Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Roman Mamedov]

[-- Attachment #1.1: Type: text/plain, Size: 899 bytes --]

On Mon, 1 Mar 2010 03:25:08 +0500
Roman Mamedov <roman@rm.pp.ru> wrote:

> I have a Pentium III with ASUS CUSL2 motherboard. When it is booted into
> the kernel 2.6.30, the sensors are detected and work fine. But when the
> kernel is 2.6.33, no sensors are detected. I did not try any kernel versions
> in between for now. Any idea what could be the cause? Attaching
> sensors-detect outputs from under both kernels.

Hm, looking at the driver's history, I believe it is caused by the second
change in that commit:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blobdiff;f=drivers/hwmon/w83781d.c;h=d38b9ede3abd1851b72d47ba0088678d01ecffb4;hp=e0fa7520400d8c565f0b7c77034a6c0e35cf4a1a;hb=6722feada7f6213ba06d510ef853e57ddcd59dae;hpb=7fce084a0b3e2bb8caef919f8f36065953655bb5
(shorter URL in case the above one breaks: http://ur1.ca/nwxs )

-- 
With respect,
Roman

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 153 bytes --]

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Jean Delvare]

Hi Roman,

On Mon, 1 Mar 2010 03:32:33 +0500, Roman Mamedov wrote:
> On Mon, 1 Mar 2010 03:25:08 +0500
> Roman Mamedov <roman@rm.pp.ru> wrote:
> 
> > I have a Pentium III with ASUS CUSL2 motherboard. When it is booted into
> > the kernel 2.6.30, the sensors are detected and work fine. But when the
> > kernel is 2.6.33, no sensors are detected. I did not try any kernel versions
> > in between for now. Any idea what could be the cause? Attaching
> > sensors-detect outputs from under both kernels.
> 
> Hm, looking at the driver's history, I believe it is caused by the second
> change in that commit:
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blobdiff;f=drivers/hwmon/w83781d.c;hÓ8b9ede3abd1851b72d47ba0088678d01ecffb4;hpàfa7520400d8c565f0b7c77034a6c0e35cf4a1a;hbg22feada7f6213ba06d510ef853e57ddcd59dae;hpb\x7fce084a0b3e2bb8caef919f8f36065953655bb5
> (shorter URL in case the above one breaks: http://ur1.ca/nwxs )

I don't think so. The commit in question basically dropped support for
I2C addresses 0x20-0x27, but according to sensors-detect, your chip
lives at I2C address 0x2d:

> Driver `w83781d':
>   * Bus `SMBus I801 adapter at e800'
>     Busdriver `i2c_i801', I2C address 0x2d
>     Chip `as99127f' (confidence: 6)

The output of sensors-detect on 2.6.33 shows that the SMBus is not even
scanned, even though the PCI device is present. So the problem is not
with the w83781d driver (hardware monitoring chip) but with the
i2c-i801 driver (SMBus controller).

That being said, there are no relevant changes to the i2c-i801 driver
between kernels 2.6.30 and 2.6.33. So most likely the problem is
elsewhere. Could be the common ACPI resource conflict issue, even
though I am surprised that such an old mainboard would be affected.

Please provide:
* The contents of /proc/ioports
* The output of "isadump -f 0x290 16"
* The kernel log messages when cycling the i2c-i801 driver ("rmmod
  i2c-i801 && modprobe i2c-i801" as root)
* The output of "dmidecode -t baseboard -t system"

-- 
Jean Delvare
http://khali.linux-fr.org/wishlist.html

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Roman Mamedov]

[-- Attachment #1.1.1: Type: text/plain, Size: 815 bytes --]

Hello.

Thanks for the reply.

On Mon, 1 Mar 2010 11:35:48 +0100
Jean Delvare <khali@linux-fr.org> wrote:

> * The contents of /proc/ioports
> * The output of "isadump -f 0x290 16"
> * The output of "dmidecode -t baseboard -t system"

Please see the attached text files.

> * The kernel log messages when cycling the i2c-i801 driver ("rmmod
>   i2c-i801 && modprobe i2c-i801" as root)

There are no messages in dmesg or syslog when removing and reloading this
driver in 2.6.30, but in 2.6.33 there are errors:

[  173.468108] ACPI: I/O resource 0000:00:1f.3 [0xe800-0xe80f] conflicts with
ACPI region SM00 [0xe800-0xe806]
[  173.468125] ACPI: If an ACPI driver is available for this device, you
should use it instead of the native driver

How can I solve this?

