I committed this change to CVS.
Now the serialmouse driver works on my desktop machine, cool :)
I had tried the original serialmouse driver on my desktop and couldn't
get it to work, but didn't have time to debug it.
Thanks!
On 2002.01.08 07:06 Arnd Bergmann wrote:
> I've had trouble with the serial mouse driver on a k6/400/linux-2.4.16
> box, the patch below fixes this.
> The problem is that the second read() would always fail silently
> with EAGAIN when there is no new data on /dev/ttyS0 yet.
> Instead of waiting for the packet to complete, I chose to return to
> the select loop, like the other mouse drivers do.
> Tested on one machine with both MS and logitech compatible
> mouses.
>
> Arnd <><
>
> --- input/serialmouse.c.orig Tue Jan 8 14:30:41 2002
> +++ input/serialmouse.c Tue Jan 8 14:50:46 2002
> @@ -89,22 +89,26 @@
> int serialmouse_fd_activate(int fd) {
> u8 buttons;
> s8 dx,dy;
> - u8 packet[3];
> + static u8 packet[3];
> + static int pos;
> s16 cursorx,cursory;
>
> if (fd != mouse_fd)
> return 0;
>
> /* Read a correctly-aligned mouse packet. If the first byte isn't
> 0x40,
> - * it isn't correctly aligned. The mouse packet is 4 bytes long.
> + * it isn't correctly aligned. The mouse packet is 3 or 4 bytes
> long.
> + * On fast machines, we can't read a whole packet at once, so we
> have
> + * to maintain the state in a static variable.
> */
>
> - if (!read(mouse_fd,packet,1))
> + if (read(mouse_fd,packet+pos,1) != 1)
> return 1;
> if (!(packet[0] & 0x40))
> return 1;
> - if (!read(mouse_fd,packet+1,2))
> + if (pos++ < 2)
> return 1;
> + pos = 0;
>
> /* Get the cursor position in physical coordinates */
> cursorx = cursor->x;
>
>
> _______________________________________________
> Pgui-devel mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/pgui-devel
>
_______________________________________________
Pgui-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/pgui-devel
