Closed jwrdegoede closed 6 years ago
Hmm, I'm testing some sensor stuff on another device now (Lenovo Miix 320), the sensor there only does polling, but it does have a sampling_frequency attribute (without a prefix) and that one defaults to 1 (Hz) which is somewhat slow. I wonder if we should do something about that too?
Yeah, the 1Hz really is no good:
[root@localhost iio:device0]# time cat inaccel?_raw -1 59 -20
real 0m11,932s user 0m0,000s sys 0m0,012s
So the reads while polling block until the device has gotten a sample, which is really really slow at 10Hz, since iio-sensor-proxy uses blocking reads while reading the sysfs files this blocks all of iio-sensor-proxy and anything making requests to iio-sensor-proxy is also waiting this long. E.g. monitor-sensor takes about 20 second before it prints: === Has accelerometer (orientation: right-up) === No ambient light sensor`
At 10Hz it is better (still not great though) : [root@localhost iio:device0]# time cat inaccel?_raw 0 57 -25
real 0m1,438s user 0m0,002s sys 0m0,007s
I'm preparing a new version of the sample_freq patch to fix this.
Ok I just pushed a new version (weird that github does not automatically adds a comment for this).
Let me know what you think of the new version.
Ok, I just pushed a new version adding the exact compile error from gcc (and dropping the .) and adding a #define for the 10 Hz minimum sample-freq.
one which actually expects arguments, causing it to throw erros like these:
typo, "errors"
Looks good to merge after those.
On 17-04-18 12:41, hadess wrote:
typo, "errors"
Fixed.
I was too quick :) Also added the / Hz / comment now.
As usual, thanks! :)
Hi Bastien,
I've been working today on getting the sensors on Asus T100HA devices to work properly, the second commit in this set is the result of this.
When I went to test my patch I would not compile, so the first patch in this set fixes compiling with gcc8.
Regards,
Hans