* [PATCH] ar9170: implement frequency calibration for one-stage/openfw
@ 2009-09-03 18:25 Christian Lamparter
2009-09-19 0:02 ` Andrew Morton
0 siblings, 1 reply; 6+ messages in thread
From: Christian Lamparter @ 2009-09-03 18:25 UTC (permalink / raw)
To: linux-wireless; +Cc: John Linville
This patch ports some code from the vendor driver, which is
supposed to upload the right calibration values for the
chosen frequency.
In theory, this should give a better range and throughput
for all users with the open, or one-stage firmware.
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
---
diff --git a/drivers/net/wireless/ath/ar9170/phy.c b/drivers/net/wireless/ath/ar9170/phy.c
index df86f70..be96f50 100644
--- a/drivers/net/wireless/ath/ar9170/phy.c
+++ b/drivers/net/wireless/ath/ar9170/phy.c
@@ -987,6 +987,124 @@ static u8 ar9170_interpolate_u8(u8 x, u8 x1, u8 y1, u8 x2, u8 y2)
#undef SHIFT
}
+static u8 ar9170_interpolate_val(u8 x, u8 *x_array, u8 *y_array)
+{
+ int i;
+
+ for (i = 0; i < 3; i++)
+ if (x <= x_array[i + 1])
+ break;
+
+ return ar9170_interpolate_u8(x,
+ x_array[i],
+ y_array[i],
+ x_array[i + 1],
+ y_array[i + 1]);
+}
+
+static int ar9170_set_freq_cal_data(struct ar9170 *ar,
+ struct ieee80211_channel *channel)
+{
+ u8 *cal_freq_pier;
+ u8 vpds[2][AR5416_PD_GAIN_ICEPTS];
+ u8 pwrs[2][AR5416_PD_GAIN_ICEPTS];
+ int chain, idx, i;
+ u8 f;
+
+ switch (channel->band) {
+ case IEEE80211_BAND_2GHZ:
+ f = channel->center_freq - 2300;
+ cal_freq_pier = ar->eeprom.cal_freq_pier_2G;
+ i = AR5416_NUM_2G_CAL_PIERS - 1;
+ break;
+
+ case IEEE80211_BAND_5GHZ:
+ f = (channel->center_freq - 4800) / 5;
+ cal_freq_pier = ar->eeprom.cal_freq_pier_5G;
+ i = AR5416_NUM_5G_CAL_PIERS - 1;
+ break;
+
+ default:
+ return -EINVAL;
+ break;
+ }
+
+ for (; i >= 0; i--) {
+ if (cal_freq_pier[i] != 0xff)
+ break;
+ }
+ if (i < 0)
+ return -EINVAL;
+
+ idx = ar9170_find_freq_idx(i, cal_freq_pier, f);
+
+ ar9170_regwrite_begin(ar);
+
+ for (chain = 0; chain < AR5416_MAX_CHAINS; chain++) {
+ for (i = 0; i < AR5416_PD_GAIN_ICEPTS; i++) {
+ struct ar9170_calibration_data_per_freq *cal_pier_data;
+ int j;
+
+ switch (channel->band) {
+ case IEEE80211_BAND_2GHZ:
+ cal_pier_data = &ar->eeprom.
+ cal_pier_data_2G[chain][idx];
+ break;
+
+ case IEEE80211_BAND_5GHZ:
+ cal_pier_data = &ar->eeprom.
+ cal_pier_data_5G[chain][idx];
+ break;
+
+ default:
+ return -EINVAL;
+ }
+
+ for (j = 0; j < 2; j++) {
+ vpds[j][i] = ar9170_interpolate_u8(f,
+ cal_freq_pier[idx],
+ cal_pier_data->vpd_pdg[j][i],
+ cal_freq_pier[idx + 1],
+ cal_pier_data[1].vpd_pdg[j][i]);
+
+ pwrs[j][i] = ar9170_interpolate_u8(f,
+ cal_freq_pier[idx],
+ cal_pier_data->pwr_pdg[j][i],
+ cal_freq_pier[idx + 1],
+ cal_pier_data[1].pwr_pdg[j][i]) / 2;
+ }
+ }
+
+ for (i = 0; i < 76; i++) {
+ u32 phy_data;
+ u8 tmp;
+
+ if (i < 25) {
+ tmp = ar9170_interpolate_val(i, &pwrs[0][0],
+ &vpds[0][0]);
+ } else {
+ tmp = ar9170_interpolate_val(i - 12,
+ &pwrs[1][0],
+ &vpds[1][0]);
+ }
+
+ phy_data |= tmp << ((i & 3) << 3);
+ if ((i & 3) == 3) {
+ ar9170_regwrite(0x1c6280 + chain * 0x1000 +
+ (i & ~3), phy_data);
+ phy_data = 0;
+ }
+ }
+
+ for (i = 19; i < 32; i++)
+ ar9170_regwrite(0x1c6280 + chain * 0x1000 + (i << 2),
+ 0x0);
+ }
+
+ ar9170_regwrite_finish();
+ return ar9170_regwrite_result();
+}
+
static int ar9170_set_power_cal(struct ar9170 *ar, u32 freq, enum ar9170_bw bw)
{
struct ar9170_calibration_target_power_legacy *ctpl;
@@ -1207,6 +1325,10 @@ int ar9170_set_channel(struct ar9170 *ar, struct ieee80211_channel *channel,
if (err)
return err;
+ err = ar9170_set_freq_cal_data(ar, channel);
+ if (err)
+ return err;
+
err = ar9170_set_power_cal(ar, channel->center_freq, bw);
if (err)
return err;
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH] ar9170: implement frequency calibration for one-stage/openfw
2009-09-03 18:25 [PATCH] ar9170: implement frequency calibration for one-stage/openfw Christian Lamparter
@ 2009-09-19 0:02 ` Andrew Morton
2009-09-19 12:21 ` Joerg Albert
0 siblings, 1 reply; 6+ messages in thread
From: Andrew Morton @ 2009-09-19 0:02 UTC (permalink / raw)
To: Christian Lamparter; +Cc: linux-wireless, linville, David S. Miller
On Thu, 3 Sep 2009 20:25:31 +0200
Christian Lamparter <chunkeey@googlemail.com> wrote:
> This patch ports some code from the vendor driver, which is
> supposed to upload the right calibration values for the
> chosen frequency.
>
> In theory, this should give a better range and throughput
> for all users with the open, or one-stage firmware.
>
> ...
>
> + for (i = 0; i < 76; i++) {
> + u32 phy_data;
> + u8 tmp;
> +
> + if (i < 25) {
> + tmp = ar9170_interpolate_val(i, &pwrs[0][0],
> + &vpds[0][0]);
> + } else {
> + tmp = ar9170_interpolate_val(i - 12,
> + &pwrs[1][0],
> + &vpds[1][0]);
> + }
> +
> + phy_data |= tmp << ((i & 3) << 3);
Clearly buggy and the compiler warns. The value of phy_data is unknown
here.
How did this get all the way into mainline?
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] ar9170: implement frequency calibration for one-stage/openfw
2009-09-19 0:02 ` Andrew Morton
@ 2009-09-19 12:21 ` Joerg Albert
2009-09-28 18:41 ` John W. Linville
0 siblings, 1 reply; 6+ messages in thread
From: Joerg Albert @ 2009-09-19 12:21 UTC (permalink / raw)
To: Andrew Morton
Cc: Christian Lamparter, linux-wireless, linville, David S. Miller
On 09/19/2009 02:02 AM, Andrew Morton wrote:
> On Thu, 3 Sep 2009 20:25:31 +0200
> Christian Lamparter <chunkeey@googlemail.com> wrote:
>
>> This patch ports some code from the vendor driver, which is
>> supposed to upload the right calibration values for the
>> chosen frequency.
