1
0
mirror of https://github.com/DigitalDevices/dddvb.git synced 2023-10-10 13:37:43 +02:00

use kernel gcd and move comment regarding calculation to correct

function
This commit is contained in:
Ralph Metzler 2017-12-11 23:04:14 +01:00
parent 2d41161cb4
commit a14cbc0eff

View File

@ -25,6 +25,7 @@
#include "ddbridge-io.h"
#include <linux/dvb/mod.h>
#include <linux/gcd.h>
/****************************************************************************/
/****************************************************************************/
@ -74,64 +75,6 @@ inline s64 RoundPCRDown(s64 a)
return a & ~(HW_LSB_MASK - 1);
}
/* Calculating KF, LF from Symbolrate
*
* Symbolrate is usually calculated as (M/N) * 10.24 MS/s
*
* Common Values for M,N
* J.83 Annex A,
* Euro Docsis 6.952 MS/s : M = 869, N = 1280
* 6.900 MS/s : M = 345, N = 512
* 6.875 MS/s : M = 1375, N = 2048
* 6.111 MS/s : M = 6111, N = 10240
* J.83 Annex B **
* QAM64 5.056941 : M = 401, N = 812
* QAM256 5.360537 : M = 78, N = 149
* J.83 Annex C **
* 5.309734 : M = 1889, N = 3643
*
* For the present hardware
* KF' = 256 * M
* LF' = 225 * N
* or
* KF' = Symbolrate in Hz
* LF' = 9000000
*
* KF = KF' / gcd(KF',LF')
* LF = LF' / gcd(KF',LF')
* Note: LF must not be a power of 2.
* Maximum value for KF,LF = 13421727 ( 0x7FFFFFF )
* ** using these M,N values will result in a small err (<5ppm)
* calculating KF,LF directly gives the exact normative result
* but with rather large KF,LF values
*/
static inline u32 gcd(u32 u, u32 v)
{
int s = 0;
while (((u | v) & 1) == 0) {
s += 1;
u >>= 1;
v >>= 1;
}
while ((u & 1) == 0)
u >>= 1;
do {
while ((v & 1) == 0)
v >>= 1;
if (u > v) {
u32 t = v;
v = u;
u = t;
}
v = v - u;
} while (v != 0);
return (u << s);
}
/****************************************************************************/
/****************************************************************************/
/****************************************************************************/
@ -310,6 +253,38 @@ static int mod_set_ibitrate(struct ddb_mod *mod, u64 ibitrate)
return 0;
}
/* Calculating KF, LF from Symbolrate
*
* Symbolrate is usually calculated as (M/N) * 10.24 MS/s
*
* Common Values for M,N
* J.83 Annex A,
* Euro Docsis 6.952 MS/s : M = 869, N = 1280
* 6.900 MS/s : M = 345, N = 512
* 6.875 MS/s : M = 1375, N = 2048
* 6.111 MS/s : M = 6111, N = 10240
* J.83 Annex B **
* QAM64 5.056941 : M = 401, N = 812
* QAM256 5.360537 : M = 78, N = 149
* J.83 Annex C **
* 5.309734 : M = 1889, N = 3643
*
* For the present hardware
* KF' = 256 * M
* LF' = 225 * N
* or
* KF' = Symbolrate in Hz
* LF' = 9000000
*
* KF = KF' / gcd(KF',LF')
* LF = LF' / gcd(KF',LF')
* Note: LF must not be a power of 2.
* Maximum value for KF,LF = 13421727 ( 0x7FFFFFF )
* ** using these M,N values will result in a small err (<5ppm)
* calculating KF,LF directly gives the exact normative result
* but with rather large KF,LF values
*/
int ddbridge_mod_output_start(struct ddb_output *output)
{
struct ddb *dev = output->port->dev;