Fix #1279, rtems allow variable length queue msgs.

[thesamprice]

Checklist (Please check before submitting)

• I reviewed the Contributing Guide.
• I signed and emailed the appropriate Contributor License Agreement to [email protected] and copied [email protected].
  NA as Gov employee? NA as removing code not adding?

Describe the contribution
A clear and concise description of what the contribution is.
Drops check on size != rtems_size to allow variable length queue messages.
Matches the posix queue implementation.

Testing performed
Steps taken to test the contribution:

1. Build steps '...'
   Built and ran a non cfs system using posix.
   built and ran the same cfs system using rtems, this fix allowed system to run.

2. Execution steps '...'
   Ran on RTEMS on microblaze.

Expected behavior changes
A clear and concise description of how this contribution will change behavior and level of impact.

• API Change: RTEMS queues changed to match posix queues for variable length messages

System(s) tested on
RTEMS / Microblaze

Additional context
Add any other context about the contribution here.

Third party code
N/A

Contributor Info - All information REQUIRED for consideration of pull request
NASA GSFC

[thesamprice]

Im not sure how to fix the commit message.
On my branch its correct. See link
Somehow the pull request is changing the commit message.

[skliper]

@thesamprice - it was the pull request title (lowercase f on fix) that was causing the job to fail. We haven't figured out yet how to get the job to pick up the corrected title to be able to rerun and have it pass.

[skliper]

Maybe make the size verification optional? Just skip the check if passed in size is zero.

[thesamprice]

@skliper
Should this check be moved up into the core code rather than in just rtems?

[skliper]

@thesamprice Copy, I should have looked at the shared code:

osal/src/os/shared/src/osapi-queue.c

Lines 174 to 181 in 6e6afb4

	if (size < queue->max_size)
	{
	/*
	** The buffer that the user is passing in is potentially too small
	*/
	*size_copied = 0;
	return_code = OS_QUEUE_INVALID_SIZE;
	}

Looks like both POSIX and VxWorks implementations allow variable size, so I'm curious why the extra check is in RTEMS. Just concerned someone may be depending on this check... although if it's only in a subset of the implementations that's a pretty fragile dependency.
VxWorks:

osal/src/os/vxworks/src/os-impl-queues.c

Lines 155 to 178 in 6e6afb4

	status = msgQReceive(impl->vxid, data, size, ticks);
	if (status == ERROR)
	{
	*size_copied = 0;
	if (errno == S_objLib_OBJ_TIMEOUT)
	{
	return_code = OS_QUEUE_TIMEOUT;
	}
	else if (errno == S_objLib_OBJ_UNAVAILABLE)
	{
	return_code = OS_QUEUE_EMPTY;
	}
	else
	{
	OS_DEBUG("msgQReceive() - vxWorks errno %d\n", errno);
	return_code = OS_ERROR;
	}
	}
	else
	{
	*size_copied = OSAL_SIZE_C(status);
	return_code = OS_SUCCESS;
	}

Linux:

osal/src/os/posix/src/os-impl-queues.c

Lines 249 to 280 in 6e6afb4

	/* Figure out the return code */
	if (sizeCopied == -1)
	{
	*size_copied = OSAL_SIZE_C(0);
	/* Map the system errno to the most appropriate OSAL return code */
	if (errno == EMSGSIZE)
	{
	return_code = OS_QUEUE_INVALID_SIZE;
	}
	else if (timeout == OS_PEND || errno != ETIMEDOUT)
	{
	/* OS_PEND was supposed to pend forever until a message arrived
	* so something else is wrong. Otherwise, at this point the only
	* "acceptable" errno is TIMEDOUT for the other cases.
	*/
	return_code = OS_ERROR;
	}
	else if (timeout == OS_CHECK)
	{
	return_code = OS_QUEUE_EMPTY;
	}
	else
	{
	return_code = OS_QUEUE_TIMEOUT;
	}
	}
	else
	{
	*size_copied = OSAL_SIZE_C(sizeCopied);
	return_code = OS_SUCCESS;
	}

[skliper]

Maybe instead of just deleting the check (and checking for success twice), it could be updated to match the pattern in POSX/VxWorks. Unless someone comes back with a reason why the check could be required for RTEMS I don't have any issues with making it consistent.

Recalling change request based on further conversation

[thesamprice]

If someone is relying on this check in rtems, their code will break when going to vxworks / posix.

I would recommend doing the following only if the check is needed.

• moving the check up into the shared area.
• and adding some sort of settings flag when its initialized
  OS_FIXED_SIZE_QUEUE
• only doing the check if OS_FIXED_SIZE_QUEUE is set.

This would keep the existing functionality common across all platforms.

The functionality should be documented also.

I use the queues to route messages to udp/tcp so this rtems check is breaking for me only on rtems.

Ill go ahead and fork osal, and make the change so our system can work on rtems.

[dzbaker]

CCB 25 Aug 2022: @dzbaker will pick up from @skliper

[thesamprice]

Closing pull request, this will be replaced by the posix mimic version.