Skip to contentSkip to conversation

Bug Report - Twinhan vp-1020, bt_8xx driver + frontend

linux-dvb Alain 4 participants 8 messages Dec 1, 2008 – Dec 2, 2008 Media (V4L/DVB)

lore
Share & tools
lore

Alain

author
#0 lore raw

Found the problem I believe.  The original code always ran dvb_frontend_swzigzag(fe) even if the algo  is 0 This fixes the issue for me.

dvb_frontend.c     2008-12-01 12:07:16.000000000 -0500 renamed from dvb_frontend.c      2008-12-01 12:07:28.000000000 -0500 +0 -0
@@ -645,6 +645,8 @@

                                break;
                        default:
                                dprintk("%s: UNDEFINED ALGO !\n", __func__); +                               dvb_frontend_swzigzag(fe); break;
                        }
                } else {

________________________________

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#1 lore raw

Digging in a little further.The dst_algo (which the twinhan uses) is set to return 0 as the default setting for the SW algo in dst.c, yet in dvb_frontend.h, the DVBFE_ALGO_SW algo is defined as 2. Which is the correct one here? Should dst.c be changed to return 2 as sw or is 0 the correct number for the SW algo and thus DVBFE_ALGO_SW be changed to return 0?
----- Original Message -----
From: Alain
To: [email protected]
Sent: Monday, December 01, 2008 12:18 PM
Subject: Re: [linux-dvb] Bug Report - Twinhan vp-1020,bt_8xx driver + frontend

Found the problem I believe. The original code always ran dvb_frontend_swzigzag(fe) even if the algo is 0 This fixes the issue for me.

  --- dvb_frontend.c      2008-12-01 12:07:28.000000000 -0500
  +++ /dvb_frontend.c     2008-12-01 12:07:16.000000000 -0500
  @@ -645,6 +645,8 @@
                                  break;
                          default:
                                  dprintk("%s: UNDEFINED ALGO !\n", __func__);
  +                               dvb_frontend_swzigzag(fe);
                                  break;
                          }
                  } else {

------------------------------------------------------------------------------

------------------------------------------------------------------------------

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#2 lore raw

2008/12/1 Alain Turbide <[email protected]>:

Quoted 6 lines

Digging in a little further.The dst_algo (which the twinhan uses) is set to return 0 as the default setting for the SW algo in dst.c, yet in dvb_frontend.h, the DVBFE_ALGO_SW algo is defined as 2. Which is the correct one here? Should dst.c be changed to return 2 as sw or is 0 the correct number for the SW algo and thus DVBFE_ALGO_SW be changed to return 0?

Is nobody else looking into this?! I would think this bug would have received a little more attention considering the number of people affected!

Please keep up the work, it's much appreciated! I, on behalf of several others who aren't subscribed to the ml, am monitoring this thread in hopes of a proper fix.

Thanks!
-Derek

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#3 lore raw

Am Dienstag 02 Dezember 2008 18:37:54 schrieb VDR User:

Quoted 11 lines

2008/12/1 Alain Turbide <[email protected]>:

Quoted 6 lines

Digging in a little further.The dst_algo (which the twinhan uses) is set to return 0 as the default setting for the SW algo in dst.c, yet in dvb_frontend.h, the DVBFE_ALGO_SW algo is defined as 2. Which is the correct one here? Should dst.c be changed to return 2 as sw or is 0 the correct number for the SW algo and thus DVBFE_ALGO_SW be changed to return 0?

Is nobody else looking into this?! I would think this bug would have received a little more attention considering the number of people affected!

Normally this is the traditionally claimed stuff of an Indian person called Manu Abraham.

But you can forget about Manu Abraham totally because he is, was and ever will be the sum of a total incapable anti-team working loudmouth and stupid idiot.

This guy may have his capabilities, but he definitely IS in the literal sense of the word nothing but an asocial incapable human being.

In fact there are people around here offering him some mental sponsoring out of invisible reasons (which is very typical for all motherfuckers that cannot go upright, you know).

But in a real honest enterprise or even real social community this guy would be eliminated without any second of hesitation.
Just sorted out as disfunctional rubbish, you know.

I think the functionary armchair gas passer Mauro Carvalho Chehab (the big Brazil boss without any legitimation, you know) will give you the appropriate answer on that.

Quoted 4 lines

Please keep up the work, it's much appreciated! I, on behalf of several others who aren't subscribed to the ml, am monitoring this thread in hopes of a proper fix.

Do not rely on stupid loud mouths and incarnations of inncapability! Wipe out the bums instead!

