Skip to contentSkip to conversation

fix tunning for r820t tunner

PATCH In review Jiří Pinkava 3 participants 8 messages Sep 29, 2013 – Sep 30, 2013 Media (V4L/DVB)

lore
Share & tools
lore
#0 lore raw

cover fix tunning for r820t tunner

Fixes range check for VCO parameters, simplifies calculation of divisor.

 drivers/media/tuners/r820t.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)
#1 lore raw

PATCH 1/2 r820t: fix nint range check

Use full range of VCO parameters, fixes tunning for some frequencies.

---
 drivers/media/tuners/r820t.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/media/tuners/r820t.c +0 -0
index 1c23666..e25c720 100644
@@ -637,7 +637,7 @@ static int r820t_set_pll(struct r820t_priv *priv,

enum v4l2_tuner_type type,

                vco_fra = pll_ref * 129 / 128;
        }

- if (nint > 63) {

+       if (nint > 76) {
                tuner_info("No valid PLL values for %u kHz!\n", freq);
                return -EINVAL;
        }
Signature · 1 line
1.8.3.2
#2 lore raw

Subject: [PATCH 2/2] [media] r820t: simplify divisor calculation

---
 drivers/media/tuners/r820t.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
drivers/media/tuners/r820t.c +0 -0
index e25c720..36dc63e 100644
@@ -596,13 +596,9 @@ static int r820t_set_pll(struct r820t_priv *priv,

enum v4l2_tuner_type type,

        while (mix_div <= 64) {
                if (((freq * mix_div) >= vco_min) &&
                   ((freq * mix_div) < vco_max)) {

- div_buf = mix_div;
- while (div_buf > 2) {
- div_buf = div_buf >> 1;
- div_num++;
- }

                        break;
                }
+               ++div_num;
                mix_div = mix_div << 1;
        }
Signature · 1 line
1.8.3.2
#3 lore raw

Re: Subject: [PATCH 2/2] [media] r820t: simplify divisor calculation

Jiří,

It's difficult to understand a patch like this, let alone merging it, without any kind of explanation. At a quick glance, the patch looks wrong. If you believe the patch is correct, then please resubmit with some sort of description and explanation for the change.

Best regards,

Mike Krufky

Quoted 29 lines · Jiří Pinkava
---
 drivers/media/tuners/r820t.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
drivers/media/tuners/r820t.c +1 -6
index e25c720..36dc63e 100644
@@ -596,13 +596,9 @@ static int r820t_set_pll(struct r820t_priv *priv,
 enum v4l2_tuner_type type,
        while (mix_div <= 64) {
                if (((freq * mix_div) >= vco_min) &&
                   ((freq * mix_div) < vco_max)) {
-                       div_buf = mix_div;
-                       while (div_buf > 2) {
-                               div_buf = div_buf >> 1;
-                               div_num++;
-                       }
                        break;
                }
+               ++div_num;
                mix_div = mix_div << 1;
        }
 
--

1.8.3.2

#4 lore raw

Re: [PATCH 1/2] [media] r820t: fix nint range check

In reply to Jiří Pinkava

Jiří,

Do you have any documentation that supports this value change? Changing this value affects the algorithm, and we'd be happier making this change if the patch included some better description and perhaps a reference explaining why the new value is correct.

Regards,

Mike Krufky

Quoted 25 lines · Jiří Pinkava

Use full range of VCO parameters, fixes tunning for some frequencies.

---
 drivers/media/tuners/r820t.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/media/tuners/r820t.c +1 -1
index 1c23666..e25c720 100644
@@ -637,7 +637,7 @@ static int r820t_set_pll(struct r820t_priv *priv,
 enum v4l2_tuner_type type,
                vco_fra = pll_ref * 129 / 128;
        }
 
-       if (nint > 63) {
+       if (nint > 76) {
                tuner_info("No valid PLL values for %u kHz!\n", freq);
                return -EINVAL;
        }

--
1.8.3.2

#5 lore raw

Re: [PATCH 1/2] [media] r820t: fix nint range check

Mike,

unfortunately no documentation can be referenced except preliminary version of
datasheet (1).This change is based on lucky guess and supported by lot of testing on real hardware.

This change add support for devices with Xtal frequency bellow 28.8MHz.

Quoted 1 line

From Nint are computed values of Ni and Si. For 28.8MHz crystal can

reach up to 12 / 3 (Ni / Si). Tuner supports crystals with frequencies
(1) 12, 16, 20, 20.48, 24, 27, 28.8, 32 MHz, but this kind of device is rare to found.
Allowing Ni to go up to 15 instead of only 12 should be safe and for 15 / 3 (Ni / Si)
we can compute limit for Nint = max(Ni) * 4 + max(Si) + 13 = 76.

If This is sufficient and acceptable explanation I can add some sort of documentation into patch and resend it (both patches, I can prove I'm right :)

(1)
http://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

Quoted 36 lines

Jiří,

Do you have any documentation that supports this value change? Changing this value affects the algorithm, and we'd be happier making this change if the patch included some better description and perhaps a reference explaining why the new value is correct.

Regards,

Mike Krufky

Quoted 24 lines · Jiří Pinkava

Use full range of VCO parameters, fixes tunning for some frequencies.

