Conversation
|
I'm trying to test in a cc2538dk board applying the same modifications without success... Why it should only work for remote boards? |
|
mhm, I'm unsure as well - because I was pretty sure I tested various settings yesterday, including 16Bit with 1MHz and it didn't work. However, today it worked 😕 |
|
I think we have cc2538dk here, too - so I'll try to test that, too |
|
|
|
I confirm that both tests are also not running on my cc2538dk. |
|
I just tested all xtimer tests and no one worked. I have a remote-reva. |
|
oops, during my debugging and tests session I found that only the cc2538 was missing the Please test again @PeterKietzmann, @kYc0o, and @A-Paul ! |
|
I think this was caused when the old |
|
to give a bit more rational on why |
| @@ -0,0 +1 @@ | |||
| FEATURES_PROVIDED += periph_pm | |||
There was a problem hiding this comment.
NO its one essential part of the fix! Read my explanation above, otherwise everything gets stuck in idle thread. I could make a separate PR for this, but its already a commit and as said its part of the fix.
|
I ran |
23010dd to
e0f7a28
Compare
|
On cc2538dk all tests/xtimer_* look good now. |
e0f7a28 to
40431fe
Compare
kYc0o
left a comment
There was a problem hiding this comment.
ACK. All xtimer tests passed.
|
Jenkins fails unrelated and both Murdocks succeeded. ACK and go |
|
finally 🎉 |
|
Thanks @smlng ! |
| #define XTIMER_SHIFT (4) | ||
| #define XTIMER_HZ (16000000UL) | ||
| #define XTIMER_SHIFT (0) | ||
| #define XTIMER_HZ (10000000UL) |
There was a problem hiding this comment.
You're sure this has to be ten times higher than on the other boards?
There was a problem hiding this comment.
this was already fixed before merge! Reload the page and have another look?
This PR fixes #6419, at least for me.
It basically reverts xtimer configuration introduces in #5608, and resets XTIMER_SHIFT to 0 and XTIMER_HZ 1000000 (1 MHz). I have to admit, I'm not sure why this helps, but it does. In the last days I made extensive tests using the hardware timer and xtimer.
The cc2538 has 4 timers each can be configured in 16Bit mode with 2 channels and 32 Bit mode with 1 channel. For the latter only a frequency of 16 MHz is allowed, as prescaler is only available in 16 Bit mode. However, using 16Bit timer with 16Mhz or a 32Bit timer with 16 MHz the tests
xtimer_driftandxtimer_periodic_wakeupalways fail, andxtimer_usleepfor 16Bit + 16MHz, too. All tests succeeded when using a 16 Bit timer with 1 MHz (which is the old setting).Can someone please confirm?