GFilterRecord- inRowBytes always set to a multiple of 16 by photoshop

Question

Hi,
I am trying to write a series of filter plugins that expect a region of memory that is exactly (filterRect.right - filterRect.left) bytes per row of image data.
As you can see I'm attempting to set inRowBytes here, as well as some other shot-in-the-dark fields of the FilterRecord to try and solve this.
I've noticed that after a call to advanceState(), inRowBytes/outRowBytes is ALWAYS the next highest multiple of 16 if my filterRect width is not a multiple of 16.  For instance, if filterRect represents a 1025x769 region, inRowBytes = outRowBytes = 1040.  The problem is my_function is assuming that inData is exactly (filterRect.right - filterRect.left) bytes per row, so everything gets all skewed and messed up.  I tried saying I don't want padding, tried setting wantsAbsolute/wantsLayout, etc, but nothing has worked and I'm not sure what to do here?  Should I adjust the filterRect by some mod 16 amount??  Or is there some other way I can ignore those bytes?
for(int plane=0; plane<3; plane++) {
        gFilterRecord->outLoPlane = gFilterRecord->inLoPlane = plane;
        gFilterRecord->outHiPlane = gFilterRecord->inHiPlane = plane;
        if (gFilterRecord->haveMask) {
            SetMaskRect(filterRect);
        gFilterRecord->wantsAbsolute = true;
        gFilterRecord->inputPadding = gFilterRecord->outputPadding = plugInDoesNotWantPadding;
        gFilterRecord->inRowBytes = gFilterRecord->outRowBytes = filterRect.right-filterRect.left;
        SetInRect(filterRect);
        SetOutRect(filterRect);
        // update the gFilterRecord with our latest request
        *gResult = gFilterRecord->advanceState();
        if (*gResult != noErr) return;
        my_function(gFilterRecord->inData,
                             gFilterRecord->outData,
                             filterRect);     

Answer 1

Ah ha.. part of my confusion was I didn't know what was in those extra bytes but after some inspection with
a debugger I've decided they are garbage values so basically when my filter algorithms need to reference memory they need to use
inRowBytes as the mulitplier when creating an 'index' to a row and not filterRect.right - filterRect.left.  Sometimes easy things confuse me.