---
 drivers/media/tuners/r820t.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/media/tuners/r820t.c +1 -1
index 1c23666..e25c720 100644
@@ -637,7 +637,7 @@ static int r820t_set_pll(struct r820t_priv *priv,
 enum v4l2_tuner_type type,
                vco_fra = pll_ref * 129 / 128;
        }
 
-       if (nint > 63) {
+       if (nint > 76) {
                tuner_info("No valid PLL values for %u kHz!\n", freq);
                return -EINVAL;
        }

--
1.8.3.2

#6 lore raw

Re: [PATCH 1/2] [media] r820t: fix nint range check

Do you have any comments on this, Mauro?

Assuming that Mauro is OK with this change, (since he is the author of this driver) then yes - please resubmit the patch with some explanation within comments inline or within the commit message.

Best regards,

Mike Krufky

Quoted 61 lines · Jiří Pinkava

Mike,

unfortunately no documentation can be referenced except preliminary version of
datasheet (1).This change is based on lucky guess and supported by lot of testing on real hardware.

This change add support for devices with Xtal frequency bellow 28.8MHz.

From Nint are computed values of Ni and Si. For 28.8MHz crystal can reach up to 12 / 3 (Ni / Si). Tuner supports crystals with frequencies
(1) 12, 16, 20, 20.48, 24, 27, 28.8, 32 MHz, but this kind of device is rare to found.
Allowing Ni to go up to 15 instead of only 12 should be safe and for 15 / 3 (Ni / Si)
we can compute limit for Nint = max(Ni) * 4 + max(Si) + 13 = 76.

If This is sufficient and acceptable explanation I can add some sort of documentation into patch and resend it (both patches, I can prove I'm right :)

(1)
http://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

Quoted 36 lines

Jiří,

Do you have any documentation that supports this value change? Changing this value affects the algorithm, and we'd be happier making this change if the patch included some better description and perhaps a reference explaining why the new value is correct.

Regards,

Mike Krufky

Quoted 24 lines · Jiří Pinkava

Use full range of VCO parameters, fixes tunning for some frequencies.

---
 drivers/media/tuners/r820t.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/media/tuners/r820t.c +1 -1
index 1c23666..e25c720 100644
@@ -637,7 +637,7 @@ static int r820t_set_pll(struct r820t_priv *priv,
 enum v4l2_tuner_type type,
                vco_fra = pll_ref * 129 / 128;
        }
 
-       if (nint > 63) {
+       if (nint > 76) {
                tuner_info("No valid PLL values for %u kHz!\n", freq);
                return -EINVAL;
        }

--
1.8.3.2

#7 lore raw

Re: [PATCH 1/2] [media] r820t: fix nint range check

Em Mon, 30 Sep 2013 12:46:52 -0400
Michael Krufky <[email protected]> escreveu:

Quoted 5 lines

Do you have any comments on this, Mauro?

Assuming that Mauro is OK with this change, (since he is the author of this driver) then yes - please resubmit the patch with some explanation within comments inline or within the commit message.

Michael,

Please don't top post. My comments are below.

Quoted 14 lines

Best regards,

Mike Krufky

Quoted 8 lines · Jiří Pinkava

Mike,

unfortunately no documentation can be referenced except preliminary version of
datasheet (1).This change is based on lucky guess and supported by lot of testing on real hardware.

This change add support for devices with Xtal frequency bellow 28.8MHz.

Hmm... I'm not seeing any test there checking if the frequency of the xtal is below to 28.8 MHz. So, if this logic apples only in this case, the patch needs to add a test for the xtal frequency.

Also, as the driver does some tests for other Raphael tuners, if this change is known to work fine only on rt820t, please add a test there for the tuner model too.

Except for that, I'm ok with the changes, provided that you add the explanations about it.

Quoted 53 lines
Quoted 53 lines

From Nint are computed values of Ni and Si. For 28.8MHz crystal can reach up to 12 / 3 (Ni / Si). Tuner supports crystals with frequencies
(1) 12, 16, 20, 20.48, 24, 27, 28.8, 32 MHz, but this kind of device is rare to found.
Allowing Ni to go up to 15 instead of only 12 should be safe and for 15 / 3 (Ni / Si)
we can compute limit for Nint = max(Ni) * 4 + max(Si) + 13 = 76.

If This is sufficient and acceptable explanation I can add some sort of documentation into patch and resend it (both patches, I can prove I'm right :)

(1)
http://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

Quoted 36 lines

Jiří,

Do you have any documentation that supports this value change? Changing this value affects the algorithm, and we'd be happier making this change if the patch included some better description and perhaps a reference explaining why the new value is correct.

Regards,

Mike Krufky

Quoted 24 lines · Jiří Pinkava

Use full range of VCO parameters, fixes tunning for some frequencies.

---
 drivers/media/tuners/r820t.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/media/tuners/r820t.c +1 -1
index 1c23666..e25c720 100644
@@ -637,7 +637,7 @@ static int r820t_set_pll(struct r820t_priv *priv,
 enum v4l2_tuner_type type,
                vco_fra = pll_ref * 129 / 128;
        }
 
-       if (nint > 63) {
+       if (nint > 76) {
                tuner_info("No valid PLL values for %u kHz!\n", freq);
                return -EINVAL;
        }

--
1.8.3.2

Signature · 2 lines
Cheers,
Mauro