public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: William Lee Irwin III <wli@holomorphy.com>
To: Zwane Mwaikambo <zwane@fsmlabs.com>
Cc: Eric BEGOT <eric_begot@yahoo.fr>, Andrew Morton <akpm@osdl.org>,
	Linux Kernel <linux-kernel@vger.kernel.org>
Subject: Re: 2.6.7-rc3-mm1
Date: Wed, 9 Jun 2004 07:48:09 -0700	[thread overview]
Message-ID: <20040609144809.GK1444@holomorphy.com> (raw)
In-Reply-To: <Pine.LNX.4.58.0406090942420.1838@montezuma.fsmlabs.com>

On Wed, 9 Jun 2004, William Lee Irwin III wrote:
>> As curious as I am as to whether that works, I'm also curious as to
>> whether it's the culprit in this case. Eric, could you also describe your
>> system? A dmesg from a working kernel would also help.

On Wed, Jun 09, 2004 at 09:43:58AM -0400, Zwane Mwaikambo wrote:
> I'm leaning more towards the cpumask stuff but at least eliminating your
> patch being a culprit gets us further.
> 	Zwane

A moment's reflection renders this obvious. phys_cpu_present_map needs
to be doublechecked for the invalid physical APIC ID's in
cpu_present_to_apicid() prior to waking cpus since I'm marking invalid
physical APIC ID's as present to drive the IO-APIC physical APIC ID
validity detection and cpuid validity + detection logic. Something like
this should help a bit.

I'm questioning whether the marking scheme is worth anything and if I
should just rely on bounds-checking against the dynamically-detected
physical APIC ID instead.

(a) attempt to fix zwane's numaq; m->mpc_apicid is wrong to look at.
(b) compilefix for CONFIG_ACPI_BOOT from zwane
(c) check physid's

Totally untested (compiletesting in progress).

-- wli

Index: mm1-2.6.7-rc3/arch/i386/kernel/mpparse.c
===================================================================
--- mm1-2.6.7-rc3.orig/arch/i386/kernel/mpparse.c	2004-06-09 07:11:51.335601000 -0700
+++ mm1-2.6.7-rc3/arch/i386/kernel/mpparse.c	2004-06-09 07:42:04.221000000 -0700
@@ -107,7 +107,7 @@
 #ifdef CONFIG_X86_NUMAQ
 static int MP_valid_apicid(int apicid, int version)
 {
-	return hweight_long(i & 0xf) == 1 && (i >> 4) != 0xf;
+	return hweight_long(apicid & 0xf) == 1 && (apicid >> 4) != 0xf;
 }
 #else
 static int MP_valid_apicid(int apicid, int version)
@@ -207,7 +207,7 @@
 	num_processors++;
 	ver = m->mpc_apicver;
 
-	if (MP_valid_apicid(m->mpc_apicid, ver))
+	if (MP_valid_apicid(apicid, ver))
 		MP_mark_version_physids(ver);
 	else {
 		printk(KERN_WARNING "Processor #%d INVALID. (Max ID: %d).\n",
@@ -871,7 +871,7 @@
 	MP_processor_info(&processor);
 }
 
-#if defined(CONFIG_X86_IO_APIC) && defined(CONFIG_ACPI_INTERPRETER)
+#if defined(CONFIG_X86_IO_APIC) && (defined(CONFIG_ACPI_INTERPRETER) || defined(CONFIG_ACPI_BOOT))
 
 #define MP_ISA_BUS		0
 #define MP_MAX_IOAPIC_PIN	127
@@ -1103,5 +1103,5 @@
 		    active_high_low == ACPI_ACTIVE_HIGH ? 0 : 1);
 }
 
-#endif /*CONFIG_X86_IO_APIC && CONFIG_ACPI_INTERPRETER*/
+#endif /*CONFIG_X86_IO_APIC && (CONFIG_ACPI_INTERPRETER || CONFIG_ACPI_BOOT)*/
 #endif /*CONFIG_ACPI_BOOT*/
Index: mm1-2.6.7-rc3/arch/i386/kernel/io_apic.c
===================================================================
--- mm1-2.6.7-rc3.orig/arch/i386/kernel/io_apic.c	2004-06-09 07:11:51.323603000 -0700
+++ mm1-2.6.7-rc3/arch/i386/kernel/io_apic.c	2004-06-09 07:30:06.936044000 -0700
@@ -722,7 +722,7 @@
 