>>
>> In theory, this should give a better range and throughput
>> for all users with the open, or one-stage firmware.
>>
>> ...
>>
>> + for (i = 0; i < 76; i++) {
>> + u32 phy_data;
>> + u8 tmp;
>> +
>> + if (i < 25) {
>> + tmp = ar9170_interpolate_val(i, &pwrs[0][0],
>> + &vpds[0][0]);
>> + } else {
>> + tmp = ar9170_interpolate_val(i - 12,
>> + &pwrs[1][0],
>> + &vpds[1][0]);
>> + }
>> +
>> + phy_data |= tmp << ((i & 3) << 3);
>
> Clearly buggy and the compiler warns. The value of phy_data is unknown
> here.
>
> How did this get all the way into mainline?
Strangely it compiles without any warning for me with the latest linux-wireless:
joerg@nc10:~/src/wireless.gits/wireless-testing$ gcc --version
gcc (Ubuntu 4.3.3-5ubuntu4) 4.3.3
joerg@nc10:~/src/wireless.gits/wireless-testing$ git describe
v2.6.31-38258-g389ce02
joerg@nc10:~/src/wireless.gits/wireless-testing/drivers/net/wireless/ath/ar9170$ grep phy_data phy.c
u32 phy_data;
phy_data |= tmp << ((i & 3) << 3);
(i & ~3), phy_data);
phy_data = 0;
joerg@nc10:~/src/wireless.gits/wireless-testing$ make modules
...
CC [M] drivers/net/wireless/ath/ar9170/mac.o
CC [M] drivers/net/wireless/ath/ar9170/phy.o
CC [M] drivers/net/wireless/ath/ar9170/led.o
make V=1 says:
gcc -Wp,-MD,drivers/net/wireless/ath/ar9170/.phy.o.d -nostdinc -isystem
/usr/lib/gcc/i486-linux-gnu/4.3.3/include -Iinclude
-I/home/joerg/src/wireless.gits/wireless-testing/arch/x86/include -include
include/linux/autoconf.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes
-Wno-trigraphs -fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Wno-format-security
-fno-delete-null-pointer-checks -O2 -m32 -msoft-float -mregparm=3
-freg-struct-return -mpreferred-stack-boundary=2 -march=i686 -mtune=pentium3
-mtune=generic -Wa,-mtune=generic32 -ffreestanding -pipe -Wno-sign-compare
-fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow
-fno-stack-protector -fno-omit-frame-pointer -fno-optimize-sibling-calls -g
-Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow -DMODULE
-D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(phy)"
-D"KBUILD_MODNAME=KBUILD_STR(ar9170usb)" -c -o
drivers/net/wireless/ath/ar9170/.tmp_phy.o drivers/net/wireless/ath/ar9170/phy.c
Adding "-Wuninitialized" didn't help, acc. to the doc it is switched on
with -Wall for -O2.
Is this a regression of gcc?
Regards,
Jörg.
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] ar9170: implement frequency calibration for one-stage/openfw
2009-09-19 12:21 ` Joerg Albert
@ 2009-09-28 18:41 ` John W. Linville
2009-09-28 19:45 ` Christian Lamparter
0 siblings, 1 reply; 6+ messages in thread
From: John W. Linville @ 2009-09-28 18:41 UTC (permalink / raw)
To: Joerg Albert
Cc: Andrew Morton, Christian Lamparter, linux-wireless,
David S. Miller
On Sat, Sep 19, 2009 at 02:21:36PM +0200, Joerg Albert wrote:
> On 09/19/2009 02:02 AM, Andrew Morton wrote:
> > On Thu, 3 Sep 2009 20:25:31 +0200
> > Christian Lamparter <chunkeey@googlemail.com> wrote:
> >
> >> This patch ports some code from the vendor driver, which is
> >> supposed to upload the right calibration values for the
> >> chosen frequency.
