Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266147AbUFIOtE (ORCPT ); Wed, 9 Jun 2004 10:49:04 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265787AbUFIOtE (ORCPT ); Wed, 9 Jun 2004 10:49:04 -0400 Received: from holomorphy.com ([207.189.100.168]:61316 "EHLO holomorphy.com") by vger.kernel.org with ESMTP id S266150AbUFIOsR (ORCPT ); Wed, 9 Jun 2004 10:48:17 -0400 Date: Wed, 9 Jun 2004 07:48:09 -0700 From: William Lee Irwin III To: Zwane Mwaikambo Cc: Eric BEGOT , Andrew Morton , Linux Kernel Subject: Re: 2.6.7-rc3-mm1 Message-ID: <20040609144809.GK1444@holomorphy.com> Mail-Followup-To: William Lee Irwin III , Zwane Mwaikambo , Eric BEGOT , Andrew Morton , Linux Kernel References: <20040609015001.31d249ca.akpm@osdl.org> <40C6F3C3.9040401@yahoo.fr> <20040609133653.GH1444@holomorphy.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.5.1+cvs20040105i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4898 Lines: 130 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) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/