Modifying sched_thread_local.c test to add corner cases.#3412
Modifying sched_thread_local.c test to add corner cases.#3412Biancaa-R wants to merge 1 commit intoapache:masterfrom
Conversation
a) Adding test to verify if all the created threads share addr. b) Checking if initial value is not corrupted. c) assert(false)->incase status is not okay. So faulty invalid object doesnt move forward. Signed-off-by: Biancaa Ramesh <biancaa2210329@ssn.edu.in>
Biancaa-R
changed the title
|
|
||
| /* In a broken system, all threads might point to the same static address */ | ||
|
|
||
| static void *g_first_thread_addr = NULL; |
There was a problem hiding this comment.
@Biancaa-R all common code (that needs to work on old microcontrollers) should follow the C89. Please read the https://nuttx.apache.org/docs/latest/contributing/coding_style.html for more info about the NuttX coding style
There was a problem hiding this comment.
Other detail: g_ is used when you are defining a global variable, that is not the case here since you are defining it inside a function
There was a problem hiding this comment.
@Biancaa-R all common code (that needs to work on old microcontrollers) should follow the C89. Please read the https://nuttx.apache.org/docs/latest/contributing/coding_style.html for more info about the NuttX coding style
Sure sir Ill read it and modify accordingly.
| /* In a broken system, all threads might point to the same static address */ | ||
|
|
||
| static void *g_first_thread_addr = NULL; | ||
| if (id == 0) |
There was a problem hiding this comment.
Did you try to print the "id" ? I think it never will be 0, if I'm not wrong (and I could be wrong) it starts from 1.
There was a problem hiding this comment.
I think it does start from 0 based on the attached log (it is printed a few times)
There was a problem hiding this comment.
I think it does start from 0 based on the attached log (it is printed a few times)
Yes.
linguini1
left a comment
linguini1
left a comment
There was a problem hiding this comment.
Hi @Biancaa-R , please read the contributing guide and modify your commit message to follow it
| printf("ERROR: Thread %d shares address with Thread 0\n", id); | ||
| ASSERT(false); | ||
|
|
||
| /* Checking 0 and other addresses */ |
| "status=%d\n", status); | ||
| ASSERT(false); | ||
|
|
||
| /* the threads will call pthread_barrier_wait on an invalid object */ |
| static void *g_first_thread_addr = NULL; | ||
| if (id == 0) | ||
| { | ||
| g_first_thread_addr = (void *)&g_tls_int; |
There was a problem hiding this comment.
What happens when the threads are released from the synchronized barrier? There is no guarantee that the first thread will set g_first_thread_addr before it is read by other threads. In fact, in your attached log, thread 2 starts first.
This check doesn't work.
3 participants
a) Adding test to verify if all the created threads share the same static address (an issue.)
b) Checking if initial value is not corrupted.
c) assert(false)->incase status is not okay.
So faulty invalid object doesnt move forward.
Note: Please adhere to Contributing Guidelines.
Summary
Impact
Makes the testing cover more corner cases , more robust.
Testing
For testing of these features I had used :
Modified test section:
CI test:
