Re: [Madwifi-users] madwifi-users@lists.sourceforge.net

[Mark R. <mr...@ma...>]

it is kernel changes, and NOT madwifi changes, that are causing the
issue. an older version of the -uc* kernel would work with any of
the versions of Madwifi you've listed (and have, I'm sure :).

all the pieces needed to figure this out have been posted on the list,
and *should* be in the archives somewhere. but, just in case that part
of the archive is unreachable/lost, here's a patch that will/should
apply to any version of the WPA branch since Sam turned on REGOPS for
the xscale-be hal target.


-mark


diff -bBpruN madwifi/hal/linux/ah_osdep.c=20
madwifi-new/hal/linux/ah_osdep.c
--- madwifi/hal/linux/ah_osdep.c	2004-05-15 15:26:25.000000000 =
-0700
+++ madwifi-new/hal/linux/ah_osdep.c	2004-05-18 16:28:39.000000000 =
-0700
@@ -153,12 +153,14 @@ ath_hal_reg_write(struct ath_hal *ah, u_
  {
  #ifdef AH_DEBUG
  	if (ath_hal_debug > 1)
-		ath_hal_printf(ah, "WRITE 0x%x <=3D 0x%x\n", reg, val);
+		ath_hal_printf(ah, "WRITE 0x%x <=3D 0x%x\n", reg,
+			(reg >=3D 0x4000 && reg < =
0x5000)?val:__bswap32(val));
  #endif
- 	if (reg >=3D 0x4000 && reg < 0x5000)
- 		*((volatile u_int32_t *)(ah->ah_sh + reg)) =3D =
__bswap32(val);
- 	else
- 		*((volatile u_int32_t *)(ah->ah_sh + reg)) =3D val;
+	if (!(reg >=3D 0x4000 && reg < 0x5000)) {
+		writel(__bswap32(val), (unsigned long)(ah->ah_sh + =
reg));
+	} else {
+		writel(val, (unsigned long)(ah->ah_sh + reg));
+	}
  }

  u_int32_t __ahdecl
@@ -166,8 +169,9 @@ ath_hal_reg_read(struct ath_hal *ah, u_i
  {
   	u_int32_t val;

-	val =3D *((volatile u_int32_t *)(ah->ah_sh + reg));
- 	if (reg >=3D 0x4000 && reg < 0x5000)
+	val =3D readl((unsigned long)(ah->ah_sh + reg));
+
+	if (!(reg >=3D 0x4000 && reg < 0x5000))
  		val =3D __bswap32(val);
  #ifdef AH_DEBUG
  	if (ath_hal_debug > 1)



On Jun 28, 2004, at 8:25 AM, Anthony Tzouris wrote:

> I also using SnapGear Linux version 3.1.1 with the Intel XScale IXP425=20=

> processor with the Atheros 5212 mini-PCI A/B/G card 0168:0013.=A0 Is=20=

> there more information available in making this work?
>
>  I have tried CVS 06-23-2004, 04-23-2004, 05-15-2004 module checkouts=20=

> and all have the same problem described below when I 'insmod=20
> ath_pci.o'.=A0 I was working on building the 01-12-2004 release too =
but=20
> am in mid-process of setting up the environment variables since the=20
> older build has some different environment variable names.=A0 Going=20
> backward in time to make this work may not be the correct solution to=20=

> my issue.
>
>  After seeing this mailing list post below, it seems as if there=20
> already is some work being performed on this driver for the same=20
> processor target platform, distribution, and toolchain.=A0 There =
appears=20
> to be some offline interchange going on.=A0 Any information would be=20=

> greatly appreciated.=A0 Thanks.
>
>  Anthony Tzouris

... <snip> ...=