Re: sys/dev/pckbc/pms.c diff

2013-07-01 Thread Martin Pieuchot
On 28/06/13(Fri) 13:08, joshua stein wrote:
 On Fri, 28 Jun 2013 at 12:14:43 -0400, David Hill wrote:
  in pms_proc_mouse, wsmouse_input is called with an uninitialized
  'dz' variable if the sc-protocol-type is not PMS_STANDARD or
  PMS_INTELLI.
 
 It should probably just return and not call wsmouse_input if the
 protocol type is unknown.

The protocol type cannot be unknown.  pms_proc_mouse() is only called
through sc-protocol-proc() and in this case type is always defined
to either PMS_STANDARD or PMS_INTELLI.

So this change only makes the code harder to understand as it suggests
there's more than two protocols supported by this function.


  Index: pms.c
  ===
  RCS file: /cvs/src/sys/dev/pckbc/pms.c,v
  retrieving revision 1.43
  diff -u -p -r1.43 pms.c
  --- pms.c   31 May 2013 19:21:09 -  1.43
  +++ pms.c   28 Jun 2013 16:13:02 -
  @@ -595,7 +595,7 @@ void
   pms_proc_mouse(struct pms_softc *sc)
   {
  u_int buttons;
  -   int  dx, dy, dz;
  +   int dx, dy, dz;
   
  buttons = butmap[sc-packet[0]  PMS_PS2_BUTTONSMASK];
  dx = (sc-packet[0]  PMS_PS2_XNEG) ?
  @@ -604,11 +604,12 @@ pms_proc_mouse(struct pms_softc *sc)
  (int)sc-packet[2] - 256 : sc-packet[2];
   
  switch (sc-protocol-type) {
  -   case PMS_STANDARD:
  -   dz = 0;
  -   break;
  case PMS_INTELLI:
  dz = (signed char)sc-packet[3];
  +   break;
  +   case PMS_STANDARD:
  +   default:
  +   dz = 0;
  break;
  }
   
  
 



viomb(4) assert with 1.35 of sys/arch/amd64/amd64/intr.c

2013-07-01 Thread Alexey E. Suslikov
Hi ports@.

See here http://article.gmane.org/gmane.os.openbsd.misc/205448

In addition, viomb_worker is doing splvm() and workq itself is

viomb_workq = workq_create(viomb, 1, IPL_VM);

Cheers,
Alexey