Re: om 2.6 Opteron last git, trying to migrate a simple task

Niels de Vos wrote:
> Hi Florian,

> as far as I know OM is _not_ preemtive safe. Checking this is on the
> ToDo list: <http://openmosix.snarc.org/wakka.php?wiki=ToDo>

>   
Ok, I tried with a non preemtible kernel, but I get the same result, on
the other side, it seems that (192.168.169.11) I get two messages on the
console :

[OMBUG] got -32 of 8
[OMBUG] remote_do_comm failed

The last version of git does not include kcomd patch ... I was expecting
this version to do at least a simple migration without crashing ...

I am setting up the omtoolkit with qemu to see if I can get some better
results (and a better understanding of the problem).

Matt, is the version with kcomd better already ? and where are you in
the 2.6.17 port ?? I don't think it is interesting to spend up all my
time trying to debug a problem that would (or should) be solved in the
kcomd version of openmosix.


> I can't help you with your debugging 
> questions though.

> Cheers,
> Niels

> On Mon, 2006-08-28 at 16:54 +0200, Florian Delizy wrote:
>   
>> Hi,

>> I am getting at it right now, for a simple test, I am trying to migrate 
>> a simple task from a node to another node, and it seems not to work at 
>> all. Instead of migrating the process, it gives me an "Oups" :


>> When I try to migrate the process using the command:

>> echo 192.168.169.11 > /proc/7623/om/where

>> I get this :

>> |-------------------- OUPS start here : 
>> ----------------------------------------------------------------------------|

>> Unable to handle kernel NULL pointer dereference at 0000000000000000 RIP:
>> <0000000000000000>{stext+2145382632}
>> PGD 13d667067 PUD 13d6fa067 PMD 0
>> Oops: 0010 [1] PREEMPT

>> CPU 0
>> Modules linked in: thermal processor fan
>> Pid: 7623, comm: eatcpu Not tainted 2.6.17-om #1
>> RIP: 0010:[<0000000000000000>] <0000000000000000>{stext+2145382632}
>> RSP: 0000:ffff81013d7cbef8  EFLAGS: 00010292
>> RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff81013e48b0c0
>> RDX: 0000000000001000 RSI: 00007fff82c0b670 RDI: ffff81013f6207e0
>> RBP: 0000000000000000 R08: 0000000000000218 R09: 000000000000fa57
>> R10: 0000000000000000 R11: ffff81013ead7c00 R12: ffff81013e48b0c0
>> R13: 00007fff82c0b670 R14: 0000000000000000 R15: 0000000000000000
>> FS:  00002b4c281f36d0(0000) GS:ffffffff807f9000(0000) knlGS:0000000000000000
>> CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
>> CR2: 0000000000000000 CR3: 000000013d798000 CR4: 00000000000006e0
>> Process eatcpu (pid: 7623, threadinfo ffff81013d7ca000, task 
>> ffff81013f6207e0)
>> Stack: 0000000a00000009 0000000000000000 0000000000000000 ffffffff8032c87e
>>        ffff81013f6207e0 ffffffff80329129 0000000000000202 ffffffff803292eb
>>        ffff81013f6207e0 ffffffff80328c67
>> Call Trace: <ffffffff8032c87e>{task_move_to_node+41}
>>        <ffffffff80329129>{task_request_move+40} 
>> <ffffffff803292eb>{task_do_request+37}
>>        <ffffffff80328c67>{openmosix_pre_usermode+28} 
>> <ffffffff8020aa13>{retint_careful+45}

>> Code:  Bad RIP value.
>> RIP <0000000000000000>{stext+2145382632} RSP <ffff81013d7cbef8>
>> CR2: 0000000000000000

>> |-------------------- OUPS ends here 
>> ----------------------------------------------------------------------------|


>> The code I am trying to migrate is a simple C program :


>> int main( int argc, char **argv )
>> {
>>         double total = 456114157974121;
>>         double min = 0;
>>         double max = total;

>>         double square = ( min + max ) / 2.0;
>>         while ( total != square * square )
>>         {
>>                 if ( (square*square) < total )
>>                 {
>>                         min = square;
>>                 } else if (square*square > total ) {
>>                         max = square;
>>                 }
>>                 square = ( min + max ) / 2;

>>         }
>>         return 0;
>> }



>> Kernel is a 2.6.17 (it gives me the same result with a 2.6.17.8), with 
>> the preempt kernel option. I will try to get some more interesting 
>> output, but first, is openmosix preemtive safe ??

>> -------------------------------------------------------------------------
>> Using Tomcat but need to do more? Need to support web services, security?
>> Get stuff done quickly with pre-integrated technology to make your job easier
>> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>> _______________________________________________
>> openMosix-devel mailing list
>> openMosix-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
>> https://lists.sourceforge.net/lists/listinfo/openmosix-devel

>>     



-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642