Date: Sun, 27 Apr 2003 13:36:44 -0700 (PDT) From: Marcel Moolenaar <marcel@FreeBSD.org> To: Perforce Change Reviews <perforce@freebsd.org> Subject: PERFORCE change 29892 for review Message-ID: <200304272036.h3RKai2j075274@repoman.freebsd.org>
index | next in thread | raw e-mail
http://perforce.freebsd.org/chv.cgi?CH=29892 Change 29892 by marcel@marcel_pluto2 on 2003/04/27 13:36:37 Step 1: Properly announce a constant timer as a bad timer. A constant timer is one that is both monotonically increasing and monotonically decreasing at the same time. This fixes the announcement on the HP rx2600 where we basicly read a bogus register due to not properly using the v2 ACPI tables and invariable read the same value. We were quite happy with this even if it made time stop completely when booting a SMP kernel. The 100% CPU usage for ntpd in an attempt to make time advance didn't make us unhappy. Except Kerberos: the time skew prevented logging in... Step 2 is to fix getting the ACPI timer info from the tables. This is in progress... Affected files ... .. //depot/projects/ia64/sys/dev/acpica/acpi_timer.c#9 edit Differences ... ==== //depot/projects/ia64/sys/dev/acpica/acpi_timer.c#9 (text+ko) ==== @@ -129,14 +129,14 @@ } if (max - min > 2) n = 0; - else if (min < 0) + else if (min < 0 || max == 0) n = 0; else n = 1; if (bootverbose) printf("ACPI timer looks %s min = %d, max = %d, width = %d\n", n ? "GOOD" : "BAD ", - min, max, max - min + 1); + min, max, max - min); return (n); }help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200304272036.h3RKai2j075274>