DPlib DLOG code change suggestion

Hi Folks,

I recently had an issue with caused by incorrect initialisation procedure when using the DPlib DLOG_4ch asm macro. Basically I had forgotten that the setting of the macros 'input' variables (size, trigger, etc) must come after the call to DPL_Init.

The only symptom of my error, was a confusing stack corruption and overflow. what had happened was that the 'DLOG_4ch_Size' input had been reset back to '0' by the DPL_Init with the ZeroNet in 'DLOG_4ch_INIT'. 

At first glance you might say that this would cause the DLOG macro to just not record any data, esentially doing nothing but triggering and counting to zero then going back to waiting for the trigger.

However, as the increment of Cnt comes before the check against the Size and the Size check only uses NEQ, what actually happens is that after the first trigger the macro starts logging data, and then never stops!! (or at least until Cnt rolls over I guess?)

The code in question (aside from my incorrect initialisation!) is on ln#s 145-146 of DLOG_4ch.asm and 108-109 of DLOG_1ch.asm (DPlib v3.3):

SUB	ACC, @_DLOG_4ch_Size	; ACC = Cnt - Size
SBF		DATALOG_EXIT, NEQ 	; If Cnt != Size banch to exit; otherwise reset Cnt

This works fine if Size is never 0 and if Cnt is never (through some code anomaly, interference, etc) greater than Size. If either of these two conditions are broken however, the macro will exceed the bounds of its output buffer and start writing data up through memory!

A suggestion may be to change a subsequent revision to use the BF instruction in place of the SBF, with the LT condition:

SUB	ACC, @_DLOG_4ch_Size	; ACC = Cnt - Size
BF		DATALOG_EXIT, LT	; If Size < Cnt branch to exit; otherwise reset Cnt

The BF instruction is of course, larger, but has the same cycle count and covers all cases for Cnt - Size.

EDIT: Could make the same change for the PreScalar and SkipCnt  on ln# 98 as similar problem exists there if PreScalar is 0.