coreboot

coreboot@coreboot.org

11 participants
39714 discussions

Overlaping IO resource for AMD K8
by Li-Ta Lo 11 Jan '05

11 Jan '05

Eric, Why I got the overlaping IO resources in the 18:0 ? PCI: 00:18.0 1ba <- [0x00f4000000 - 0x00f5ffffff] prefmem <node 0 link2> PCI: 00:18.0 1c2 <- [0x0000001000 - 0x0000001fff] io <node 0 link 2> PCI: 00:18.0 1d8 <- [0x0000002000 - 0x0000001fff] io <node 0 link 0> PCI: 00:18.0 1b0 <- [0x00e0000000 - 0x00efffffff] prefmem <node 0 link0> PCI: 00:18.0 1a8 <- [0x00f6300000 - 0x00f63fffff] mem <node 0 link 0> PCI: 00:18.0 1a2 <- [0x00f6000000 - 0x00f62fffff] mem <node 0 link 2> Shouldn't the 1d8 be 0x2000 to 0x2ffff? Ollie

2 1

RE: Overlaping IO resource for AMD K8
by YhLu 11 Jan '05

11 Jan '05

Complie error You missed sth? Tue Jan 11 10:17:11 PST 2005 FAILED! Log excerpt: /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c: In function `setup_system_bios': /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c:219: warning: implicit declaration of function `memset' /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c: At top level: /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c:232: error: conflicting types for 'reset_int_vect' /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c:179: error: previous implicit declaration of 'reset_int_vect' was here /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c: In function `run_bios': /home/yhlu/xx/xx/freebios2/src/devices/emulator/biosemu.c:289: warning: implicit declaration of function `X86EMU_setMemBase' make[1]: *** [biosemu.o] Error 1 -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Tuesday, January 11, 2005 11:13 AM To: YhLu Cc: LinuxBIOS; Eric Biederman Subject: RE: Overlaping IO resource for AMD K8 On Tue, 2005-01-11 at 12:02, Li-Ta Lo wrote: > On Tue, 2005-01-11 at 11:56, YhLu wrote: > > Ollie, > > > > I recall sth about mtrr setting, I wonder if I enable the NC could affect > > your code. Please check cpu/x86/mtrr/mtrr.c about enable NC below 4G. > > > > I think there is some problem with the resource allocation. If the vga > card has PCI IO resource then there will be no overlaping. If it does > not have PCI IO resource then there will be overlaping. > There is really a but in the resource allocation. Please take a look at the attached log files. The ATI card has PCI IO resource so there is no bogus resource in the northbridge. As the Nvidia card it does not have PCI IO resource so we got the 20000 to 1ffff thing. Ollie

2 1

RE: Overlaping IO resource for AMD K8
by YhLu 11 Jan '05

11 Jan '05

Ollie, I recall sth about mtrr setting, I wonder if I enable the NC could affect your code. Please check cpu/x86/mtrr/mtrr.c about enable NC below 4G. YH -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Tuesday, January 11, 2005 10:21 AM To: LinuxBIOS; Eric Biederman Subject: Overlaping IO resource for AMD K8 Eric, Why I got the overlaping IO resources in the 18:0 ? PCI: 00:18.0 1ba <- [0x00f4000000 - 0x00f5ffffff] prefmem <node 0 link2> PCI: 00:18.0 1c2 <- [0x0000001000 - 0x0000001fff] io <node 0 link 2> PCI: 00:18.0 1d8 <- [0x0000002000 - 0x0000001fff] io <node 0 link 0> PCI: 00:18.0 1b0 <- [0x00e0000000 - 0x00efffffff] prefmem <node 0 link0> PCI: 00:18.0 1a8 <- [0x00f6300000 - 0x00f63fffff] mem <node 0 link 0> PCI: 00:18.0 1a2 <- [0x00f6000000 - 0x00f62fffff] mem <node 0 link 2> Shouldn't the 1d8 be 0x2000 to 0x2ffff? Ollie _______________________________________________ Linuxbios mailing list Linuxbios(a)clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios

2 2

RE: Enable apic ext id to use 256 apic id for AMD64
by YhLu 11 Jan '05

11 Jan '05

I don't think it is caused by the change about nodeid. In s2885, it only change lapicid() to lapicid() & 0xf, and will not affect non apic lifting problem. What't the card that you are using on S2885?, and If I get time, I would try that on s2885. YH -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Tuesday, January 11, 2005 9:12 AM To: YhLu Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: RE: Enable apic ext id to use 256 apic id for AMD64 On Mon, 2005-01-10 at 15:27, YhLu wrote: > For dual core and lifting apic id, you can not just use apic id as node id. > > And you have to use > Lapicid() & 0xf > > Or cupid to get it. > > Regards > > YH It breaks Expansion ROM loading from PCI slot to the main memory. The old code does not have this execprtion. pci_rom_load, copying VGA ROM Image from f8100000 to c0000, 8000 bytes Unexpected Exception: 2 @ 10:00009526 - Halting Code: 0 eflags: 00000083 eax: 000000ff ebx: 00008000 ecx: 000c0000 edx: 00000051 edi: 0001acd0 esi: f8100000 ebp: 00037f74 esp: 00037f6c Ollie > > -----Original Message----- > From: Li-Ta Lo [mailto:ollie@lanl.gov] > Sent: Monday, January 10, 2005 2:14 PM > To: YhLu > Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; > discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt > Domsch > Subject: RE: Enable apic ext id to use 256 apic id for AMD64 > > On Mon, 2005-01-10 at 15:18, YhLu wrote: > > Please refer s4882 to change the MB that you support. > > > > nodeid = lapicid() & 0xf; > > > > orm with initial apic id > > nodeid = (cpuid_ebx(1) >> 24) & 0xf; > > > > What is this thing all about ? > > Ollie >

