Hi Michael,
Now that the got my more basic gremlins are worked out, I find the i686
box in the shop also freezes with the latency-test, so clearly the
problem is with my setup and not a 32-bit vs. 64-bit issue.
I don't see the [permanent] flag set when loading the rtapi modules.
I'm going back to try to recreate your setup. Found your kernel config
file. If you have one around, please send a LinuxCNC build log to help
guess what the differences might be.
John
On 11/10/2012 12:30 PM, Michael Haberler wrote:
> John,
>
> the HAL_LIB error message you got and the divide error raise my gut
> feeling that something is wrong with in-kernel 64bit integer division on
> x32_64; clock periods are long long; maybe there's some inline assembly
> somewhere which breaks
>
> maybe that can be verified in a standalone program?
>
> ---
>
> I'd be interested if you can verify the below error as well
>
> also, can you verify you have the same strange '[permanent]' tag in lsmod as
> me?
>
> (post failure described below):
>
> root@atom:~# lsmod
> Module Size Used by
> sampler 14143 1 [permanent]
> hal_lib 39412 1 sampler,[permanent]
> rtapi 46610 2 sampler,hal_lib,[permanent]
> xeno_math 12602 1 rtapi,[permanent]
> binfmt_misc 16829 1
>
>
> -----
>
>
> meanwhile I could produce an oops during module load, but different
> from yours (null pointer deref in-kernel), with the 32bit
> 2.6.38.8-xenomai+
>
> to trigger the error:
>
> mah@atom:~/emc2-dev/tests/abs.0$ realtime start
> mah@atom:~/emc2-dev/tests/abs.0$ halcmd -f -k
> halcmd: loadrt sampler cfg=ff depth=350
> <stdin>:1: child did not exit normally
> <stdin>:1: insmod failed, returned -1
> See the output of 'dmesg' for more information.
> halcmd:
>
> (which is part of tests/abs.0, so runtests should cause it)
>
> root@atom:~# dmesg
> [ 252.764152] RTAI[math]: loaded.
> [ 260.200891] BUG: unable to handle kernel NULL pointer dereference at
> (null)
> [ 260.201030] IP: [<c1072256>] sys_init_module+0x2c6/0x1b80
> [ 260.201132] *pde = 00000000
> [ 260.201184] Oops: 0000 [#1] SMP
> [ 260.201244] last sysfs file:
> /sys/devices/pci0000:00/0000:00:1e.0/0000:05:00.0/local_cpus
> [ 260.201381] Modules linked in: sampler(+) hal_lib rtapi xeno_math
> binfmt_misc nfsd exportfs nfs lockd fscache nfs_acl auth_rpcgss sunrpc
> dm_crypt snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep snd_pcm
> snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq ppdev snd_timer
> snd_seq_device snd psmouse parport_pc serio_raw soundcore snd_page_alloc
> parport usbhid hid i915 drm_kms_helper drm ahci r8169 libahci i2c_algo_bit
> video
> [ 260.202135]
> [ 260.202164] Pid: 1779, comm: insmod Not tainted 2.6.38.8-xenomai+ #1
> /D525MW
> [ 260.202321] EIP: 0060:[<c1072256>] EFLAGS: 00010246 CPU: 0
> [ 260.202415] EIP is at sys_init_module+0x2c6/0x1b80
> [ 260.204015] EAX: 00000000 EBX: f90d9000 ECX: 00000000 EDX: 00000001
> [ 260.204015] ESI: 00000000 EDI: 00000001 EBP: f0d9bfac ESP: f0d9bf24
> [ 260.204015] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
> [ 260.204015] Process insmod (pid: 1779, ti=f0d9a000 task=f0d3dee0
> task.ti=f0d9a000)
> [ 260.204015] I-pipe domain Linux
> [ 260.204015] Stack:
> [ 260.204015] 000006dc c16d39a4 00000000 c16575e6 f90d900c f13debc4
> f90be460 00000004
> [ 260.204015] f90d84e8 0000000b f90d9048 00000025 00000001 f0d9bf5c
> f90bd000 00017cc8
> [ 260.204015] f90cf3c4 f90cf262 f90d4a4c f13d8180 00003000 00003040
> 00000000 00000000
> [ 260.204015] Call Trace:
> [ 260.204015] [<c1002ce1>] sysenter_do_call+0x12/0x16
> [ 260.204015] Code: 00 00 8b 93 e8 00 00 00 8b b3 e0 00 00 00 89 34 24 e8 1f
> e3 ff ff 8b 83 80 01 00 00 85 c0 74 19 31 f6 8d 76 00 8b 83 7c 01 00 00 <ff>
> 14 b0 83 c6 01 3b b3 80 01 00 00 72 ec 8b 83 d4 00 00 00 85
> [ 260.204015] EIP: [<c1072256>] sys_init_module+0x2c6/0x1b80 SS:ESP
> 0068:f0d9bf24
> [ 260.204015] CR2: 0000000000000000
> [ 260.366000] ---[ end trace 7dbf30905148abaa ]---
>
>
>
>
>
> Am 10.11.2012 um 08:06 schrieb John Morris:
>
>> Hi Michael,
>>
>> I've been whacking away ('hacking' is too glamorous a term for my clumsy
>> bumbling) at getting xenomai-kernel to work today. I did make some
>> progress; there's a new commit that fixes a problem where symbols in
>> xeno_math.ko are not being shared with other modules, resulting in
>> rtapi.ko not loading.
