Hi, if I make any calls to mailboxes or semaphores, my application crashes. I can't even debug what happenned, all I get is "Could not stop ARM device!".
If I comment out the calls to mailboxes (e.g. os_mbx_send) and semaphores (e.g. os_sem_wait) everything works fine.
And yes, I am calling the init functions for the mailboxes and semaphores, so that's not the problem.
Any ideas? I can post some of the code if need be. Thanks.
-Mike
Welcome to the club. See my threads, here:
http://www.keil.com/forum/docs/thread15346.asp
and here:
http://www.keil.com/forum/docs/thread15337.asp
Keil are working on it. Have faith!
Thanks Tamir but I cannot wait for Keil to fix this I have to have it done right away, this is for my job.
I am using my own "mailboxes" and "semaphores" (if you can even call it that, they're really just shared variables).
But this is a pain in the rear and I would rather use the product we paid for. Not too happy right now.
well, I trust Keil to get it done. My product should run for possibly YEARS without a reset - now it dies after 3 days at most. Our client is going frigging haywire (also in the US) to get Keil to do it even faster, but we better let them do it and do it right! If you have the source code (RL-ARM license) you can try to do it yourself, but I would not recommend it. On the other hand, my test program destroys the kernel in 5 seconds of operation at 50 microseconds tick rate. Hmm, I need it to run for years and it is actually urgent. All this is a little disappointing, especially when RTX is part of a larger commercial product that you pay a lot of money for.
I understand Tamir, thanks for the input. By the way, when your application dies, do you get the same issue I do with the debugger? ("Could not stop ARM device" message)
we first saw this problem as part of a very complex application that died with no apparent reason so we could not really tell why (it was all one large spaghetti - our own mistakes were interleaved!). sometimes an undef exception, sometimes the kernel hung in setting of a task priority, you name it. so just 3 days ago I decided that I had enough of this and managed to give Keil what they have been asking me for months now (this is an old problem!) - a simple test case. it took me 15 minutes to write. I swear.
yes, i too started with a large application, but have isolated the issue, i can provide a very simple program which will cause the crash. in fact if i have time i will post it.
Please do, and please also send it to Keil - Franc Urbanc is returning from his vacation soon. He does not yet know what awaits him :-)