2 1

RE: Enable apic ext id to use 256 apic id for AMD64
by YhLu 11 Jan '05

11 Jan '05

For dual core and lifting apic id, you can not just use apic id as node id. And you have to use Lapicid() & 0xf Or cupid to get it. Regards YH -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Monday, January 10, 2005 2:14 PM To: YhLu Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: RE: Enable apic ext id to use 256 apic id for AMD64 On Mon, 2005-01-10 at 15:18, YhLu wrote: > Please refer s4882 to change the MB that you support. > > nodeid = lapicid() & 0xf; > > orm with initial apic id > nodeid = (cpuid_ebx(1) >> 24) & 0xf; > What is this thing all about ? Ollie

2 1

PCI Expansion ROM POST committed
by Li-Ta Lo 11 Jan '05

11 Jan '05

Hello, I just committed the VGABIOS support. It works for some ATI and Nvidia card. Currently it will try to POST any PCI card it can find with PCI expansion ROM. I will add the #ifdef PCI_EXPANSION_ROM or if (class == vga_class) later. Ollie

1 1

RE: Enable apic ext id to use 256 apic id for AMD64
by YhLu 10 Jan '05

10 Jan '05

I changed that for all tyan, amd, and arima MB. YH -----Original Message----- From: YhLu Sent: Monday, January 10, 2005 2:28 PM To: Li-Ta Lo Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: RE: Enable apic ext id to use 256 apic id for AMD64 OK. -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Monday, January 10, 2005 2:16 PM To: YhLu Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: RE: Enable apic ext id to use 256 apic id for AMD64 On Mon, 2005-01-10 at 15:18, YhLu wrote: > Please refer s4882 to change the MB that you support. > > nodeid = lapicid() & 0xf; > > orm with initial apic id > nodeid = (cpuid_ebx(1) >> 24) & 0xf; > > Can you do that to the s2885 for me? I am busy working on the VGABIOS support and have to brain for that. Ollie _______________________________________________ Linuxbios mailing list Linuxbios(a)clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios

2 1

RE: Enable apic ext id to use 256 apic id for AMD64
by YhLu 10 Jan '05

10 Jan '05

OK. -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Monday, January 10, 2005 2:16 PM To: YhLu Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: RE: Enable apic ext id to use 256 apic id for AMD64 On Mon, 2005-01-10 at 15:18, YhLu wrote: > Please refer s4882 to change the MB that you support. > > nodeid = lapicid() & 0xf; > > orm with initial apic id > nodeid = (cpuid_ebx(1) >> 24) & 0xf; > > Can you do that to the s2885 for me? I am busy working on the VGABIOS support and have to brain for that. Ollie

1 0

RE: Enable apic ext id to use 256 apic id for AMD64
by YhLu 10 Jan '05

10 Jan '05

Please refer s4882 to change the MB that you support. nodeid = lapicid() & 0xf; orm with initial apic id nodeid = (cpuid_ebx(1) >> 24) & 0xf; YH -----Original Message----- From: Li-Ta Lo [mailto:ollie@lanl.gov] Sent: Monday, January 10, 2005 1:24 PM To: YhLu Cc: Stefan Reinauer; Ronald G. Minnich; LinuxBIOS; Andi Kleen; discuss(a)x86-64.org; jamesclv(a)us.ibm.com; suresh.b.siddha(a)intel.com; Matt Domsch Subject: Re: Enable apic ext id to use 256 apic id for AMD64 On Fri, 2005-01-07 at 14:35, YhLu wrote: > Stefan, > > Please check out the code from the CVS code. > > After LinuxBIOS keeps the bsp to still use apicid 0. Kernel (2.6.10) x86-64 > can work well. > > Andi suggest that. But I wonder why the normal BIOS (kernel with acpi > enabled) can lift that to 0x10. --- normal BIOS (kernel with acpi disabled) > will fail too. > YhLU, You change the cpu_init_detected(void) to cpu_init_detected(nodeid), what should I put in the nodeid? You break EVERY K8 mainboard we have. Ollie

2 2

Enable apic ext id to use 256 apic id for AMD64
by YhLu 10 Jan '05

10 Jan '05

Stefan, Please check out the code from the CVS code. After LinuxBIOS keeps the bsp to still use apicid 0. Kernel (2.6.10) x86-64 can work well. Andi suggest that. But I wonder why the normal BIOS (kernel with acpi enabled) can lift that to 0x10. --- normal BIOS (kernel with acpi disabled) will fail too. I also CC the kernel guys relating to this. YH -----Original Message----- From: YhLu Sent: Wednesday, January 05, 2005 12:43 PM To: Ronald G. Minnich; Stefan Reinauer Cc: linuxbios(a)clustermatic.org Subject: RE: mptable.c Enable apic ext id committed. Changes Auto.c nodeid = lapicid() & 0xf; #if ENABLE_APIC_EXT_ID == 1 enable_apic_ext_id(nodeid); lapic_write(LAPIC_ID, ( lapic_read(LAPIC_ID) | (APIC_ID_OFFSET<<24) ) ); // CPU apicid is from 0x10 #endif if (cpu_init_detected(nodeid)) { asm volatile ("jmp __cpu_reset"); } distinguish_cpu_resets(nodeid); also reset_test.c change cpu_init_detected and distinguish_cpu_resets to take node id. YH

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

coreboot