On 10/04/2026 08:07, Michael Kelly wrote:
On 10/04/2026 05:33, Amos Jeffries wrote:
Better make the glibc depend on existence of the new
MACH_ADJTIME_SECS_INVALID macro to enable support at built-time.
Yes, that's a good precaution, thanks. I was also concerned that
someone could run a new glibc with an old gnumach and so actually end
up adjusting the time by that huge value. Can that scenario can be
prevented ?
Thanks for the feedback, Amos. There are other issues:
1) MACH_ADJTIME_SECS_INVALID has the wrong value (by far!). The total
adjustment (in microseconds) is actually maintained as a signed int
within gnumach and the maximum adjustment is plus or minus 2145 seconds.
2) There seems to be something amiss with certain time adjustments. This
is an existing issue that is perhaps connected with signed/unsigned or
64 to 32 bit conversions that are occurring. I'll be checking this and
fixing if necessary.
I'll issue a revised v2 patch in the days ahead.
Mike.