-- 
With respect,
Roman

[-- Attachment #1.1.2: 2.6.30-dmidecode-baseboard-system.txt --]
[-- Type: text/plain, Size: 536 bytes --]

# dmidecode 2.9
SMBIOS 2.3 present.

Handle 0x0001, DMI type 1, 25 bytes
System Information
	Manufacturer: System Manufacturer
	Product Name: System Name
	Version: System Version
	Serial Number: SYS-1234567890
	UUID: Not Settable
	Wake-up Type: Power Switch

Handle 0x0002, DMI type 2, 8 bytes
Base Board Information
	Manufacturer: ASUSTeK Computer INC.
	Product Name: CUSL2
	Version: REV 1.xx
	Serial Number: xxxxxxxxxxx

Handle 0x002E, DMI type 32, 11 bytes
System Boot Information
	Status: No errors detected


[-- Attachment #1.1.3: 2.6.30-ioports.txt --]
[-- Type: text/plain, Size: 998 bytes --]

0000-001f : dma1
0020-0021 : pic1
0040-0043 : timer0
0050-0053 : timer1
0060-0060 : keyboard
0064-0064 : keyboard
0070-0073 : rtc0
0080-008f : dma page reg
00a0-00a1 : pic2
00c0-00df : dma2
00f0-00ff : fpu
0170-0177 : 0000:00:1f.1
  0170-0177 : piix
01f0-01f7 : 0000:00:1f.1
  01f0-01f7 : piix
0376-0376 : 0000:00:1f.1
  0376-0376 : piix
03c0-03df : vga+
03f0-03f1 : pnp 00:03
03f6-03f6 : 0000:00:1f.1
  03f6-03f6 : piix
04d0-04d1 : pnp 00:03
0cf8-0cff : PCI conf1
b000-b01f : 0000:00:1f.4
  b000-b01f : uhci_hcd
b400-b41f : 0000:00:1f.2
  b400-b41f : uhci_hcd
b800-b80f : 0000:00:1f.1
  b800-b80f : piix
d000-dfff : PCI Bus 0000:01
  d800-d83f : 0000:01:0c.0
    d800-d83f : e1000
e400-e47f : 0000:00:1f.0
  e400-e47f : pnp 00:04
    e400-e403 : ACPI PM1a_EVT_BLK
    e404-e405 : ACPI PM1a_CNT_BLK
    e408-e40b : ACPI PM_TMR
    e428-e42b : ACPI GPE0_BLK
e800-e80f : 0000:00:1f.3
  e800-e80f : i801_smbus
ec00-ec3f : 0000:00:1f.0
  ec00-ec3f : pnp 00:04

[-- Attachment #1.1.4: 2.6.30-isadump.txt --]
[-- Type: text/plain, Size: 274 bytes --]

# isadump -f 0x290 16
WARNING! Running this program can cause system crashes, data loss and worse!
I will probe address range 0x290 to 0x29f.
Continue? [Y/n] y
       0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
0290: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff

[-- Attachment #1.1.5: 2.6.33-dmidecode-baseboard-system.txt --]
[-- Type: text/plain, Size: 536 bytes --]

# dmidecode 2.9
SMBIOS 2.3 present.

Handle 0x0001, DMI type 1, 25 bytes
System Information
	Manufacturer: System Manufacturer
	Product Name: System Name
	Version: System Version
	Serial Number: SYS-1234567890
	UUID: Not Settable
	Wake-up Type: Power Switch

Handle 0x0002, DMI type 2, 8 bytes
Base Board Information
	Manufacturer: ASUSTeK Computer INC.
	Product Name: CUSL2
	Version: REV 1.xx
	Serial Number: xxxxxxxxxxx

Handle 0x002E, DMI type 32, 11 bytes
System Boot Information
	Status: No errors detected


[-- Attachment #1.1.6: 2.6.33-ioports.txt --]
[-- Type: text/plain, Size: 972 bytes --]

0000-001f : dma1
0020-0021 : pic1
0040-0043 : timer0
0050-0053 : timer1
0060-0060 : keyboard
0064-0064 : keyboard
0070-0073 : rtc0
0080-008f : dma page reg
00a0-00a1 : pic2
00c0-00df : dma2
00f0-00ff : fpu
0170-0177 : 0000:00:1f.1
  0170-0177 : piix
01f0-01f7 : 0000:00:1f.1
  01f0-01f7 : piix
0376-0376 : 0000:00:1f.1
  0376-0376 : piix
03c0-03df : vga+
03f0-03f1 : pnp 00:03
03f6-03f6 : 0000:00:1f.1
  03f6-03f6 : piix
04d0-04d1 : pnp 00:03
0cf8-0cff : PCI conf1
b000-b01f : 0000:00:1f.4
  b000-b01f : uhci_hcd
b400-b41f : 0000:00:1f.2
  b400-b41f : uhci_hcd
b800-b80f : 0000:00:1f.1
  b800-b80f : piix
d000-dfff : PCI Bus 0000:01
  d800-d83f : 0000:01:0c.0
    d800-d83f : e1000
e400-e47f : 0000:00:1f.0
  e400-e47f : pnp 00:04
    e400-e403 : ACPI PM1a_EVT_BLK
    e404-e405 : ACPI PM1a_CNT_BLK
    e408-e40b : ACPI PM_TMR
    e428-e42b : ACPI GPE0_BLK
e800-e80f : 0000:00:1f.3
ec00-ec3f : 0000:00:1f.0
  ec00-ec3f : pnp 00:04

