gettimeofday() has 1-second granularity

[Jiri Svoboda]

It appears that when I call gettimeofday() and a clock device is available (ia32 in Qemu, so that it does not merely call getuptime()), the time returned has 1-second granularity (I always get useconds == 0).

[Jakub Jermář]

This issue is preventing Jainja JVM from working properly:

<guillaum11> I found the reason of the strange behavior of Jainja.
<guillaum11> gettimeofday in uspace/lib/c/generic/time.c has a granularity of 1 second.
<guillaum11> tv->tv_usec is always 0 on my build of HelenOS 0.6
...
<guillaum11> It seems to be voluntary implemented as is:
<guillaum11> tv->tv_usec = 0;
<guillaum11> tv->tv_sec = mktime(&time);
...
<jermar_> guillaum11, what kind of problems is this causing?
<guillaum11> jermar_: green threads doesn't work well because it needs accurate time

[Jakub Jermář]

Fixed in mainline,2302.