 __setup("pirq=", ioapic_pirq_setup);
 
-static int get_physical_broadcast(void)
+int get_physical_broadcast(void)
 {
 	unsigned int lvr, version;
 	lvr = apic_read(APIC_LVR);
Index: mm1-2.6.7-rc3/include/asm-i386/mach-default/mach_apic.h
===================================================================
--- mm1-2.6.7-rc3.orig/include/asm-i386/mach-default/mach_apic.h	2004-06-09 07:11:51.513574000 -0700
+++ mm1-2.6.7-rc3/include/asm-i386/mach-default/mach_apic.h	2004-06-09 07:42:51.696783000 -0700
@@ -79,7 +79,10 @@
 
 static inline int cpu_present_to_apicid(int mps_cpu)
 {
-	return  mps_cpu;
+	if (mps_cpu < get_physical_broadcast())
+		return  mps_cpu;
+	else
+		return BAD_APICID;
 }
 
 static inline physid_mask_t apicid_to_cpu_present(int phys_apicid)
Index: mm1-2.6.7-rc3/include/asm-i386/mach-visws/mach_apic.h
===================================================================
--- mm1-2.6.7-rc3.orig/include/asm-i386/mach-visws/mach_apic.h	2004-06-09 07:11:51.635555000 -0700
+++ mm1-2.6.7-rc3/include/asm-i386/mach-visws/mach_apic.h	2004-06-09 07:29:30.131639000 -0700
@@ -60,7 +60,10 @@
 
 static inline int cpu_present_to_apicid(int mps_cpu)
 {
-	return mps_cpu;
+	if (mps_cpu < get_physical_broadcast())
+		return mps_cpu;
+	else
+		return BAD_APICID;
 }
 
 static inline physid_mask_t apicid_to_cpu_present(int apicid)
Index: mm1-2.6.7-rc3/include/asm-i386/apic.h
===================================================================
--- mm1-2.6.7-rc3.orig/include/asm-i386/apic.h	2004-06-07 12:15:12.000000000 -0700
+++ mm1-2.6.7-rc3/include/asm-i386/apic.h	2004-06-09 07:29:54.104995000 -0700
@@ -41,6 +41,8 @@
 	do { } while ( apic_read( APIC_ICR ) & APIC_ICR_BUSY );
 }
 
+int get_physical_broadcast(void);
+
 #ifdef CONFIG_X86_GOOD_APIC
 # define FORCE_READ_AROUND_WRITE 0
 # define apic_read_around(x)

  reply	other threads:[~2004-06-09 14:49 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-06-09  8:50 2.6.7-rc3-mm1 Andrew Morton
2004-06-09 11:25 ` 2.6.7-rc3-mm1 Eric BEGOT
2004-06-09 13:13   ` 2.6.7-rc3-mm1 Zwane Mwaikambo
2004-06-09 13:36     ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 13:42       ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 13:43       ` 2.6.7-rc3-mm1 Zwane Mwaikambo
2004-06-09 14:48         ` William Lee Irwin III [this message]
2004-06-09 14:58           ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 15:16             ` 2.6.7-rc3-mm1 William Lee Irwin III
     [not found]             ` <40C73198.4080700@yahoo.fr>
2004-06-09 15:50               ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 16:35 ` 2.6.7-rc3-mm1 Norberto Bensa
2004-06-09 17:05   ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 17:46     ` 2.6.7-rc3-mm1 Norberto Bensa
2004-06-10  3:38     ` 2.6.7-rc3-mm1 Norberto Bensa
2004-06-09 17:59 ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 19:50   ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 23:44 ` 2.6.7-rc3-mm1 (compile stats) John Cherry
2004-06-10  2:10 ` 2.6.7-rc3-mm1 Phil Brunner
2004-06-10  2:22   ` 2.6.7-rc3-mm1 Andrew Morton
2004-06-10  2:52   ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-10  8:59     ` 2.6.7-rc3-mm1 Phil Brunner
2004-06-10  5:16 ` 2.6.7-rc3-mm1 Clemens Schwaighofer
2004-06-10  5:31   ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-10  6:54     ` 2.6.7-rc3-mm1 Clemens Schwaighofer
2004-06-11  4:36 ` 2.6.7-rc3-mm1 Paul Jackson
2004-06-11  5:47   ` 2.6.7-rc3-mm1 Andrew Morton
2004-06-11  5:53     ` 2.6.7-rc3-mm1 Paul Jackson
  -- strict thread matches above, loose matches on Subject: below --
2004-06-09 15:33 2.6.7-rc3-mm1 Peter Maas
2004-06-09 15:52 ` 2.6.7-rc3-mm1 William Lee Irwin III
2004-06-09 20:03 2.6.7-rc3-mm1 Nguyen, Tom L
2004-06-10  0:51 2.6.7-rc3-mm1 Nguyen, Tom L
     [not found] <2576k-4hW-13@gated-at.bofh.it>
     [not found] ` <25LZK-88C-17@gated-at.bofh.it>
2004-06-11 10:48   ` 2.6.7-rc3-mm1 Andi Kleen
2004-06-11 11:32     ` 2.6.7-rc3-mm1 Paul Jackson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20040609144809.GK1444@holomorphy.com \
    --to=wli@holomorphy.com \
    --cc=akpm@osdl.org \
    --cc=eric_begot@yahoo.fr \
    --cc=linux-kernel@vger.kernel.org \
    --cc=zwane@fsmlabs.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox