On Sun, Mar 15, 2009 at 8:00 PM, Bob Copeland <[email protected]> wrote: > Hi all, > > A user is experiencing a MIB storm with bug 12647: > > http://bugzilla.kernel.org/show_bug.cgi?id=12647 > > But he's seeing numbers like this: > > [ 813.480007] ATH5K: AR5K_FCS_FAIL = 49152 > [ 813.487655] ATH5K: AR5K_BEACON_CNT = 7703 > [ 813.491712] ATH5K: AR5K_PROFCNT_TX = 28375 > [ 813.495712] ATH5K: AR5K_PROFCNT_RX = 1407146494 > [ 813.497653] ATH5K: AR5K_PROFCNT_RXCLR = 1434465504 > [ 813.501653] ATH5K: AR5K_PROFCNT_CYCLE = 2909062697 > [ 813.506456] ATH5K: AR5K_PROFCNT_RX = 1221387 > [ 813.511296] ATH5K: AR5K_PROFCNT_RXCLR = 1129347 > [ 813.515296] ATH5K: AR5K_PROFCNT_CYCLE = 1164599 > > [ 998.600008] ATH5K: AR5K_FCS_FAIL = 49152 > [ 998.606662] ATH5K: AR5K_BEACON_CNT = 1779 > [ 998.610719] ATH5K: AR5K_PROFCNT_TX = 53946 > [ 998.615037] ATH5K: AR5K_PROFCNT_RX = 1707360992 > [ 998.619037] ATH5K: AR5K_PROFCNT_RXCLR = 1749437580 > [ 998.623058] ATH5K: AR5K_PROFCNT_CYCLE = 3853124565 > [ 998.627058] ATH5K: AR5K_PROFCNT_RX = 302530 > [ 998.630041] ATH5K: AR5K_PROFCNT_RXCLR = 272468 > [ 998.635404] ATH5K: AR5K_PROFCNT_CYCLE = 1141769 > > It's curious that FCS_FAIL is always 49152, which seems to indicate either > that the register is not read-and-clear on all parts, or else the threshold > is 49152 which also doesn't make any sense. > > Any thoughts on this? I think we could add this patch to be on the safe > side, but it would be nice to know what the deal is. He says the patch > fixes it for him. > > (The patch is whitespace damaged but you get the idea). > > diff --git a/drivers/net/wireless/ath5k/pcu.c > b/drivers/net/wireless/ath5k/pcu.c > index f8a4a69..16bd1d9 100644 > --- a/drivers/net/wireless/ath5k/pcu.c > +++ b/drivers/net/wireless/ath5k/pcu.c > @@ -143,6 +143,9 @@ void ath5k_hw_update_mib_counters(struct ath5k_hw *ah, > } > > /* TODO: Handle ANI stats */ > + > + /* clear all counters */ > + ath5k_hw_reg_write(ah, AR5K_MIBC_CMC, AR5K_MIBC); > }
I'm not sure of the details of the counters being the same but I'd recommend to disable MIB interrupt until ANI is supported. Luis _______________________________________________ ath5k-devel mailing list [email protected] https://lists.ath5k.org/mailman/listinfo/ath5k-devel
