![](https://seccdn.libravatar.org/avatar/3035b38ff33cf86f480bb169b8500b80.jpg?s=120&d=mm&r=g)
https://bugzilla.novell.com/show_bug.cgi?id=399342
User koenig@linux.de added comment
https://bugzilla.novell.com/show_bug.cgi?id=399342#c11
Harald Koenig
Do have any news around? Btw: which HZ values in user and kernel space are used on this Toshiba Portege?
it's the plain newly installed 11.0, noting special using the tickless kernel -- so HZ is somewhat... here is the problem from debugging procinfo: in get_HZ() clock_getres() returns 1e-9 for the tickless kernel giving sys_hz = (long) 1e9 with this huge value of sys_hz in procinfo.c:465 elapsed = (DIFF(intr[0]) * usr_hz) / sys_hz; gives elapsed = 2 (for tickless kernel the DIFF(intr[0]) doesn't really make much sense anymore). elapsed=2 is passed to perc() as t=2 causing a divide by zero in the last term of 438 v = (unsigned int) (i < 1000000 ? 439 ((1000 * i + t / 2) / t) : 440 ((i + t / 2000) / (t / 1000))); with i=2729038 using the following small patch "fixes" procinfo for my i386 system: ------------------------------------------------------------------ --- procinfo.c~ 2008-06-17 20:39:39.000000000 +0200 +++ procinfo.c 2008-06-17 21:24:41.000000000 +0200 @@ -459,12 +459,14 @@ /* XXX Is this stuff still relevant/true? */ +#ifdef NO_TICKLESS_KERNEL #ifdef __i386__ /* IRQ 0 is timer tick on i386's... */ if (nr_irqs && new.intr[0]) { if (fs && old.uptime) elapsed = (DIFF(intr[0]) * usr_hz) / sys_hz; } else #endif +#endif #ifdef __sparc__ /* IRQ 10 is timer tick on sparc's... */ if (nr_irqs && new.intr[10]) { if (fs && old.uptime) ------------------------------------------------------------------ obviously this is a 32bit-only problem and you have to run the system for a while to get vaules > 1e6 to be displayed... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.