Skip to contentSkip to conversation

Kernel 2.66.33-rC1 boot problem bisected to commit 99935a7a59eaca0292c1a5880e10bae03f4a5e3d

DISCUSSION Vincent ETIENNE 2 participants 3 messages Dec 19, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

I am experiencing boot problem with kernel 2.6.33-rc1. Bisection points to

commit 99935a7a59eaca0292c1a5880e10bae03f4a5e3d

Author: Yinghai Lu <[email protected]>
Date:   Sun Oct 4 21:54:24 2009 -0700

x86/PCI: read root resources from IOH on Intel

For intel systems with multi IOH, we should read peer root resources directly from PCI config space, and don't trust _CRS.

V2.6.32 is OK

V2.6.33 is bad. A brief summary of the problem is that very quickly kernel ( less than 1.5 seconds) try to mount the root device without doing any probing on disk leading to a kernel panic ( could not mount root devices). Alsa devices and usb devices is also not detected / not probed. Maybe other devices is also not detected but that just from the message seen on the monitor before the panic so it's not very accurate (no log as it doesn't boot and only one computer so no serial console). Sounds like the kernel didn't see any devices (disk, usb, sounds).

I have a intel system so the commit log seems possibly relevant for the my computer (Not sure of what is IOH, i have take it for I/O Hub in the lspci out output and i have 3 of these according to lspci).

Have try to revert the commit on a 2.6.33_rc1 base but this lead to some compilation problem (symbol not defined and so on). So I'm not completely sure of my bisection (also note that it's the first time i use git so an error is very possible).

Attached a dmesg log with kernel 2.6.32-rc8 and lspci -vvv output.

If you need further information/testing, i will my best to provide these.

Regards
Vincent ETIENNE

#1 lore raw
Quoted 32 lines · Vincent ETIENNE

I am experiencing boot problem with kernel 2.6.33-rc1. Bisection points to

commit 99935a7a59eaca0292c1a5880e10bae03f4a5e3d

Author: Yinghai Lu <[email protected]>
Date:   Sun Oct 4 21:54:24 2009 -0700

x86/PCI: read root resources from IOH on Intel

For intel systems with multi IOH, we should read peer root resources directly from PCI config space, and don't trust _CRS.

V2.6.32 is OK

V2.6.33 is bad. A brief summary of the problem is that very quickly kernel ( less than 1.5 seconds) try to mount the root device without doing any probing on disk leading to a kernel panic ( could not mount root devices). Alsa devices and usb devices is also not detected / not probed. Maybe other devices is also not detected but that just from the message seen on the monitor before the panic so it's not very accurate (no log as it doesn't boot and only one computer so no serial console). Sounds like the kernel didn't see any devices (disk, usb, sounds).

I have a intel system so the commit log seems possibly relevant for the my computer (Not sure of what is IOH, i have take it for I/O Hub in the lspci out output and i have 3 of these according to lspci).

Quoted 8 lines

Have try to revert the commit on a 2.6.33_rc1 base but this lead to some compilation problem (symbol not defined and so on). So I'm not completely sure of my bisection (also note that it's the first time i use git so an error is very possible).

Attached a dmesg log with kernel 2.6.32-rc8 and lspci -vvv output.

If you need further information/testing, i will my best to provide these.

please check

Subject: [PATCH 1/7] x86/pci: intel ioh bus num reg accessing fix

it is above 0x100, so if mmconf is not enable, need to skip it

---
 arch/x86/pci/intel_bus.c |    4 ++++
 1 file changed, 4 insertions(+)
arch/x86/pci/intel_bus.c +4 -0
Index: linux-2.6/arch/x86/pci/intel_bus.c
===================================================================
@@ -49,6 +49,10 @@ static void __devinit pci_root_bus_res(s
 	u64 mmioh_base, mmioh_end;
 	int bus_base, bus_end;
 
+	/* some sys doesn't get mmconf enabled */
+	if (dev->cfg_size < 0x200)
+		return;
+
 	if (pci_root_num >= PCI_ROOT_NR) {
 		printk(KERN_DEBUG "intel_bus.c: PCI_ROOT_NR is too small\n");
 		return;
#2 lore raw
Quoted 24 lines · Yinghai Lu
Quoted 3 lines · Vincent ETIENNE

I am experiencing boot problem with kernel 2.6.33-rc1. Bisection points to

please check

Subject: [PATCH 1/7] x86/pci: intel ioh bus num reg accessing fix

it is above 0x100, so if mmconf is not enable, need to skip it

---
 arch/x86/pci/intel_bus.c |    4 ++++
 1 file changed, 4 insertions(+)

Index: linux-2.6/arch/x86/pci/intel_bus.c
===================================================================

arch/x86/pci/intel_bus.c static void __devinit pci_root_bus_res(s · line 58

Quoted 10 lines
arch/x86/pci/intel_bus.c +4 -0
@@ -49,6 +49,10 @@ static void __devinit pci_root_bus_res(s
 	u64 mmioh_base, mmioh_end;
 	int bus_base, bus_end;
 
+	/* some sys doesn't get mmconf enabled */
+	if (dev->cfg_size < 0x200)
+		return;
+
 	if (pci_root_num >= PCI_ROOT_NR) {
 		printk(KERN_DEBUG "intel_bus.c: PCI_ROOT_NR is too small\n");
 		return;

It works : v2.6;33_rc1 + your patch booot sucessfully on my machine Thanks a lot

Vincent