Re: [PATCH v4] UefiCpuPkg/PiSmmCpuDxeSmm: Avoid allocate Token every time

["Ni, Ray" <ray.ni@intel.com>]

Some comments.
> +  //
> 
> +  // Not free the buffer, just clear the UsedTokenNum. In order to
> 
> +  // avoid triggering allocate action when we need to use the token,
> 
> +  // do not free the buffer.
1.  Can you reword to below?
"Only free the token buffer recorded in the OldTOkenBufList upon exiting SMI.
Current token buffer stays allocated so next SMI doesn't need to re-allocate."

> +  ASSERT_EFI_ERROR (TokenCountPerChunk != 0);

2. This is not right. Should use ASSERT().

> +  DEBUG ((DEBUG_INFO, "CpuSmm: SpinLock Size = 0x%x,
> PreAllocateTokenNum = 0x%x\n", SpinLockSize, TokenCountPerChunk));

3. "PreAllocateTokenNum" -> "PcdTokenCountPerChunk". Please directly use the PCD name in debug log.

> 
> +  UINT8                           *CurrentTokenBuf;
> 
> +  UINTN                           UsedTokenNum;     // Only record tokens used in
> CurrentTokenBuf.

4. I see a mismatch between types of UsedTokenNum and the PCD. One is UINTN, the other is UINT32.
Better to use UINT32 here.

> gUefiCpuPkgTokenSpaceGuid.PcdTokenCountPerChunk|64|UINT32|0x30002002

5. The PCD name is too generic. It's hard to tell when/where the PCD is used by reading
the name only.
Maybe "PcdCpuSmmMpTokenCountPerChunk"?

> +#string
> STR_gUefiCpuPkgTokenSpaceGuid_PcdTokenCountPerChunk_PROMPT
> #language en-US "Specify the size of pre allocated token count per chunk.\n"

6. "Size of pre allocated token count per chunk" is a bit confusing.
How about "Count of pre allocated SMM MP tokens"?
Then maybe the PCD name can be "PcdCpuSmmMpPreAllocateTokenCount"?

Thanks,
Ray