Quoted 7 lines

Thanks!
-Derek

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#4 lore raw
In reply to VDR User

Well, it's not a difficult fix now that I see it. The issue was that the original default for FE_ALGO_SW was 0 while FE_ALGO_HW was 1. Since there is an older documented option for the dst module called dst_algo that some people might still be using to force the tuning algo to sofware by setting dst_algo=0, there is no choice but to also make the default of DVBFE_ALGO_SW to also be 0 so that the values will match and the new code will still function with users who force dst_algo=0 on the dst module load.. The fix would thus be to go from: this in dvb_frontend.h
enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = (1 << 1),
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)

};

to this:
enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = 0,
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)

};

This is what I've done now and works well. This is the only change required
to fix the issue. In dst.c we could also default dst_algo to
DVB_FRONTEND_SW instead of 0 to make it more robust. I can't see any code else where that depends on DVBFE_ALGO_SW being set to 2.

For those that do not want to patch code, the alternate way to get the cards to work is to simply load the dst module with the dst_algo parm set. to 2:
ie. modprobe dst dst_algo=2 ( to have the dst module return the current
value of DVBFE_ALGO_SW) back to the front end code.

----- Original Message -----
From: "VDR User" <[email protected]>
To: "Alain Turbide" <[email protected]>
Cc: <[email protected]>
Sent: Tuesday, December 02, 2008 12:37 PM
Subject: Re: [linux-dvb] [FIXEd] Bug Report - Twinhan vp-1020, bt_8xx driver
+ frontend

Quoted 14 lines

2008/12/1 Alain Turbide <[email protected]>:

Quoted 8 lines

Digging in a little further.The dst_algo (which the twinhan uses) is set to
return 0 as the default setting for the SW algo in dst.c, yet in dvb_frontend.h, the DVBFE_ALGO_SW algo is defined as 2. Which is the correct one here? Should dst.c be changed to return 2 as sw or is 0 the correct number for the SW algo and thus DVBFE_ALGO_SW be changed to return
0?

Is nobody else looking into this?! I would think this bug would have received a little more attention considering the number of people affected!

Quoted 6 lines

Please keep up the work, it's much appreciated! I, on behalf of several others who aren't subscribed to the ml, am monitoring this thread in hopes of a proper fix.

Thanks!
-Derek

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#5 lore raw

modprobe dst dst_algo=1 (to select the HW tuning algo ) works as well.. Didnt realize the hw algo worked for those cards.

----- Original Message -----
From: "Alain Turbide" <[email protected]>
To: <[email protected]>
Sent: Tuesday, December 02, 2008 2:05 PM
Subject: Re: [linux-dvb] [FIXEd] Bug Report - Twinhan vp-1020,bt_8xx driver
+ frontend

Quoted 79 lines

Well, it's not a difficult fix now that I see it. The issue was that the original default for FE_ALGO_SW was 0 while FE_ALGO_HW was 1. Since there is an older documented option for the dst module called dst_algo
that some people might still be using to force the tuning algo to sofware by
setting dst_algo=0, there is no choice but to also make the default of DVBFE_ALGO_SW to also be 0 so that the values will match and the new code will still function with users who force dst_algo=0 on the dst module load..
The fix would thus be to go from: this in dvb_frontend.h
enum dvbfe_algo {

       DVBFE_ALGO_HW                   = (1 <<  0),
       DVBFE_ALGO_SW                   = (1 << 1),
       DVBFE_ALGO_CUSTOM               = (1 <<  2),
       DVBFE_ALGO_RECOVERY             = (1 << 31)

};

to this:
enum dvbfe_algo {

       DVBFE_ALGO_HW                   = (1 <<  0),
       DVBFE_ALGO_SW                   = 0,
       DVBFE_ALGO_CUSTOM               = (1 <<  2),
       DVBFE_ALGO_RECOVERY             = (1 << 31)

};

This is what I've done now and works well. This is the only change required
to fix the issue. In dst.c we could also default dst_algo to
DVB_FRONTEND_SW instead of 0 to make it more robust. I can't see any code else where that depends on DVBFE_ALGO_SW being set to 2.

For those that do not want to patch code, the alternate way to get the cards
to work is to simply load the dst module with the dst_algo parm set. to 2:
ie. modprobe dst dst_algo=2 ( to have the dst module return the current
value of DVBFE_ALGO_SW) back to the front end code.