[-- Attachment #1.1.7: 2.6.33-isadump.txt --]
[-- Type: text/plain, Size: 274 bytes --]

# isadump -f 0x290 16
WARNING! Running this program can cause system crashes, data loss and worse!
I will probe address range 0x290 to 0x29f.
Continue? [Y/n] y
       0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
0290: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 153 bytes --]

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Luca Tettamanti]

On Mon, Mar 1, 2010 at 1:48 PM, Roman Mamedov <roman@rm.pp.ru> wrote:
>> * The kernel log messages when cycling the i2c-i801 driver ("rmmod
>>   i2c-i801 && modprobe i2c-i801" as root)
>
> There are no messages in dmesg or syslog when removing and reloading this
> driver in 2.6.30, but in 2.6.33 there are errors:
>
> [  173.468108] ACPI: I/O resource 0000:00:1f.3 [0xe800-0xe80f] conflicts with
> ACPI region SM00 [0xe800-0xe806]
> [  173.468125] ACPI: If an ACPI driver is available for this device, you
> should use it instead of the native driver

From ioports:
e800-e80f : 0000:00:1f.3
  e800-e80f : i801_smbus

So the ports used by i801_smbus are actually claimed by ACPI; 2.6.30
did not enforce the reservation, .33 does.
You might be able to use asus_atk0110 ACPI driver on your board; if it
does not work please send me a copy of /sys/firmware/acpi/tables/DSDT,

Luca

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Roman Mamedov]

[-- Attachment #1.1.1: Type: text/plain, Size: 1347 bytes --]

On Mon, 1 Mar 2010 13:55:48 +0100
Luca Tettamanti <kronos.it@gmail.com> wrote:

> On Mon, Mar 1, 2010 at 1:48 PM, Roman Mamedov <roman@rm.pp.ru> wrote:
> >> * The kernel log messages when cycling the i2c-i801 driver ("rmmod
> >>   i2c-i801 && modprobe i2c-i801" as root)
> >
> > There are no messages in dmesg or syslog when removing and reloading this
> > driver in 2.6.30, but in 2.6.33 there are errors:
> >
> > [  173.468108] ACPI: I/O resource 0000:00:1f.3 [0xe800-0xe80f] conflicts
> > with ACPI region SM00 [0xe800-0xe806]
> > [  173.468125] ACPI: If an ACPI driver is available for this device, you
> > should use it instead of the native driver
> 
> From ioports:
> e800-e80f : 0000:00:1f.3
>   e800-e80f : i801_smbus
> 
> So the ports used by i801_smbus are actually claimed by ACPI; 2.6.30
> did not enforce the reservation, .33 does.
> You might be able to use asus_atk0110 ACPI driver on your board;

If I understood you correctly:

# modprobe asus_atk0110
# lsmod | grep asus
asus_atk0110            9890  0 
# sensors
No sensors found!
Make sure you loaded all the kernel drivers you need.
Try sensors-detect to find out which these are.

...this didn't seem to work.

> if it
> does not work please send me a copy of /sys/firmware/acpi/tables/DSDT,

Attached.

-- 
With respect,
Roman

[-- Attachment #1.1.2: DSDT.gz --]
[-- Type: application/x-gzip, Size: 4782 bytes --]

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 153 bytes --]

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Luca Tettamanti]

On Mon, Mar 1, 2010 at 2:49 PM, Roman Mamedov <roman@rm.pp.ru> wrote:
> On Mon, 1 Mar 2010 13:55:48 +0100
> Luca Tettamanti <kronos.it@gmail.com> wrote:
>> On Mon, Mar 1, 2010 at 1:48 PM, Roman Mamedov <roman@rm.pp.ru> wrote:
>> >> * The kernel log messages when cycling the i2c-i801 driver ("rmmod
>> >>   i2c-i801 && modprobe i2c-i801" as root)
>> >
>> > There are no messages in dmesg or syslog when removing and reloading this
>> > driver in 2.6.30, but in 2.6.33 there are errors:
>> >
>> > [  173.468108] ACPI: I/O resource 0000:00:1f.3 [0xe800-0xe80f] conflicts
>> > with ACPI region SM00 [0xe800-0xe806]
>> > [  173.468125] ACPI: If an ACPI driver is available for this device, you
>> > should use it instead of the native driver
>>
>> From ioports:
>> e800-e80f : 0000:00:1f.3
>>   e800-e80f : i801_smbus
>>
>> So the ports used by i801_smbus are actually claimed by ACPI; 2.6.30
>> did not enforce the reservation, .33 does.
>> You might be able to use asus_atk0110 ACPI driver on your board;
>
> If I understood you correctly:
>
> # modprobe asus_atk0110
> # lsmod | grep asus
> asus_atk0110            9890  0
> # sensors
> No sensors found!
> Make sure you loaded all the kernel drivers you need.
> Try sensors-detect to find out which these are.
>
> ...this didn't seem to work.

