Modifies error text reported from RTX from 'underflow' to 'overflow' #6173
Conversation
| @@ -45,7 +45,7 @@ __NO_RETURN uint32_t osRtxErrorNotify (uint32_t code, void *object_id) | |||
| switch (code) { | |||
| case osRtxErrorStackUnderflow: | |||
| // Stack underflow detected for thread (thread_id=object_id) | |||
| error("CMSIS-RTOS error: Stack underflow (status: 0x%X, task ID: 0x%X, task name: %s)\n\r", | |||
| error("CMSIS-RTOS error: Stack overflow (status: 0x%X, task ID: 0x%X, task name: %s)\n\r", | |||
There was a problem hiding this comment.
is the name from RTX wrong or just something related to a user clarity? If I read the code here - underflow case, the comment also states underflow, then overflow in error msg. This would make me think it's typo here
There was a problem hiding this comment.
Correct, all references are for an underflow error. As best that I can tell though, this error is invoked when stack space is exhausted. I think the reason it's referred to an underflow error is because the direction the stack grows is descending.
From a user's perspective, they know what an overflow error is, but the same can't be said for an underflow (which would be a different error anyways).
@neil-tan could likely give more info since this PR was initially created for him.
There was a problem hiding this comment.
@cmonr is correct from my knowledge, the underflow error in RTX is actually an overflow and the misnaming has been a pain for a while...Can we change the define to overflow or is that something out of our reach?
There was a problem hiding this comment.
+1, I've seen a lot of new users confused about this. They usually think the rtos itself is broken or they didn't configure the thread correctly. Once told it's a stack overflow, things click and they usually move to a completely different stage of debugging.
@kegilbert, no, changing the define breaks compatibility with RTX. See the conversation in #5966.
I'm just going to leave these here:
https://os.mbed.com/questions/79344/Thread-stack-underflow-i-have-no-clue/
http://www.keil.com/forum/22972/stack-underflow/
https://groups.google.com/forum/#!topic/nanopb/-pEAygzmgp4
There was a problem hiding this comment.
Figured as much. Maybe a comment saying as much would be nice to point out that the print statement isn't a typo would help future confusion at least a bit.
There was a problem hiding this comment.
👍 I know understand it better , thanks for the pointers!
|
Anyone opposed to seeing this move to the next stage of the PR, or is the consensus that it should get a comment akin to @kegilbert's suggestion? |
|
That comment there would be good (state it also in the commit msg the intention is there, even those pointers above ppl shared, feel free to reference there in the commit message ). |
92d9b7f to
f70ab45
Compare
End users are more familiar with handling overflow errors, even if underflow may be technically correct
9d0ea51 to
05dd446
Compare
|
/morph build |
Build : SUCCESSBuild number : 1344 Triggering tests/morph test |
|
Awesome, thanks for this! |
Exporter Build : SUCCESSBuild number : 1000 |
|
ಠ_ಠ /morph test |
Description
Modifies the underflow error in RTX to be reported as an overflow error, for user clarity.
Pull request type