On Sat, 2010-06-05 at 22:48 +0200, Jan Kiszka wrote: > Gilles Chanteperdrix wrote: > > Jan Kiszka wrote: > >> Philippe Gerum wrote: > >>> On Thu, 2010-06-03 at 16:22 +0200, Jan Kiszka wrote: > >>>> From: Jan Kiszka <[email protected]> > >>>> > >>>> Implement the x86 arch bits for ipipe_get_irq_regs support. This allows > >>>> to drop __ipipe_tick_regs and use the new service instead. > >>> I'm unsure whether this patch would actually replace __ipipe_tick_regs > >>> properly, particularly regarding how the profiling code works. > >> tick_regs are a "workaround", this approach appears to me way closer to > >> how native works. > > > > We do not want ipipe_tick_regs to work the way the native kernel works. > > We want to fool the kernel by passing it the value of the registers at > > the moment of the real timer tick so that its accounting works more or > > less reliably. When the kernel used the real registers it got the > > accounting wrong. > > Right, though current tick_regs approach is not accurate either: the > preempted task gets the full time slice accounted, neglecting the actual > schedule inside the preempting domain. >
Indeed. This trade-off was made back in 2002. > > Jan > > _______________________________________________ > Adeos-main mailing list > [email protected] > https://mail.gna.org/listinfo/adeos-main -- Philippe. _______________________________________________ Adeos-main mailing list [email protected] https://mail.gna.org/listinfo/adeos-main
