Re: [Xenomai] Fwd: Kernel Compilation Problems

[Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>]

On 01/17/2014 09:44 PM, Andy Pugh wrote:
>> On 16 January 2014 22:43, Gilles Chanteperdrix
>> <gilles.chanteperdrix@xenomai.org> wrote:
>>
>>> Does the
>>> unpatched kernel, with exactly the same configuration, compile correctly?
> 
> I had a feeling that I was very close, so I commented out the two
> calls each to the troublesome functions, figuring that for the
> time-being I would live without hdmi-audio and fec ethernet.
> 
> After that, the patched Udoo git repository kernel did compile, and
> boots the board. I need to figure out how to tidy up the kernel string
> (all changes _are_ committed).
> 
> How does this look?
> 
> 
> [    1.397848] I-pipe: Domain Xenomai registered.
> [    1.402368] Xenomai: hal/arm started.
> [    1.406363] Xenomai: scheduling class idle registered.
> [    1.411520] Xenomai: scheduling class rt registered.
> [    1.424750] Xenomai: real-time nucleus v2.6.3 (Lies and Truths) loaded.
> [    1.431387] Xenomai: debug mode enabled.
> [    1.435777] Xenomai: starting native API services
> [    1.440587] Xenomai: starting POSIX services.
> [    1.445072] Xenomai: starting RTDM services.
> 
> andypugh@udoo:~/git/Kernel_Unico$ ls /proc/xenomai/
> 
> acct  affinity  apc  faults  heap  interfaces  irq  latency  lock
> registry  rtdm  sched  schedclasses  stat  timebases  timer  timerstat
>  version
> 
> And the testsuite:
> 
> andypugh@udoo:~/git/Xenomai$ uname -a
> 
> Linux udoo 3.0.35-Xenomai-g58cf7fe-dirty #13 SMP PREEMPT Fri Jan 17
> 03:26:53 UTC 2014 armv7l armv7l armv7l GNU/Linux
> 
> andypugh@udoo:~/git/Xenomai$ sudo xeno-regression-test -l
> "/usr/lib/xenomai/testsuite/dohell -m /tmp 100" -t 2
> 
> Started child 9158: /bin/bash
> /usr/lib/xenomai/testsuite/xeno-test-run-wrapper
> /usr/bin/xeno-regression-test -t 2
> 
> ++ echo 0
> ++ /usr/lib/xenomai/testsuite/arith
> mul: 0x79364d93, shft: 26
> integ: 30, frac: 0x4d9364d9364d9364
> signed positive operation: 0x03ffffffffffffff * 1000000000 / 33000000
> inline calibration: 0x0000000000000000: 45.615 ns, rejected 0/10000
> inlined llimd: 0x79364d9364d9362f: 1727.421 ns, rejected 2/10000
> inlined llmulshft: 0x79364d92ffffffe1: 32.735 ns, rejected 0/10000
> inlined nodiv_llimd: 0x79364d9364d9362f: 45.375 ns, rejected 0/10000
> out of line calibration: 0x0000000000000000: 48.075 ns, rejected 0/10000
> out of line llimd: 0x79364d9364d9362f: 1788.676 ns, rejected 2/10000
> out of line llmulshft: 0x79364d92ffffffe1: 27.764 ns, rejected 0/10000
> out of line nodiv_llimd: 0x79364d9364d9362f: 40.441 ns, rejected 0/10000
> signed negative operation: 0xfc00000000000001 * 1000000000 / 33000000
> inline calibration: 0x0000000000000000: 45.565 ns, rejected 0/10000
> inlined llimd: 0x86c9b26c9b26c9d1: 1743.871 ns, rejected 1/10000
> inlined llmulshft: 0xd45d172d0000001e: 22.707 ns, rejected 0/10000
> inlined nodiv_llimd: 0x86c9b26c9b26c9d1: 50.464 ns, rejected 0/10000
> out of line calibration: 0x0000000000000000: 70.583 ns, rejected 0/10000
> out of line llimd: 0x86c9b26c9b26c9d1: 1751.943 ns, rejected 2/10000
> out of line llmulshft: 0xd45d172d0000001e: 10.312 ns, rejected 0/10000
> out of line nodiv_llimd: 0x86c9b26c9b26c9d1: 30.519 ns, rejected 0/10000
> 
> unsigned operation: 0x03ffffffffffffff * 1000000000 / 33000000
> inline calibration: 0x0000000000000000: 45.570 ns, rejected 0/10000
> inlined nodiv_ullimd: 0x79364d9364d9362f: 42.910 ns, rejected 0/10000
> out of line calibration: 0x0000000000000000: 48.037 ns, rejected 0/10000
> out of line nodiv_ullimd: 0x79364d9364d9362f: 40.478 ns, rejected 0/10000
> ++ /usr/lib/xenomai/testsuite/clocktest -C 42 -T 30
> == Tested clock: 42 (CLOCK_HOST_REALTIME)
> CPU      ToD offset [us] ToD drift [us/s]      warps max delta [us]
> --- -------------------- ---------------- ---------- --------------
>   0                  1.3            0.024          0            0.0
>   1                  1.0            0.040          0            0.0
>   2                  0.6           -0.046          0            0.0
>   3                  1.2            0.046          0            0.0
> 

It looks ok, you should run xeno-test instead of xeno-regression-test,
and see if you get acceptable latencies.

-- 
                                                                Gilles.