On Sat, May 31, 2008 at 8:03 AM, Krzysztof Helt [off-list ref] wrote:

quoted

@@ -569,13 +579,16 @@ static int __devinit metronomefb_probe(struct platform_device *dev)
 {

(...)

quoted

+     fw = epd_frame_table[epd_dt_index].fw;
+     fh = epd_frame_table[epd_dt_index].fh;
+     videomemorysize = fw * fh;
      videomemory = vmalloc(videomemorysize);
      if (!videomemory)
              return -ENOMEM;

I suppose the put_module should be done before return (see the whole
function code).

Ah, yes. Ok, will correct that.

In order to reduce number of goto labels you can do something
like this (for allocated pointers):

err_exit:
   if (pointer)
       free(pointer)
  do_something_more

instead

err_exit1:
 free(pointer)
err_exit2:
  do_something_more

Ok. I'll take a look at doing it that way.

quoted

 static int __devexit metronomefb_remove(struct platform_device *dev)

@@ -734,14 +764,13 @@ static int __devexit metronomefb_remove(struct platform_device *dev)
      if (info) {
              struct metronomefb_par *par = info->par;
              fb_deferred_io_cleanup(info);
-             dma_free_writecombine(&dev->dev, par->metromemsize,
-                                     par->metromem, par->metromem_dma);
              fb_dealloc_cmap(&info->cmap);
              vfree(par->csum_table);
              unregister_framebuffer(info);

I am not sure but to release the color map before
unregistering the framebuffer is probably a mistake.

Agreed. I'll correct that.

Thanks,
jaya

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/