On Sun, Mar 27, 2011 at 8:37 AM, Roman Mamedov <r...@romanrm.ru> wrote:
>
> However I have applied your patch [1] to my copy of 2.6.38.1 and unfortunately
> still can not unblock rfkill completely by software.
>
> After doing
>
> echo 1 > /sys/class/rfkill/rfkill0/state
>
> this value does not change, it just stays at 2.
>
> To reproduce: turn off Yeeloong, turn it back on, and without ever touching
> Fn+F5 try to unblock rfkill via the above "echo" or using the "rfkill"
> program, and establish a wifi connection.
>
> Maybe I am missing something else, or was that patch not supposed to fix this
> yet (it only mentions the LED)?
Just got the root cause, the default state of rkfill is 0 and the
default 'hard' state(determined by the related GPIO state) is also
blocked, so, nobody is allowed to change the soft state this time.
After pressing Fn+F5, the hard state is unblocked(GPIO state change),
the related EC interrupt is triggered and the KEY_WLAN is sent to
user-space by the YeeLoong platform driver, the user-space daemon can
be able to change the soft state through the /sys interface, then, the
soft state is also unblocked, at last WLAN works.
So, with the current rfkill driver, we can not change the rfkill state
without pressing Fn+F5.
Before in rtl8187b, we have changed the hard RF state when the soft
state is changed, that's why we can use the /sys interface, So, our
old implementation more or less have broken the hard & soft logic but
which really have brought us the convenience.
If we really want the current rfkill support of rtl8187 work as
before, we may need to change its whole design, so, no plan to touch
it ;-)
BTW, perhaps you can check if this patch work?
diff --git a/drivers/net/wireless/rtl818x/rtl8187/rfkill.c
b/drivers/net/wireless/rtl818x/rtl8187/rfkill.c
index 3411671..b2fe03e 100644
--- a/drivers/net/wireless/rtl818x/rtl8187/rfkill.c
+++ b/drivers/net/wireless/rtl818x/rtl8187/rfkill.c
@@ -21,7 +21,10 @@
#include "rfkill.h"
static bool rtl8187_is_radio_enabled(struct rtl8187_priv *priv)
-{
+{
+#ifdef CONFIG_LEMOTE_MACH2F
+ return 1;
+#else
u8 gpio;
gpio = rtl818x_ioread8(priv, &priv->map->GPIO0);
@@ -29,6 +32,7 @@ static bool rtl8187_is_radio_enabled(struct
rtl8187_priv *priv)
gpio = rtl818x_ioread8(priv, &priv->map->GPIO1);
return gpio & priv->rfkill_mask;
+#endif
}
void rtl8187_rfkill_init(struct ieee80211_hw *hw)
Or we may really need to turn on the 'hard' state by default in
rtl8187_init_hw().
Best Regards,
Wu Zhangjin
--
You received this message because you are subscribed to the Google Groups
"loongson-dev" group.
To post to this group, send email to loongson-dev@googlegroups.com.
To unsubscribe from this group, send email to
loongson-dev+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/loongson-dev?hl=en.
