On Mon, 18 Nov 2002, Oliver Neukum wrote:
> OK, here we go.
>
> case CPAD_CGID:
>
> - Race condition. You must drop the lock only after the last put_user
fixed.
>
> /* see if we are already in the middle of a write */
> if (dev->write_urb->status == -EINPROGRESS) {
> dbg (__FUNCTION__ " - already writing");
> goto exit;
> }
>
> - Bad. Wait for completion in that case, or, if you are opened non blocking,
> return the right error code
ok, need to look through more example code to work out wait for
completion, but have put in better error codes.
> for (i=bytes_written,wcount=2;i>1 && wcount<CPAD_BUFMAX;)
> ((char *) dev->write_urb->transfer_buffer)[wcount++] = tmpbuf[--i];
>
> - You might better copy into the buffer in first place instead of another copy
agreed, however at this step bytes from 2 onward are reversed so that the
user API looks like the LCD controller datasheet says and the bits/bytes
of the bitmap lies increment correctly as it goes out to the display
(don't know why these bytes are reversed, didn't build it :-)
> unsigned char mdata[8]; /* cpad mouse input buffer */
>
> - You must not do that. You violate DMA requirements on some architectures.
> Allocate the buffer sperately with kmalloc.
thank you, fixed.
cheers,
rob.
-------------------------------------------------------
This sf.net email is sponsored by: To learn the basics of securing
your web site with SSL, click here to get a FREE TRIAL of a Thawte
Server Certificate: http://www.gothawte.com/rd524.html
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