> >>
> >> In theory, this should give a better range and throughput
> >> for all users with the open, or one-stage firmware.
> >>
> >> ...
> >>
> >> + for (i = 0; i < 76; i++) {
> >> + u32 phy_data;
> >> + u8 tmp;
> >> +
> >> + if (i < 25) {
> >> + tmp = ar9170_interpolate_val(i, &pwrs[0][0],
> >> + &vpds[0][0]);
> >> + } else {
> >> + tmp = ar9170_interpolate_val(i - 12,
> >> + &pwrs[1][0],
> >> + &vpds[1][0]);
> >> + }
> >> +
> >> + phy_data |= tmp << ((i & 3) << 3);
> >
> > Clearly buggy and the compiler warns. The value of phy_data is unknown
> > here.
> >
> > How did this get all the way into mainline?
>
> Strangely it compiles without any warning for me with the latest linux-wireless:
That is strange -- no warning here (F-11) either...
Christian, care to propose a patch? Or maybe just something like this?
diff --git a/drivers/net/wireless/ath/ar9170/phy.c b/drivers/net/wireless/ath/ar9170/phy.c
index b3e5cf3..49c10cb 100644
--- a/drivers/net/wireless/ath/ar9170/phy.c
+++ b/drivers/net/wireless/ath/ar9170/phy.c
@@ -1220,7 +1220,7 @@ static int ar9170_set_freq_cal_data(struct ar9170 *ar,
&vpds[1][0]);
}
- phy_data |= tmp << ((i & 3) << 3);
+ phy_data = tmp << ((i & 3) << 3);
if ((i & 3) == 3) {
ar9170_regwrite(0x1c6280 + chain * 0x1000 +
(i & ~3), phy_data);
John
--
John W. Linville Someday the world will need a hero, and you
linville@tuxdriver.com might be all we have. Be ready.
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH] ar9170: implement frequency calibration for one-stage/openfw
2009-09-28 18:41 ` John W. Linville
@ 2009-09-28 19:45 ` Christian Lamparter
2009-09-28 19:50 ` John W. Linville
0 siblings, 1 reply; 6+ messages in thread
From: Christian Lamparter @ 2009-09-28 19:45 UTC (permalink / raw)
To: John W. Linville
Cc: Joerg Albert, Andrew Morton, linux-wireless, David S. Miller
2009/9/28 John W. Linville <linville@tuxdriver.com>:
> On Sat, Sep 19, 2009 at 02:21:36PM +0200, Joerg Albert wrote:
>> On 09/19/2009 02:02 AM, Andrew Morton wrote:
>> > On Thu, 3 Sep 2009 20:25:31 +0200
>> > Christian Lamparter <chunkeey@googlemail.com> wrote:
>> >
>> >> This patch ports some code from the vendor driver, which is
>> >> supposed to upload the right calibration values for the
>> >> chosen frequency.
>> >>
>> >> In theory, this should give a better range and throughput
>> >> for all users with the open, or one-stage firmware.
>> >>
>> >> ...
>> >>
>> >> + for (i = 0; i < 76; i++) {
>> >> + u32 phy_data;
>> >> + u8 tmp;
>> >> +
>> >> + if (i < 25) {
>> >> + tmp = ar9170_interpolate_val(i, &pwrs[0][0],
>> >> + &vpds[0][0]);
>> >> + } else {
>> >> + tmp = ar9170_interpolate_val(i - 12,
>> >> + &pwrs[1][0],
>> >> + &vpds[1][0]);
>> >> + }
>> >> +
>> >> + phy_data |= tmp << ((i & 3) << 3);
>> >
>> > Clearly buggy and the compiler warns. The value of phy_data is unknown
>> > here.
>> >
>> > How did this get all the way into mainline?