Unfortunately the ATK0110 is not present. The registers in the ICH are
used by ACPI for \_SI (System Indicators). Linux uses \_SI._SST to
change the system status indicator when suspending and resuming, so
the race condition window is pretty small. The enforcement of ACPI
reservations was done to avoid situations where the same ports could
be accessed by both ACPI and native drivers (without coordination).
To revert to the old behaviour you can pass
"acpi_enforce_resources=lax" on the kernel command line, but I cannot
guarantee that it's a safe operation.

Luca

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Roman Mamedov]

[-- Attachment #1.1: Type: text/plain, Size: 614 bytes --]

On Mon, 1 Mar 2010 15:25:39 +0100
Luca Tettamanti <kronos.it@gmail.com> wrote:

> To revert to the old behaviour you can pass
> "acpi_enforce_resources=lax" on the kernel command line, but I cannot
> guarantee that it's a safe operation.

Thanks, this helped. I wonder if similar situations could be detected and
pointed at during sensors-detect. E.g. failures to load modules because of
resource conflicts, if you see my earlier sensors-detect log, it is
completely silent about being unable to insert the module because of the
conflict (which is reported in dmesg only).

-- 
With respect,
Roman

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 153 bytes --]

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Jean Delvare]

Hi Roman,

On Mon, 1 Mar 2010 19:46:58 +0500, Roman Mamedov wrote:
> On Mon, 1 Mar 2010 15:25:39 +0100
> Luca Tettamanti <kronos.it@gmail.com> wrote:
> 
> > To revert to the old behaviour you can pass
> > "acpi_enforce_resources=lax" on the kernel command line, but I cannot
> > guarantee that it's a safe operation.
> 
> Thanks, this helped. I wonder if similar situations could be detected and
> pointed at during sensors-detect. E.g. failures to load modules because of
> resource conflicts, if you see my earlier sensors-detect log, it is
> completely silent about being unable to insert the module because of the
> conflict (which is reported in dmesg only).

This is on my to-do list, but I have no idea how to implement it. The
problem is that the list of ACPI-reserved resources is not exposed to
user-space, contrary to the standard I/O port requests (which show
in /proc/ioports). This prevents sensors-detect from warning the user.

So we would need either ACPI to export its resource list to user-space,
or a helper kernel driver basically exposing the ACPI resource tester
function to user-space. I don't know what the ACPI people will find
acceptable.

Luca, do you have any idea on the matter? Maybe I am overlooking a more
elegant or simple solution...

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] ASUS CUSL2: sensors work with 2.6.30,

[Luca Tettamanti]

On Mon, Mar 1, 2010 at 3:53 PM, Jean Delvare <khali@linux-fr.org> wrote:
> Hi Roman,
>
> On Mon, 1 Mar 2010 19:46:58 +0500, Roman Mamedov wrote:
>> On Mon, 1 Mar 2010 15:25:39 +0100
>> Luca Tettamanti <kronos.it@gmail.com> wrote:
>>
>> > To revert to the old behaviour you can pass
>> > "acpi_enforce_resources=lax" on the kernel command line, but I cannot
>> > guarantee that it's a safe operation.
>>
>> Thanks, this helped. I wonder if similar situations could be detected and
>> pointed at during sensors-detect. E.g. failures to load modules because of
>> resource conflicts, if you see my earlier sensors-detect log, it is
>> completely silent about being unable to insert the module because of the
>> conflict (which is reported in dmesg only).
>
> This is on my to-do list, but I have no idea how to implement it. The
> problem is that the list of ACPI-reserved resources is not exposed to
> user-space, contrary to the standard I/O port requests (which show
> in /proc/ioports). This prevents sensors-detect from warning the user.
>
> So we would need either ACPI to export its resource list to user-space,
> or a helper kernel driver basically exposing the ACPI resource tester
> function to user-space. I don't know what the ACPI people will find
> acceptable.

It should be easy to modify acpi_os_{,in}validate_address to expose
the internal resources list through sysfs.

> Luca, do you have any idea on the matter? Maybe I am overlooking a more
> elegant or simple solution...

I could be done in userspace disassembling the DSDT (but it probably
does not meet the "simple" requirement).

Luca

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors