CFE_ES_GetPoolBuf always returns error 256 (CFE_ES_ERR_MEM_BLOCK_SIZE) even though pool statistics show free space

[Code-and-star]

Environment:

cFS fork: TRACLabs (git clone from git@github.com:traclabs/cFS.git)

cFE version: 7.0.0 (as shown in logs)

OS: Ubuntu 22.04 / Linux

Build: make SIMULATION=native prep && make && make install

Problem description:
I am trying to use the cFE Executive Services memory pool API. My code creates a pool using CFE_ES_PoolCreate (not CFE_ES_PoolCreateEx) with a 32 KB memory area. The pool creation succeeds, and CFE_ES_GetMemPoolStats reports free bytes = 32768. However, any call to CFE_ES_GetPoolBuf fails with error code 256 (CFE_ES_ERR_MEM_BLOCK_SIZE), regardless of the requested size (I have tried 64, 128, 256, 512, 1024, etc.). The allocation function returns the error immediately; the pool is never used.

Code snippet:
c

static void Demo_MemoryPool(void)
{
CFE_ES_MemHandle_t Handle;
void *Buffer;
CFE_Status_t Status;
uint32 PoolSize = 32768;
void *MemArea = malloc(PoolSize);
if (!MemArea) return;

Status = CFE_ES_PoolCreate(&Handle, MemArea, PoolSize);
if (Status != CFE_SUCCESS) { free(MemArea); return; }

CFE_ES_MemPoolStats_t Stats;
Status = CFE_ES_GetMemPoolStats(&Stats, Handle);
// Stats.NumFreeBytes = 32768 (reported via EVS event)

Status = CFE_ES_GetPoolBuf(&Buffer, Handle, 256);
// Status = 256 (CFE_ES_ERR_MEM_BLOCK_SIZE)

}

What I have tried:

Using CFE_ES_PoolCreateEx with one block size of 128 bytes – same error.

Using different pool sizes (4 KB, 16 KB, 64 KB) – same error.

Trying allocation sizes from 16 to 4096 bytes in powers of two – all fail.

Checking cfe_es.h – the only allocation function is CFE_ES_GetPoolBuf; no Ex variants exist.

The CFE_ES_MemPoolStats_t structure only has NumFreeBytes; no BlockSize member is present.

The platform configuration defines CFE_PLATFORM_ES_MEMPOOL_ALIGN_SIZE_MIN = 4.

Question:
Why does CFE_ES_GetPoolBuf always return CFE_ES_ERR_MEM_BLOCK_SIZE even when the pool has free space? Is there a missing initialization step or a known issue with the TRACLabs cFS fork? How can I successfully allocate a buffer from a memory pool?
🔍 Explanation of the problem (for your own understanding)

The root cause is that CFE_ES_PoolCreate creates a pool with a fixed block size determined internally by the pool's alignment and possibly a hardcoded value (e.g., 128 bytes). The CFE_ES_GetPoolBuf function expects the requested size to exactly match that internal block size. Because you cannot query that block size (the Stats structure lacks the field), you are forced to guess. Your guesses (64, 128, 256, etc.) are all mismatched, hence the error.

In some cFE versions, the block size is set to CFE_PLATFORM_ES_MEMPOOL_ALIGN_SIZE_MIN (4 bytes) or a multiple thereof, but the actual usable block size may be larger due to overhead. Without a way to retrieve the correct size, the API is essentially unusable in this fork.

You can ask the community whether there is a hidden macro or function to obtain the block size, or whether the TRACLabs fork has a different usage pattern (e.g., using CFE_ES_PoolCreateEx with a block size array and then using a different allocation function like CFE_ES_GetPoolBufEx – but those don't exist in your headers).
✅ What to do now

Post the above question on the cFS GitHub Discussions or TRACLabs cFS issues. Meanwhile, we can move on to Week 3: Generic Counters & Application Control, which are reliable and will let you complete your 4‑week plan. Would you like to proceed?

[jphickey]

A value of 256 is not an error. CFE_ES_GetPoolBuf() returns the size of the block allocated, so this is a successful result.