----- Original Message -----
From: "VDR User" <[email protected]>
To: "Alain Turbide" <[email protected]>
Cc: <[email protected]>
Sent: Tuesday, December 02, 2008 12:37 PM
Subject: Re: [linux-dvb] [FIXEd] Bug Report - Twinhan vp-1020, bt_8xx driver
+ frontend

Quoted 14 lines

2008/12/1 Alain Turbide <[email protected]>:

Quoted 8 lines

Digging in a little further.The dst_algo (which the twinhan uses) is set to
return 0 as the default setting for the SW algo in dst.c, yet in dvb_frontend.h, the DVBFE_ALGO_SW algo is defined as 2. Which is the correct one here? Should dst.c be changed to return 2 as sw or is 0 the correct number for the SW algo and thus DVBFE_ALGO_SW be changed to return
0?

Is nobody else looking into this?! I would think this bug would have received a little more attention considering the number of people affected!

Quoted 6 lines

Please keep up the work, it's much appreciated! I, on behalf of several others who aren't subscribed to the ml, am monitoring this thread in hopes of a proper fix.

Thanks!
-Derek

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

Quoted 4 lines

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#6 lore raw
In reply to Alain Turbide
Quoted 21 lines · Alain Turbide

Well, it's not a difficult fix now that I see it. The issue was that the original default for FE_ALGO_SW was 0 while FE_ALGO_HW was 1. Since there is an older documented option for the dst module called dst_algo that some people might still be using to force the tuning algo to sofware by setting dst_algo=0, there is no choice but to also make the default of DVBFE_ALGO_SW to also be 0 so that the values will match and the new code will still function with users who force dst_algo=0 on the dst module load.. The fix would thus be to go from: this in dvb_frontend.h enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = (1 << 1),
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)

};

to this:
enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = 0,
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)
Quoted 7 lines

};

This is what I've done now and works well. This is the only change required
to fix the issue. In dst.c we could also default dst_algo to
DVB_FRONTEND_SW instead of 0 to make it more robust. I can't see any code else where that depends on DVBFE_ALGO_SW being set to 2.

Why should we rely on exact values, and not just modify the only place where these values are exposed to userspace, and change dst_get_tuning_algo to something similar to:

static int dst_get_tuning_algo(struct dvb_frontend *fe)

{
        return dst_algo ? DVBFE_ALGO_HW : DVBFE_ALGO_SW;
}

The other (not backward compatible) fix is to adjust the parameter description and use 1 and 2 as allowed settings.

Regards
Matthias

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

#7 lore raw

Good one Matthias. Your first option seems like the most practical one and
I can also confirm works. 1=HW, any other value will set SW

----- Original Message -----
From: "Matthias Schwarzott" <[email protected]>
To: <[email protected]>
Cc: "Alain Turbide" <[email protected]>
Sent: Tuesday, December 02, 2008 2:21 PM
Subject: Re: [linux-dvb] [FIXEd] Bug Report - Twinhan vp-1020, bt_8xx driver
+ frontend

Quoted 44 lines · Alain Turbide
Quoted 23 lines

Well, it's not a difficult fix now that I see it. The issue was that the original default for FE_ALGO_SW was 0 while FE_ALGO_HW was 1. Since there is an older documented option for the dst module called dst_algo that some people might still be using to force the tuning algo to
sofware by setting dst_algo=0, there is no choice but to also make the default of DVBFE_ALGO_SW to also be 0 so that the values will match and the
new code will still function with users who force dst_algo=0 on the dst module load.. The fix would thus be to go from: this in dvb_frontend.h enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = (1 << 1),
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)

};

to this:
enum dvbfe_algo {

        DVBFE_ALGO_HW                   = (1 <<  0),
        DVBFE_ALGO_SW                   = 0,
        DVBFE_ALGO_CUSTOM               = (1 <<  2),
        DVBFE_ALGO_RECOVERY             = (1 << 31)
Quoted 9 lines

};

This is what I've done now and works well. This is the only change required
to fix the issue. In dst.c we could also default dst_algo to
DVB_FRONTEND_SW instead of 0 to make it more robust. I can't see any code
else where that depends on DVBFE_ALGO_SW being set to 2.

Why should we rely on exact values, and not just modify the only place where
these values are exposed to userspace, and change dst_get_tuning_algo to something similar to:

static int dst_get_tuning_algo(struct dvb_frontend *fe)

{
       return dst_algo ? DVBFE_ALGO_HW : DVBFE_ALGO_SW;
}
Quoted 6 lines

The other (not backward compatible) fix is to adjust the parameter description
and use 1 and 2 as allowed settings.

Regards
Matthias

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

_______________________________________________
linux-dvb mailing list
[email protected]
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb