2010/3/3 Slava Pestov <[email protected]>
> A similar issue occurs on Windows under VirtualBox. Doug and I tried a
> workaround similar to yours but it doesn't work too well if the Factor
> side of things doesn't poll the nano-counter often enough.
>
Another solution would be to retry until the counter is indeed monotonic,
with some kind of hard limit, as in:
void factor_vm::primitive_nano_count()
{
- u64 nanos = nano_count();
- if(nanos < last_nano_count) critical_error("Monotonic counter
decreased",0);
- last_nano_count = nanos;
- ctx->push(from_unsigned_8(nanos));
+ // Jump through hoops to get a monotonic value even when
+ // running from inside VirtualBox.
+ // http://www.virtualbox.org/ticket/6318
+ for (int i = 0; i < 8; i++) {
+ u64 nanos = nano_count();
+ if(nanos < last_nano_count) continue;
+ last_nano_count = nanos;
+ ctx->push(from_unsigned_8(nanos));
+ return;
+ }
+ critical_error("Monotonic counter decreased",0);
}
as it looks like the counter sometimes goes backwards but doesn't jump far
ahead.
That should work best (at the price of more system calls in buggy situation)
but is not 100% foolproof, even if the probability of failure when the
system reports 8 misplaced values in a row is kinda low.
Perhaps the best thing to do is to report this to the VirtualBox guys
> and wait for them to fix it.
>
I reported it just after sending my mail:
http://www.virtualbox.org/ticket/6318
In the meantime, we can accomodate a local patch to Factor.
Sam
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Factor-talk mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/factor-talk
