i2c: lost sensors with 2.6.10(-mm1)
Hi...
I have lost my sensors info with 2.6.10, in particular -mm1. They work fine with 2.6.9-mm1 (current state of the box, booted on 2.6.9 or 10, no other difference).
Some info follows.
With 2.6.9-mm1, I get this:
Module Size Used by w83627hf 25128 0 i2c_sensor 3456 1 w83627hf i2c_isa 2304 0 i2c_i801 7820 0 i2c_dev 8192 0 i2c_core 19712 5 w83627hf,i2c_sensor,i2c_isa,i2c_i801,i2c_dev
sensors:
w83627thf-isa-0290
Adapter: ISA adapter
VCore: +1.49 V (min = +1.93 V, max = +1.93 V)
...
With 2.6.10-mm1:
Module Size Used by w83627hf 24224 0 i2c_sensor 3968 1 w83627hf i2c_isa 2304 0 i2c_i801 9996 0 i2c_dev 8960 0 i2c_core 20992 5 w83627hf,i2c_sensor,i2c_isa,i2c_i801,i2c_dev
sensors:
No sensors found!
I have noticed different contents in /sys:
under 2.6.9:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/1-0290:/sys/devices/platform/i2c-1/1-0290/power:/sys/devices/platform/i2c-1/power:
under 2.6.10:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/power:
So some /sys nodes are missing !!!
(the isa bus)
Debug output from 2.6.10-mm1:
Jan 7 01:33:11 werewolf kernel: i2c /dev entries driver Jan 7 01:33:11 werewolf kernel: i2c-core: driver dev_driver registered. Jan 7 01:33:11 werewolf kernel: i801_smbus 0000:00:1f.3: I801 using PCI Interrupt for SMBus. Jan 7 01:33:11 werewolf kernel: i801_smbus 0000:00:1f.3: SMBREV = 0x2 Jan 7 01:33:11 werewolf kernel: i801_smbus 0000:00:1f.3: I801_smba = 0x500 Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-0: Registered as minor 0 Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-0: registered as adapter #0 Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: Registered as minor 1 Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: registered as adapter #1 Jan 7 01:33:11 werewolf kernel: i2c-core: driver w83627hf registered. Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: found normal isa entry for adapter 9191, addr 0290 Jan 7 01:33:12 werewolf sensord: sensord started Jan 7 01:33:12 werewolf lm_sensors: sensord startup succeeded
Some ideas ?
TIA
--
J.A. Magallon <jamagallon()able!es> \ Software is like sex:
werewolf!able!es \ It's better when it's free
Mandrakelinux release 10.2 (Cooker) for i586
Linux 2.6.10-jam1 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-2mdk)) #2
Quoted 11 lines · J.A. Magallon
I have lost my sensors info with 2.6.10, in particular -mm1. They work fine with 2.6.9-mm1 (current state of the box, booted on 2.6.9 or 10, no other difference).
(...)
I have noticed different contents in /sys:
under 2.6.9:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/1-0290:/sys/devices/platform/i2c-1/1-0290/power:/sys/devices/platform/i2c-1/power:
Quoted 6 lines
under 2.6.10:
/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/power:So some /sys nodes are missing !!!
(the isa bus)
This basically means that the i2c client was not registered.
Quoted 4 lines
Debug output from 2.6.10-mm1:
(...)Jan 7 01:33:11 werewolf kernel: i2c-core: driver w83627hf registered. Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: found normal isa entry for adapter 9191, addr 0290
However, this suggests that the driver loaded properly and the base address was correctly read from Super-I/O space. This would mean that the problem happened later, in w83627hf_detect(). The most likely reason for this would be if the region request failed (unfortunately we have no message, not even debug, if this happens).
Quoted 1 line
Some ideas ?
Three things to try, in order:
1* Compare /proc/ioports in 2.6.9-mm1 and 2.6.10-mm1. I suspect that the 0x290-0x297 range is held by some device in 2.6.10-mm1.