>>
>> Xenomai seems to work properly. The Xenomai 'latency' and 'xeno-test'
>> programs look great, with <3uS/<9uS latency in kernel task and user task
>> modes, respectively. I followed all the guides I could find, and added
>> '--enable-dlopen-skins' to the user tools build, per your instructions.
>>
>> The LinuxCNC latency-test, however, results in one of a few bad endings.
>>
>> - Exits after writing the lines from 'Symptom 1' below into syslog; this
>> probably happens on the first run after a reboot.
>>
>> - (One time only) a kernel oops, see 'Symptom 2' below; system kept running
>>
>> - Machine hangs; CPU fan spins at max speed until system reset.
>>
>> Let me know what bells these ring.
>>
>> John
>>
>>
>>
>> Symptom 1:
>> RTAI[math]: loaded.
>> HAL_LIB: ERROR: clock period too long: 1000000000000
>> THREADS: ERROR: could not create thread 'slow'
>> RTAI[math]: unloaded.
>>
>> Symptom 2:
>> RTAI[math]: loaded.
>> divide error: 0000 [#1] PREEMPT SMP
>> last sysfs file: /sys/module/xeno_nucleus/parameters/xenomai_gid
>> CPU 0
>> Modules linked in: threads(+) hal_lib rtapi xeno_math nfs fscache
>> nfs_acl auth_rpcgss autofs4 lockd sunrpc ipv6 uinput ppdev parport_pc
>> parport sg pcspkr k8temp edac_core edac_mce_amd forcedeth i2c_nforce2
>> ext4 mbcache jbd2 raid1 firewire_ohci firewire_core crc_itu_t sd_mod
>> crc_t10dif sr_mod cdrom pata_amd pata_acpi ata_generic sata_nv radeon
>> ttm drm_kms_helper drm i2c_algo_bit i2c_core dm_mirror dm_region_hash
>> dm_log dm_mod [last unloaded: xeno_math]
>>
>> Pid: 17143, comm: insmod Not tainted 2.6.38.8-2.xenomai.el6.x86_64 #1
>> /M61P-S3
>> RIP: 0010:[<ffffffffa04aad0c>] [<ffffffffa04aad0c>]
>> hal_create_thread+0x1fc/0x530 [hal_lib]
>> RSP: 0000:ffff8801c09afea8 EFLAGS: 00010256
>> RAX: 00000000000186a0 RBX: ffffc90004ffde58 RCX: 0000000000000000
>> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000063
>> RBP: ffff8801c09aff08 R08: 0000000000000000 R09: 0000000000000000
>> R10: 0000000000000000 R11: 0000000000000000 R12: ffff8801c097f410
>> R13: 00000000000186a0 R14: 0000000000000001 R15: ffffc90004ffde80
>> FS: 00007f19885c1700(0000) GS:ffff8800dfc00000(0000) knlGS:0000000000000000
>> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 00007f7109cc90f0 CR3: 00000001c09e2000 CR4: 00000000000006f0
>> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
>> Process insmod (pid: 17143, threadinfo ffff8801c09ac000, task
>> ffff88020e728e40)
>> Stack:
>> ffff8801c09afee8 ffffffff814b67c4 657268745f4c4148 ffffffff00736461
>> ffffc90004fbe000 0000000000000001 ffffffffa04bd540 ffffffffa04bd540
>> 0000000000000000 ffffffffa04bd020 0000000001276050 0000000000000004
>> Call Trace:
>> [<ffffffff814b67c4>] ? notifier_call_chain+0x54/0x80
>> [<ffffffffa04bd020>] ? init_module+0x0/0x1f4 [threads]
>> [<ffffffffa04bd121>] init_module+0x101/0x1f4 [threads]
>> [<ffffffff810001e3>] do_one_initcall+0x43/0x1a0
>> [<ffffffff8109678a>] sys_init_module+0xba/0x200
>> [<ffffffff81002fff>] system_call_fastpath+0x16/0x1b
>> Code: 48 8b 48 08 8b 78 10 48 8b b2 98 00 00 00 49 39 f5 0f 82 61 02 00
>> 00 48 89 f0 31 d2 48 c1 e8 3f 48 01 f0 48 d1 f8 49 8d 44 05 00 <48> f7
>> f6 48 98 48 0f af c6 48 39 c1 48 89 43 08 0f 8f cb 01 00
>> RIP [<ffffffffa04aad0c>] hal_create_thread+0x1fc/0x530 [hal_lib]
>> RSP <ffff8801c09afea8>
>> ---[ end trace 3a9286738b179516 ]---
>>
>> ------------------------------------------------------------------------------
>> Everyone hates slow websites. So do we.
>> Make your web apps faster with AppDynamics
>> Download AppDynamics Lite for free today:
>> http://p.sf.net/sfu/appdyn_d2d_nov
>> _______________________________________________
>> Emc-developers mailing list
>> Emc-developers@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-developers
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_nov
> _______________________________________________
> Emc-developers mailing list
> Emc-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-developers
>
------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov
_______________________________________________
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
