0
mirror of https://github.com/genodelabs/genode.git synced 2025-01-18 18:56:29 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

hw cortex-a9 timer: raise translation precision

Browse Source 
If we do the tics-to-us translation with one division and multiplication
over the whole argument, like before, we loose microseconds granularity
although the timer frequency allows for such granularity. Thus, we treat
the most significant half and the least significant half of the tics
value separate. Each half is shifted to the best bit position for the
translation, then translated, and then shifted back.

Ref #2347
This commit is contained in:
Martin Stein 2017-03-16 17:57:57 +01:00 committed by Christian Helmuth
parent 0b30cf0ab6
commit 05655a9d88
1 changed files with 30 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
repos/base-hw/src/core/include/spec/cortex_a9/timer.h
View File

@ -85,8 +85,36 @@ class Genode::Timer : public Mmio
*/
void start_one_shot(time_t const tics, unsigned const);
time_t tics_to_us(time_t const tics) const {
return (tics / TICS_PER_MS) * 1000; }
time_t tics_to_us(time_t const tics) const
{
/*
* If we would do the translation with one division and
* multiplication over the whole argument, we would loose
* microseconds granularity although the timer frequency would
* allow for such granularity. Thus, we treat the most significant
* half and the least significant half of the argument separate.
* Each half is shifted to the best bit position for the
* translation, then translated, and then shifted back.
*/
static_assert(TICS_PER_MS >= 1000, "Bad TICS_PER_MS value");
enum {
TICS_WIDTH = sizeof(time_t) * 8,
TICS_HALF_WIDTH = TICS_WIDTH / 2,
TICS_MSB_MASK = ~0UL << TICS_HALF_WIDTH,
TICS_LSB_MASK = ~0UL >> TICS_HALF_WIDTH,
TICS_MSB_RSHIFT = 10,
TICS_LSB_LSHIFT = TICS_HALF_WIDTH - TICS_MSB_RSHIFT,
};
time_t const tics_msb = (tics & TICS_MSB_MASK) >> TICS_MSB_RSHIFT;
time_t const tics_lsb = (tics & TICS_LSB_MASK) << TICS_LSB_LSHIFT;
time_t const us_msb = ((tics_msb * 1000) / TICS_PER_MS) << TICS_MSB_RSHIFT;
time_t const us_lsb = ((tics_lsb * 1000) / TICS_PER_MS) >> TICS_LSB_LSHIFT;
return us_msb | us_lsb;
}
time_t us_to_tics(time_t const us) const {
return (us / 1000) * TICS_PER_MS; }