Date: Tue, 26 Sep 2006 12:05:57 +0300 From: Viktor Rosendahl <viktor.rosendahl@saunalahti.fi> To: stable@freebsd.org Subject: /usr/src/sys/i386/i386/local_apic.c and status of ULE Message-ID: <200609261205.57901.viktor.rosendahl@saunalahti.fi>
next in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Hi all,
I wonder, what is the status of the ULE scheduler? Is it supposed to work at
all?
I upgraded my -stable system some time ago and it started to hang at the end
of the boot process, right before the mounting of the root device. I found
out that the immediate cause was a change
in /usr/src/sys/i386/i386/local_apic.c.
I attach a patch that reverts the change, so that you can see what I am
talking about. However, it only hangs if I use the ULE scheduler. That's why
I wonder whether the ULE scheduler is supposed to work or not.
Now, I understand that it might very well depend on some additional local
condition. Thus, I also attach my dmesg (of a successful boot) and my kernel
configuration file.
best regards,
Viktor
[-- Attachment #2 --]
diff -Naur src/sys/i386/i386/local_apic.c src.fixed/sys/i386/i386/local_apic.c
--- src/sys/i386/i386/local_apic.c Sun Sep 24 02:56:22 2006
+++ src.fixed/sys/i386/i386/local_apic.c Sun Mar 19 13:53:42 2006
@@ -610,18 +610,6 @@
lapic_handle_timer(struct clockframe frame)
{
struct lapic *la;
-
-#ifdef SMP
- /*
- * Don't do any accounting for the disabled HTT cores, since it
- * will provide misleading numbers for the userland.
- *
- * No locking is necessary here, since even if we loose the race
- * when hlt_cpus_mask changes it is not a big deal, really.
- */
- if ((hlt_cpus_mask & (1 << PCPU_GET(cpuid))) != 0)
- return;
-#endif
la = &lapics[PCPU_GET(apic_id)];
(*la->la_timer_count)++;
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200609261205.57901.viktor.rosendahl>