imon vfd driver complexity question

[Jim Bryant]

ok, here's a dilemma...

the imon vfd has a bargraph mode (called equalizer mode using the 
soundgraph imon software in winblowz) has a certain level of wierdness 
in it.

i am leaning towards inclucing it's data massaging into the kernel 
driver, as it's use is a bit touchy.

it has off and sixteen vertical levels 1..f bottom to top.  each char 
column on both lines is used to form a single bar in the graph.

the code to do this in my test program is as follows (he question is 
should i or should i not include this in the kernel driver, as it is 
totally undocumented by anyone other than myself.  the manufacturer 
won't even answer questions or provide requested documentation. and 
given the recent supreme court ruling, i can reverse-engineer the 
interface to my heart's content legally.  i bought hardware, and want to 
use it):

/* im is a descriptor container structure pointer.

    ib is the input buffer of length 16 containing values of 0..16

    dt is a delay to prevent overrun, and may not be in the final code, 
this is a quick hack without error checking for non-deliverables
*/

int imonSendBargraph(struct imon *im, unsigned char *ib, int dt)
{
  int i, j, odd;
  unsigned char nh, nl, c, ob[16];
  unsigned char xt[9] = { 0x00, 0x01, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 
0x0e, 0x0f };

  bzero(ob, 16);

  j = 8;

  for(i = 0; i < 16; i++)
  {
    if ((c = ib[i]) > 16) return(-1);

    if (c > 8)
    {
      nh = c - 8;
      nl = 8;
    }

    else
    {
      nh = 0;
      nl = c;
    }

    if (i % 2)
    {
      ob[j] |= xt[nl];
      ob[(j++) - 8] |= xt[nh];
    }

    else
    {
      ob[j] |= (xt[nl] << 4);
      ob[j - 8] |= (xt[nh] << 4);
    }

    /*
    int odd;

    ob[j] |= (odd = (i % 2)) ? xt[nl] : (xt[nl] << 4);
    ob[j - 8] |= odd ? xt[nh] : (xt[nh] << 4);

    if (odd)
    {
      ++j;
      odd = 0;
    }
    */
  }

  imonSendCommand(im, ob[0x00], ob[0x01], ob[0x02], ob[0x03], ob[0x04], 
ob[0x05], 0xff, 0x40); delay(dt);
  imonSendCommand(im, ob[0x06], ob[0x07], ob[0x08], ob[0x09], ob[0x0a], 
ob[0x0b], 0xff, 0x42); delay(dt);
  imonSendCommand(im, ob[0x0c], ob[0x0d], ob[0x0e], ob[0x0f],     
0x00,     0xff, 0xff, 0x44); delay(dt);

  return(0);
}

after the driver is in place, i might add a plugin for xmms to take 
advantage of this.

given that all of the marketing photos of this (the antec veris elite) 
display this "equalizer" bargraph, i'm highly surprised that none of the 
linux people have bothered to reverse-engineer it and take advantage of it.

oh well, they can port it from freebsd when i'm done.  heh!

is this level of bit-fiddling acceptable in the kernel?  this is the 
most efficient version of doing this that i have come up with so far.  
the bit with the rval conditionals was just fucking around, and not 
intended for production code.

jim