Attention is currently required from: neels, pespin, dexter.

fixeria has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmocore/+/29475 )

Change subject: fix gsm0808_sc_cfg <-> gsm48_mr_cfg conversion ......................................................................

Patch Set 4:

(1 comment)

File include/osmocom/gsm/protocol/gsm_04_08.h:

https://gerrit.osmocom.org/c/libosmocore/+/29475/comment/65ec494f_715300db PS2, Line 627: return ((uint8_t *)cfg)[1];

gsm48_multi_rate_conf has two uint8_t members, AFAICT adding a union would require adding another ex […]

You don't really need to name the union itself, it can be anonymous (just like structures). The problem is that bit-fields are treated as individual fields in a union, so doing what Pau suggested:

``` /* I removed the LE variant for the sake of readability * see https://people.osmocom.org/fixeria/union_bfield_test.c */ struct gsm48_multi_rate_conf { uint8_t ver:3, nscb:1, icmi:1, spare:1, smod:2; union { uint8_t m12_2:1, m10_2:1, m7_95:1, m7_40:1, m6_70:1, m5_90:1, m5_15:1, m4_75:1; uint8_t mode; }; } __attribute__((packed)); ```

will result in the following behavior:

``` struct gsm48_multi_rate_conf mrc = { 0 }; // mrc.mode = 0x00 // m4_75=0, m5_15=0, m5_90=0, m6_70=0, m7_40=0, m7_95=0, m10_2=0, m12_2=0

mrc.m4_75 = 1; // mrc.mode = 0x01 // m4_75=1, m5_15=1, m5_90=1, m6_70=1, m7_40=1, m7_95=1, m10_2=1, m12_2=1

mrc.m10_2 = 0; // mrc.mode = 0x00 // m4_75=0, m5_15=0, m5_90=0, m6_70=0, m7_40=0, m7_95=0, m10_2=0, m12_2=0

mrc.m12_2 = 1; // mrc.mode = 0x01 // m4_75=1, m5_15=1, m5_90=1, m6_70=1, m7_40=1, m7_95=1, m10_2=1, m12_2=1 ```

Definitely not what we want. Here is an alternative approach:

``` union gsm48_multi_rate_conf_u { struct gsm48_multi_rate_conf mrc; struct { uint8_t hdr; uint8_t mode; }; }; ```

This should work as expected, but I am fine with the current approach. Keep it as it is.