>>
>> Strangely it compiles without any warning for me with the latest linux-wireless:
>
> That is strange -- no warning here (F-11) either...
>
> Christian, care to propose a patch? Or maybe just something like this?
>
> diff --git a/drivers/net/wireless/ath/ar9170/phy.c b/drivers/net/wireless/ath/ar9170/phy.c
> index b3e5cf3..49c10cb 100644
> --- a/drivers/net/wireless/ath/ar9170/phy.c
> +++ b/drivers/net/wireless/ath/ar9170/phy.c
> @@ -1220,7 +1220,7 @@ static int ar9170_set_freq_cal_data(struct ar9170 *ar,
> &vpds[1][0]);
> }
>
> - phy_data |= tmp << ((i & 3) << 3);
> + phy_data = tmp << ((i & 3) << 3);
> if ((i & 3) == 3) {
> ar9170_regwrite(0x1c6280 + chain * 0x1000 +
> (i & ~3), phy_data);
>
> John
no? Andrew Morton's mail included a tiny patch which initializes the
phy_data to 0. I assumed you would pick it up right away...
Or is there something which would prevent it from including?
Regards,
Chr
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] ar9170: implement frequency calibration for one-stage/openfw
2009-09-28 19:45 ` Christian Lamparter
@ 2009-09-28 19:50 ` John W. Linville
0 siblings, 0 replies; 6+ messages in thread
From: John W. Linville @ 2009-09-28 19:50 UTC (permalink / raw)
To: Christian Lamparter
Cc: Joerg Albert, Andrew Morton, linux-wireless, David S. Miller
On Mon, Sep 28, 2009 at 09:45:28PM +0200, Christian Lamparter wrote:
> 2009/9/28 John W. Linville <linville@tuxdriver.com>:
> > On Sat, Sep 19, 2009 at 02:21:36PM +0200, Joerg Albert wrote:
> >> On 09/19/2009 02:02 AM, Andrew Morton wrote:
> >> > How did this get all the way into mainline?
> >>
> >> Strangely it compiles without any warning for me with the latest linux-wireless:
> >
> > That is strange -- no warning here (F-11) either...
> >
> > Christian, care to propose a patch? Or maybe just something like this?
> >
> > diff --git a/drivers/net/wireless/ath/ar9170/phy.c b/drivers/net/wireless/ath/ar9170/phy.c
> > index b3e5cf3..49c10cb 100644
> > --- a/drivers/net/wireless/ath/ar9170/phy.c
> > +++ b/drivers/net/wireless/ath/ar9170/phy.c
> > @@ -1220,7 +1220,7 @@ static int ar9170_set_freq_cal_data(struct ar9170 *ar,
> > &vpds[1][0]);
> > }
> >
> > - phy_data |= tmp << ((i & 3) << 3);
> > + phy_data = tmp << ((i & 3) << 3);
> > if ((i & 3) == 3) {
> > ar9170_regwrite(0x1c6280 + chain * 0x1000 +
> > (i & ~3), phy_data);
> >
> > John
> no? Andrew Morton's mail included a tiny patch which initializes the
> phy_data to 0. I assumed you would pick it up right away...
> Or is there something which would prevent it from including?
Don't know what email you got, but there was no patch in my inbox...
So, I presume you have no objection to the patch above? Initializing
phy_data just to unconditionally assign to it doesn't seem any better.
John
--
John W. Linville Someday the world will need a hero, and you
linville@tuxdriver.com might be all we have. Be ready.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2009-09-28 20:00 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-09-03 18:25 [PATCH] ar9170: implement frequency calibration for one-stage/openfw Christian Lamparter
2009-09-19 0:02 ` Andrew Morton
2009-09-19 12:21 ` Joerg Albert
2009-09-28 18:41 ` John W. Linville
2009-09-28 19:45 ` Christian Lamparter
2009-09-28 19:50 ` John W. Linville
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).