bcm43xx: Hide if bcm43xx-mac80211 is builtin
From: Michael Wu <[email protected]>
This prevents users from compiling bcm43xx (softmac) and bcm43xx-mac80211
into the kernel at the same time. In the case that it is attempted, make
bcm43xx (softmac) not build.
Signed-off-by: Michael Wu <[email protected]>
---
drivers/net/wireless/bcm43xx/Kconfig | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/wireless/bcm43xx/Kconfig b/drivers/net/wireless/bcm43xx/Kconfig
index ce397e4..f69510d 100644
--- a/drivers/net/wireless/bcm43xx/Kconfig
+++ b/drivers/net/wireless/bcm43xx/Kconfig
@@ -1,6 +1,7 @@
config BCM43XX
tristate "Broadcom BCM43xx wireless support"
depends on PCI && IEEE80211 && IEEE80211_SOFTMAC && WLAN_80211 && EXPERIMENTAL
+ depends on BCM43XX_MAC80211 != 'y'
select WIRELESS_EXT
select FW_LOADER
select HW_RANDOM
On Wed, 2007-04-11 at 21:20 -0400, John W. Linville wrote:
> On Wed, Apr 11, 2007 at 07:01:06PM -0400, Michael Wu wrote:
> > bcm43xx: Hide if bcm43xx-mac80211 is builtin
> >
> > From: Michael Wu <[email protected]>
> >
> > This prevents users from compiling bcm43xx (softmac) and bcm43xx-mac80211
> > into the kernel at the same time. In the case that it is attempted, make
> > bcm43xx (softmac) not build.
>
> Since the two drivers use different firmware and the (currently)
> softmac-based driver will evolve into bcm4301 with support for older
> hardware, I'd prefer to leave this one out.
It's possible to bind and unbind drivers, so having two drivers for the
same hardware is not a problem per se.
The problem is that such configuration won't compile as it stands now.
It can be fixed by massive renaming of the symbols in one of the
drivers, but I'm not sure we settled on a specific name. I suggested
bcm4301, which was well received, but it sound like it won't happen
until bcm43xx_mac80211 matures.
If "bcm4301" prefix is too confusing or "humiliating" for the
softmac-based driver, we could consider something more acceptable in the
short term, like bcm4k, broadcom, bcom, brcm (Broadcom's stock ticker),
bcm43sm (after softmac), bcmwifi, bcmw, bcw (just to amuse Theo) and so
on.
If the symbols are not renamed, we need some protection in Kconfig, or
we'll fail "make allyesconfig".
--
Regards,
Pavel Roskin
Dan Williams wrote:
> On Wed, 2007-04-11 at 19:01 -0400, Michael Wu wrote:
>> bcm43xx: Hide if bcm43xx-mac80211 is builtin
>>
>> From: Michael Wu <[email protected]>
>>
>> This prevents users from compiling bcm43xx (softmac) and bcm43xx-mac80211
>> into the kernel at the same time. In the case that it is attempted, make
>> bcm43xx (softmac) not build.
>
> What kernel version is this targetted at? Have the mac80211 bits
> reached feature parity with the softmac bits? I'm still hearing quite a
> few reports of mac80211 being flaky WRT WEXT implementation and
> compatibility. I don't necessarily object, I just want to make sure
> that when this patch hits released kernels, that the mac80211 port is in
> the same or better shape than the softmac one.
This is aimed at Linville's wireless-dev and the -mm tree where both softmac and mac80211 versions
coexist.
At the moment, mac80211 is not in as good shape as softmac, and a date for its addition to mainline
has not been set.
Larry
On Wed, Apr 11, 2007 at 09:55:08PM -0400, Michael Wu wrote:
> On Wednesday 11 April 2007 21:20, John W. Linville wrote:
> > Since the two drivers use different firmware and the (currently)
> > softmac-based driver will evolve into bcm4301 with support for older
> > hardware, I'd prefer to leave this one out.
> >
> This is just to make allyesconfig work. (wasn't Andrew Morton telling you to
> do this?) Both drivers can still be compiled as modules.
OK, I see that now. FWIW, the namespace changes of the old bcm43xx
driver to bcm4301 would fulfill the same goal. But I suppose this
patch is fine for the time being.
John
--
John W. Linville
[email protected]
On Wed, 2007-04-11 at 19:01 -0400, Michael Wu wrote:
> bcm43xx: Hide if bcm43xx-mac80211 is builtin
>
> From: Michael Wu <[email protected]>
>
> This prevents users from compiling bcm43xx (softmac) and bcm43xx-mac80211
> into the kernel at the same time. In the case that it is attempted, make
> bcm43xx (softmac) not build.
What kernel version is this targetted at? Have the mac80211 bits
reached feature parity with the softmac bits? I'm still hearing quite a
few reports of mac80211 being flaky WRT WEXT implementation and
compatibility. I don't necessarily object, I just want to make sure
that when this patch hits released kernels, that the mac80211 port is in
the same or better shape than the softmac one.
Dan
> Signed-off-by: Michael Wu <[email protected]>
> ---
>
> drivers/net/wireless/bcm43xx/Kconfig | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/net/wireless/bcm43xx/Kconfig b/drivers/net/wireless/bcm43xx/Kconfig
> index ce397e4..f69510d 100644
> --- a/drivers/net/wireless/bcm43xx/Kconfig
> +++ b/drivers/net/wireless/bcm43xx/Kconfig
> @@ -1,6 +1,7 @@
> config BCM43XX
> tristate "Broadcom BCM43xx wireless support"
> depends on PCI && IEEE80211 && IEEE80211_SOFTMAC && WLAN_80211 && EXPERIMENTAL
> + depends on BCM43XX_MAC80211 != 'y'
> select WIRELESS_EXT
> select FW_LOADER
> select HW_RANDOM
On Wed, Apr 11, 2007 at 07:01:06PM -0400, Michael Wu wrote:
> bcm43xx: Hide if bcm43xx-mac80211 is builtin
>
> From: Michael Wu <[email protected]>
>
> This prevents users from compiling bcm43xx (softmac) and bcm43xx-mac80211
> into the kernel at the same time. In the case that it is attempted, make
> bcm43xx (softmac) not build.
Since the two drivers use different firmware and the (currently)
softmac-based driver will evolve into bcm4301 with support for older
hardware, I'd prefer to leave this one out.
Thanks,
John
--
John W. Linville
[email protected]
On Wednesday 11 April 2007 21:20, John W. Linville wrote:
> Since the two drivers use different firmware and the (currently)
> softmac-based driver will evolve into bcm4301 with support for older
> hardware, I'd prefer to leave this one out.
>
This is just to make allyesconfig work. (wasn't Andrew Morton telling you to
do this?) Both drivers can still be compiled as modules.
-Michael Wu