Porting Fiasco+L4re to DaVinci DA850 - Linux time wrong
I've now got L4Linux running on top of Fiasco+L4re on the DA850. However there appears to be something wrong with the time. Time seems to be passing at a rate of 10-1 (i.e. 1 elapsed second == 10 reported seconds) in Linux. If I change the default Hz rating the L4Linux config (from 100Hz) in either direction L4Linux either hangs or slows down dramatically. If I change the period in the timer driver (in the Fiasco bsp directory) things hang earlier. For the Fiasco BSP I'm using: * timer_tick-single-vector for timer_tick_IMPL, * clock_generic for clock_IMPL, * a da850 timer driver for timer_IMPL. Can someone enlighten me as to how it should work? Thanks -- Peter Howard <pjh@northern-ridge.com.au>
On Tue May 20, 2014 at 10:47:03 +1000, Peter Howard wrote:
I've now got L4Linux running on top of Fiasco+L4re on the DA850. However there appears to be something wrong with the time. Time seems to be passing at a rate of 10-1 (i.e. 1 elapsed second == 10 reported seconds) in Linux. If I change the default Hz rating the L4Linux config (from 100Hz) in either direction L4Linux either hangs or slows down dramatically. If I change the period in the timer driver (in the Fiasco bsp directory) things hang earlier.
For the Fiasco BSP I'm using: * timer_tick-single-vector for timer_tick_IMPL, * clock_generic for clock_IMPL, * a da850 timer driver for timer_IMPL.
Can someone enlighten me as to how it should work?
I wonder whether Fiasco's timer has the proper frequency. hello's "Hello World" should be displayed once a second. Is that the case or is it more/less often? Adam -- Adam adam@os.inf.tu-dresden.de Lackorzynski http://os.inf.tu-dresden.de/~adam/
On Wed, 2014-05-21 at 00:50 +0200, Adam Lackorzynski wrote:
On Tue May 20, 2014 at 10:47:03 +1000, Peter Howard wrote:
I've now got L4Linux running on top of Fiasco+L4re on the DA850. However there appears to be something wrong with the time. Time seems to be passing at a rate of 10-1 (i.e. 1 elapsed second == 10 reported seconds) in Linux. If I change the default Hz rating the L4Linux config (from 100Hz) in either direction L4Linux either hangs or slows down dramatically. If I change the period in the timer driver (in the Fiasco bsp directory) things hang earlier.
For the Fiasco BSP I'm using: * timer_tick-single-vector for timer_tick_IMPL, * clock_generic for clock_IMPL, * a da850 timer driver for timer_IMPL.
Can someone enlighten me as to how it should work?
I wonder whether Fiasco's timer has the proper frequency. hello's "Hello World" should be displayed once a second. Is that the case or is it more/less often?
It spits out "Hello World!" _way_ faster than once per second. Something like 15-20 times a second. So the problem is in Fiasco.
Adam
-- Peter Howard <pjh@northern-ridge.com.au>
On Wed, 2014-05-21 at 09:03 +1000, Peter Howard wrote:
On Wed, 2014-05-21 at 00:50 +0200, Adam Lackorzynski wrote:
On Tue May 20, 2014 at 10:47:03 +1000, Peter Howard wrote:
I've now got L4Linux running on top of Fiasco+L4re on the DA850. However there appears to be something wrong with the time. Time seems to be passing at a rate of 10-1 (i.e. 1 elapsed second == 10 reported seconds) in Linux. If I change the default Hz rating the L4Linux config (from 100Hz) in either direction L4Linux either hangs or slows down dramatically. If I change the period in the timer driver (in the Fiasco bsp directory) things hang earlier.
For the Fiasco BSP I'm using: * timer_tick-single-vector for timer_tick_IMPL, * clock_generic for clock_IMPL, * a da850 timer driver for timer_IMPL.
Can someone enlighten me as to how it should work?
I wonder whether Fiasco's timer has the proper frequency. hello's "Hello World" should be displayed once a second. Is that the case or is it more/less often?
It spits out "Hello World!" _way_ faster than once per second. Something like 15-20 times a second. So the problem is in Fiasco.
OK, tuning the Interval value in my timer class so that "Hello World!" started coming out at the proper 1 second interval fixed the time in L4Linux. I'd played with that value before my post, but clearly over-compensated. Thanks. -- Peter Howard <pjh@northern-ridge.com.au>
participants (2)
-
Adam Lackorzynski -
Peter Howard