Date: Tue, 15 Jun 1999 21:22:26 +0200 From: Poul-Henning Kamp <phk@critter.freebsd.dk> To: Tor.Egge@fast.no Cc: jgreco@ns.sol.net, james@ehlo.com, freebsd-smp@FreeBSD.ORG, aaron-fbsd@mutex.org Subject: Re: CPU states at 0.0% on 3.2-R SMP box ? Message-ID: <7236.929474546@critter.freebsd.dk> In-Reply-To: Your message of "Tue, 15 Jun 1999 21:02:10 %2B0200." <199906151902.VAA96921@midten.fast.no>
next in thread | previous in thread | raw e-mail | index | archive | help
I think that this kind of surgery should be avoided. It is far
too much hazzle to be worth the effort. If the bios traps RTC
access, we shouldn't use it for statistics anyway, since it will
no longer have decent properties for statistical sampling.
Poul-Henning
In message <199906151902.VAA96921@midten.fast.no>, Tor.Egge@fast.no writes:
>----Next_Part(Tue_Jun_15_20:53:50_1999)--
>Content-Type: Text/Plain; charset=us-ascii
>Content-Transfer-Encoding: 7bit
>
>>
>> I would advise that you first try adding the line
>>
>> device apm0 at isa? flags 0x20 # Advanced Power Management
>
>That disables the RTC interrupt and causes hardclock() to call statclock().
>
>Some bioses seems to trap direct accesses to the RTC chip in order to
>pass Y2K tests (Some Y2K test programs access the RTC chip directly).
>
>When the CPU accesses IO port 0x70, the power management chip
>generates an SMI, and the physical RTC chip is accessed from inside
>SMM.
>
>This method sometimes fails when using an SMP kernel. A workaround is
>to disable the trap SMI for access to IO port 0x70.
>
>When using an SMP kernel on Asus P2B-DS mainboard with BIOS revision
>1007 or newer, you either have to disable use of RTC interrupts or
>disable the trap SMIs.
>
>- Tor Egge
>
>
>----Next_Part(Tue_Jun_15_20:53:50_1999)--
>Content-Type: Text/Plain; charset=us-ascii
>Content-Transfer-Encoding: 7bit
>
>Index: sys/pci/pcisupport.c
>===================================================================
>RCS file: /home/ncvs/src/sys/pci/pcisupport.c,v
>retrieving revision 1.86.2.8
>diff -u -r1.86.2.8 pcisupport.c
>--- pcisupport.c 1999/05/26 16:39:42 1.86.2.8
>+++ pcisupport.c 1999/06/06 20:31:02
>@@ -213,7 +213,87 @@
> tag->secondarybus = tag->subordinatebus = secondarybus + 1;
> }
>
>+#ifdef SMP
> static void
>+fix_82371ab_power_management(pcici_t tag)
>+{
>+ int pmba;
>+ int devctl;
>+ int devctl_changed;
>+ int devrese;
>+ int devresg;
>+
>+#define PMBA_CONFIG_OFFSET 0x40
>+#define PMBA_TO_IO(addr) (addr & 0xffc0)
>+#define DEVRESE_OFFSET 0x68
>+#define DEVRESG_OFFSET 0x70
>+#define DEVRES_MONITOR_ENABLE (1 << 20)
>+#define DEVCTL_IO_OFFSET 0x2c
>+#define DEVCTL_TRAP_DEV12 (1 << 24)
>+#define DEVCTL_TRAP_DEV13 (1 << 25)
>+#define RTC_IOADDR 0x70
>+
>+ pmba = pci_cfgread(tag, PMBA_CONFIG_OFFSET, 4);
>+
>+ devctl = inl(PMBA_TO_IO(pmba) + DEVCTL_IO_OFFSET);
>+ devctl_changed = 0;
>+
>+ devrese = pci_cfgread(tag, DEVRESE_OFFSET, 2) |
>+ (pci_cfgread(tag, DEVRESE_OFFSET + 2, 1) << 16);
>+
>+ if ((devrese & DEVRES_MONITOR_ENABLE) != 0 &&
>+ ((devrese >> 16) & 15) <= 7 &&
>+ (devrese & 0xffff) == RTC_IOADDR &&
>+ (devctl & DEVCTL_TRAP_DEV12) != 0) {
>+
>+ devrese &= ~DEVRES_MONITOR_ENABLE;
>+ devctl &= ~DEVCTL_TRAP_DEV12;
>+ devctl_changed = 1;
>+
>+ pci_cfgwrite(tag, DEVRESE_OFFSET, 2, (devrese & 0xffff));
>+ pci_cfgwrite(tag, DEVRESE_OFFSET + 2, 1,
>+ (devrese >> 16) & 0xff);
>+
>+ printf("Disabled Device 12 trap SMI for access to RTC chip\n");
>+ }
>+
>+ devresg = pci_cfgread(tag, DEVRESG_OFFSET, 2) |
>+ (pci_cfgread(tag, DEVRESG_OFFSET + 2, 1) << 16);
>+
>+ if ((devresg & DEVRES_MONITOR_ENABLE) != 0 &&
>+ ((devresg >> 16) & 15) <= 7 &&
>+ (devresg & 0xffff) == RTC_IOADDR &&
>+ (devctl & DEVCTL_TRAP_DEV13) != 0) {
>+
>+ devresg &= ~DEVRES_MONITOR_ENABLE;
>+ devctl &= ~DEVCTL_TRAP_DEV13;
>+ devctl_changed = 1;
>+
>+ pci_cfgwrite(tag, DEVRESG_OFFSET, 2, (devresg & 0xffff));
>+ pci_cfgwrite(tag, DEVRESG_OFFSET + 2, 1,
>+ (devresg >> 16) & 0xff);
>+
>+ printf("Disabled Device 13 trap SMI for access to RTC chip\n");
>+ }
>+
>+ if (devctl_changed != 0) {
>+ outl(PMBA_TO_IO(pmba) + DEVCTL_IO_OFFSET, devctl);
>+ }
>+
>+#undef PMBA_CONFIG_OFFSET
>+#undef PMBA_TO_IO
>+#undef DEVRESE_OFFSET
>+#undef DEVRESG_OFFSET
>+#undef DEVRES_MONITOR_ENABLE
>+#undef DEVCTL_IO_OFFSET
>+#undef DEVCTL_TRAP_DEV12
>+#undef DEVCTL_TRAP_DEV13
>+#undef RTC_IOADDR
>+}
>+#endif
>+
>+
>+static void
> fixwsc_natoma(pcici_t tag)
> {
> int pmccfg;
>@@ -284,6 +364,9 @@
> case 0x71108086:
> return ("Intel 82371AB PCI to ISA bridge");
> case 0x71138086:
>+#if defined(SMP)
>+ fix_82371ab_power_management(tag);
>+#endif
> return ("Intel 82371AB Power management controller");
> case 0x71808086:
> return ("Intel 82443LX host to PCI bridge");
>
>----Next_Part(Tue_Jun_15_20:53:50_1999)----
>
>
>To Unsubscribe: send mail to majordomo@FreeBSD.org
>with "unsubscribe freebsd-smp" in the body of the message
>
--
Poul-Henning Kamp FreeBSD coreteam member
phk@FreeBSD.ORG "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?7236.929474546>