linux at horizon.com wrote:
> On my motherboard, bank 1 input 13 is not connected to 3VDual, but rather
> to the +12V line via a 23/96 voltage divider. (At least, that's the
> ratio that makes the raw in10 limit voltages of 2.3V and 3.45V match
> the 9.6 and 14.4 V limits reported by the BIOS.)
>
> I had more significant problems with +5SB on input 9 (in6). This is
> labelled "AGP VDDQ", and appears to be connected straight through,
> but is a 3.3V level on my system. That's very close to the 3.5V Vref,
> leading to a raw reading of 241.
>
> That makes the sensor type determination say:
>
>> abituguru: bank1-sensor: 9 reading (241) too close to limits, unable to determine sensor type, skipping sensor
>
> Changing the code to have a limit of 245 (using a min and max of 250 and 253)
> seems to work, and saves me an annoying bank1_types=-1,...,-1,0 parameter.
>
> Then I get:
> AGP VDDQ Voltage: +3.30 V (min +3.00 V, max +3.49 V)
> ATX +5V: +5.10 V (min +4.75 V, max +5.24 V)
> ATX +3.3V: +3.33 V (min +3.13 V, max +3.47 V)
>
> Not that I have an AGP slot, but it's something.
>
> Actually, couldn't the code be amended to set the maximum very low if the
> reading is high and detect a voltage sensor with ANY reading that way?
>
> Something like (diff-like markers inserted by hand):
>
> ABIT_UGURU_DEBUG(2, "testing bank1 sensor %d\n", (int)sensor_addr);
> /* Volt sensor test: enable volt low alarm, set min value ridicously
> high. If its a volt sensor this should always give us an alarm.
> If the voltage is high, do the opposite. */
>
> ! if (val < 128) {
> ! buf[0] = ABIT_UGURU_VOLT_LOW_ALARM_ENABLE;
> ! buf[1] = 245; /* Very high minimum value */
> ! buf[2] = 250;
> ! flag = ABIT_UGURU_VOLT_LOW_ALARM_FLAG;
> ! } else {
> ! buf[0] = ABIT_UGURU_VOLT_HIGH_ALARM_ENABLE;
> ! buf[1] = 5;
> ! buf[2] = 10; /* Very low maximum value */
> ! flag = ABIT_UGURU_VOLT_HIGH_ALARM_FLAG;
> ! }
> if (abituguru_write(data, ABIT_UGURU_SENSOR_BANK1 + 2, sensor_addr,
> buf, 3) != 3)
> goto abituguru_detect_bank1_sensor_type_exit;
> /* Now we need 20 ms to give the uguru time to read the sensors
> and raise a voltage alarm */
> set_current_state(TASK_UNINTERRUPTIBLE);
> schedule_timeout(HZ/50);
> /* Check for alarm and check the alarm is a volt low alarm. */
> if (abituguru_read(data, ABIT_UGURU_ALARM_BANK, 0, buf, 3,
> ABIT_UGURU_MAX_RETRIES) != 3)
> goto abituguru_detect_bank1_sensor_type_exit;
> if (buf[sensor_addr/8] & (0x01 << (sensor_addr % 8))) {
> if (abituguru_read(data, ABIT_UGURU_SENSOR_BANK1 + 1,
> sensor_addr, buf, 3,
> ABIT_UGURU_MAX_RETRIES) != 3)
> goto abituguru_detect_bank1_sensor_type_exit;
> ! if (buf[0] & flag) {
> ABIT_UGURU_DEBUG(2, " found volt sensor\n");
> ret = ABIT_UGURU_IN_SENSOR;
> goto abituguru_detect_bank1_sensor_type_exit;
> } else
> ABIT_UGURU_DEBUG(2, " alarm raised during volt "
> "sensor test, but volt range flag not set\n");
> } else
> ABIT_UGURU_DEBUG(2, " alarm not raised during volt sensor "
> "test\n");
>
Something like that should work yes, good idea. Attached is a version with this
implemented. I've attached a makefile for easy out of tree compilation too,
just drop the makefile and abituguru.c in a dir, type make and you should get
an abituguru.ko to insmod.
Please let me know how this works for you. Notice that I only use the volt high
test for readings > 240, as this is delicated code and I do not want to change
the code path for already working setups.
Regards,
Hans
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile
Url: http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070617/5a162df4/attachment.pl
-------------- next part --------------
A non-text attachment was scrubbed...
Name: abituguru.c
Type: text/x-csrc
Size: 53133 bytes
Desc: not available
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070617/5a162df4/attachment.bin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hwmon-abituguru-volt-test.patch
Type: text/x-patch
Size: 2366 bytes
Desc: not available
Url : http://lists.lm-sensors.org/pipermail/lm-sensors/attachments/20070617/5a162df4/attachment-0001.bin