2* Try reverting this patch in 2.6.10-mm1:
http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.10/2.6.10-mm1/broken-out/bk-i2c.patch
It does indeed include a change in the way the I/O region is requested. It should not make any difference, but maybe we are missing something and it actually does.
3* Try a vanilla 2.6.10 kernel and report how it is going.
Thanks,
Signature · 1 line
Jean Delvare
Quoted 33 lines · Jean Delvare
Quoted 11 lines · J.A. Magallon
I have lost my sensors info with 2.6.10, in particular -mm1. They work fine with 2.6.9-mm1 (current state of the box, booted on 2.6.9 or 10, no other difference).
(...)
I have noticed different contents in /sys:
under 2.6.9:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/1-0290:/sys/devices/platform/i2c-1/1-0290/power:/sys/devices/platform/i2c-1/power:Quoted 6 lines
under 2.6.10:
/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/power:So some /sys nodes are missing !!!
(the isa bus)This basically means that the i2c client was not registered.
Quoted 4 lines
Debug output from 2.6.10-mm1:
(...)Jan 7 01:33:11 werewolf kernel: i2c-core: driver w83627hf registered. Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: found normal isa entry for adapter 9191, addr 0290However, this suggests that the driver loaded properly and the base address was correctly read from Super-I/O space. This would mean that the problem happened later, in
w83627hf_detect(). The most likely reason for this would be if the region request failed (unfortunately we have no message, not even debug, if this happens).
Quoted 7 lines
Quoted 1 line
Some ideas ?
Three things to try, in order:
1* Compare
/proc/ioportsin 2.6.9-mm1 and 2.6.10-mm1. I suspect that the 0x290-0x297 range is held by some device in 2.6.10-mm1.
Good start. 2.6.10-mm1 shows this:
@@ -45,6 +44,7 @@
00f0-00ff : fpu
0170-0177 : ide1
01f0-01f7 : ide0
+0295-0296 : pnp 00:0d
02f8-02ff : serial
0376-0376 : ide1
03c0-03df : vga+
Differences in dmesg show this (apart from others):
@@ -120,6 +120,7 @@
Normal zone: 257760 pages, LIFO batch:16
HighMem zone: 0 pages, LIFO batch:1
DMI 2.3 present.
+__iounmap: bad address b00f0000
ACPI: RSDP (v000 IntelR ) @ 0x000f71f0
ACPI: RSDT (v001 IntelR AWRDACPI 0x42302e31 AWRD 0x00000000) @ 0x3fee3000
ACPI: FADT (v001 IntelR AWRDACPI 0x42302e31 AWRD 0x00000000) @ 0x3fee3040
@@ -149,25 +150,27 @@
Enabling APIC mode: Flat. Using 1 I/O APICs
Using ACPI (MADT) for SMP configuration information
Built 1 zonelists
+mapped APIC to ffffd000 (fee00000)
+mapped IOAPIC to ffffc000 (fec00000)
Initializing CPU#0
@@ -236,7 +239,9 @@
PCI: PCI BIOS revision 2.10 entry at 0xfb7c0, last bus=3
PCI: Using configuration type 1
mtrr: v2.0 (20020519)
-ACPI: Subsystem revision 20040816
+ACPI: Subsystem revision 20041210
+ ACPI-1138: *** Error: Method execution failed [\STRC] (Node efedc380), AE_AML_BUFFER_LIMIT
+ ACPI-1138: *** Error: Method execution failed [\_SB_.PCI0._INI] (Node efedd620), AE_AML_BUFFER_LIMIT
ACPI: Interpreter enabled
ACPI: Using IOAPIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (00:00)
@@ -254,6 +259,8 @@
ACPI: PCI Interrupt Link [LNK0] (IRQs 3 4 5 7 9 10 *11 12 14 15)
ACPI: PCI Interrupt Link [LNK1] (IRQs 3 4 5 7 *9 10 11 12 14 15) Linux Plug and Play Support v0.97 (c) Adam Belay
+pnp: PnP ACPI init
+pnp: PnP ACPI: found 16 devices
SCSI subsystem initialized
PCI: Using ACPI for IRQ routing
@@ -263,6 +270,8 @@
** behavior. If this argument makes the device work again, ** please email the output of "lspci" to [email protected] ** so I can fix the driver.
+pnp: 00:0d: ioport range 0x400-0x4bf could not be reserved +pnp: 00:0d: ioport range 0x295-0x296 has been reserved
IA-32 Microcode Update Driver: v1.14 <[email protected]>
isapnp: Scanning for PnP cards...
isapnp: No Plug & Play device found
So I guess it is "Plug and Play ACPI support", I'm just going to disable it and try. If it is the case, there is an IO port conflict...
TIA
--
J.A. Magallon <jamagallon()able!es> \ Software is like sex:
werewolf!able!es \ It's better when it's free
Mandrakelinux release 10.2 (Cooker) for i586
Linux 2.6.10-jam1 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-2mdk)) #2
Fwd: Re: i2c: lost sensors with 2.6.10(-mm1) [[email protected]]
On 2005.01.08, J.A. Magallon wrote:
Quoted 33 lines · Jean Delvare
Quoted 11 lines · J.A. Magallon
I have lost my sensors info with 2.6.10, in particular -mm1. They work fine with 2.6.9-mm1 (current state of the box, booted on 2.6.9 or 10, no other difference).
(...)
I have noticed different contents in /sys:
under 2.6.9:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/1-0290:/sys/devices/platform/i2c-1/1-0290/power:/sys/devices/platform/i2c-1/power:Quoted 6 lines
under 2.6.10:
/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/power:So some /sys nodes are missing !!!
(the isa bus)This basically means that the i2c client was not registered.
Quoted 4 lines
Debug output from 2.6.10-mm1:
(...)Jan 7 01:33:11 werewolf kernel: i2c-core: driver w83627hf registered. Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: found normal isa entry for adapter 9191, addr 0290However, this suggests that the driver loaded properly and the base address was correctly read from Super-I/O space. This would mean that the problem happened later, in
w83627hf_detect(). The most likely reason for this would be if the region request failed (unfortunately we have no message, not even debug, if this happens).
Quoted 7 lines
Quoted 1 line
Some ideas ?
Three things to try, in order:
1* Compare
/proc/ioportsin 2.6.9-mm1 and 2.6.10-mm1. I suspect that the 0x290-0x297 range is held by some device in 2.6.10-mm1.
Good start. 2.6.10-mm1 shows this:
@@ -45,6 +44,7 @@
00f0-00ff : fpu
0170-0177 : ide1
01f0-01f7 : ide0
+0295-0296 : pnp 00:0d
02f8-02ff : serial
0376-0376 : ide1
03c0-03df : vga+
Differences in dmesg show this (apart from others):
@@ -120,6 +120,7 @@
Normal zone: 257760 pages, LIFO batch:16
HighMem zone: 0 pages, LIFO batch:1
DMI 2.3 present.
+__iounmap: bad address b00f0000
ACPI: RSDP (v000 IntelR ) @ 0x000f71f0
ACPI: RSDT (v001 IntelR AWRDACPI 0x42302e31 AWRD 0x00000000) @ 0x3fee3000
ACPI: FADT (v001 IntelR AWRDACPI 0x42302e31 AWRD 0x00000000) @ 0x3fee3040
@@ -149,25 +150,27 @@
Enabling APIC mode: Flat. Using 1 I/O APICs
Using ACPI (MADT) for SMP configuration information
Built 1 zonelists
+mapped APIC to ffffd000 (fee00000)
+mapped IOAPIC to ffffc000 (fec00000)
Initializing CPU#0
@@ -236,7 +239,9 @@
PCI: PCI BIOS revision 2.10 entry at 0xfb7c0, last bus=3
PCI: Using configuration type 1
mtrr: v2.0 (20020519)
-ACPI: Subsystem revision 20040816
+ACPI: Subsystem revision 20041210
+ ACPI-1138: *** Error: Method execution failed [\STRC] (Node efedc380), AE_AML_BUFFER_LIMIT
+ ACPI-1138: *** Error: Method execution failed [\_SB_.PCI0._INI] (Node efedd620), AE_AML_BUFFER_LIMIT
ACPI: Interpreter enabled
ACPI: Using IOAPIC for interrupt routing
ACPI: PCI Root Bridge [PCI0] (00:00)
@@ -254,6 +259,8 @@
ACPI: PCI Interrupt Link [LNK0] (IRQs 3 4 5 7 9 10 *11 12 14 15)
ACPI: PCI Interrupt Link [LNK1] (IRQs 3 4 5 7 *9 10 11 12 14 15) Linux Plug and Play Support v0.97 (c) Adam Belay
+pnp: PnP ACPI init
+pnp: PnP ACPI: found 16 devices
SCSI subsystem initialized
PCI: Using ACPI for IRQ routing
@@ -263,6 +270,8 @@
** behavior. If this argument makes the device work again, ** please email the output of "lspci" to [email protected] ** so I can fix the driver.
+pnp: 00:0d: ioport range 0x400-0x4bf could not be reserved +pnp: 00:0d: ioport range 0x295-0x296 has been reserved
IA-32 Microcode Update Driver: v1.14 <[email protected]>
isapnp: Scanning for PnP cards...
isapnp: No Plug & Play device found
So I guess it is "Plug and Play ACPI support", I'm just going to disable it and try. If it is the case, there is an IO port conflict...
TIA
--
J.A. Magallon <jamagallon()able!es> \ Software is like sex:
werewolf!able!es \ It's better when it's free
Mandrakelinux release 10.2 (Cooker) for i586
Linux 2.6.10-jam1 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-2mdk)) #2
--
J.A. Magallon <jamagallon()able!es> \ Software is like sex:
werewolf!able!es \ It's better when it's free
Mandrakelinux release 10.2 (Cooker) for i586
Linux 2.6.10-jam1 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-2mdk)) #3
Quoted 33 lines · Jean Delvare
Quoted 11 lines · J.A. Magallon
I have lost my sensors info with 2.6.10, in particular -mm1. They work fine with 2.6.9-mm1 (current state of the box, booted on 2.6.9 or 10, no other difference).
(...)
I have noticed different contents in /sys:
under 2.6.9:/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/1-0290:/sys/devices/platform/i2c-1/1-0290/power:/sys/devices/platform/i2c-1/power:Quoted 6 lines
under 2.6.10:
/sys/devices/platform/i2c-1:/sys/devices/platform/i2c-1/power:So some /sys nodes are missing !!!
(the isa bus)This basically means that the i2c client was not registered.
Quoted 4 lines
Debug output from 2.6.10-mm1:
(...)Jan 7 01:33:11 werewolf kernel: i2c-core: driver w83627hf registered. Jan 7 01:33:11 werewolf kernel: i2c_adapter i2c-1: found normal isa entry for adapter 9191, addr 0290However, this suggests that the driver loaded properly and the base address was correctly read from Super-I/O space. This would mean that the problem happened later, in
w83627hf_detect(). The most likely reason for this would be if the region request failed (unfortunately we have no message, not even debug, if this happens).
Quoted 7 lines
Quoted 1 line
Some ideas ?
Three things to try, in order:
1* Compare
/proc/ioportsin 2.6.9-mm1 and 2.6.10-mm1. I suspect that the 0x290-0x297 range is held by some device in 2.6.10-mm1.
Disabling "Plug and Play ACPI support" freed the ioport range and sensors worked again.
Thanks for the info.
Is this a problem/conflict ? Or I can't use PnP-ACPI (if I ever guess what is it useful for) and sensors at the same time ?
TIA
--
J.A. Magallon <jamagallon()able!es> \ Software is like sex:
werewolf!able!es \ It's better when it's free
Mandrakelinux release 10.2 (Cooker) for i586
Linux 2.6.10-jam2 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-2mdk)) #1