Discrepancy in the module amr

[janakj]

Function encode_wb makes sure that the target array is at least NB_SERIAL_MAX:

...
if (*len < NB_SERIAL_MAX)
  return ENOMEM;
...

but the resiprocal decode_wb actually checks that the input is no more than NB_SERIAL_MAX+1:

...
if (len > (1+NB_SERIAL_MAX))
  return EINVAL;

I am not sure if this is an actual error, but the discrepancy looks suspicious.

[juha-h]

Tests go the same way also in nb encode/decode. Surely looks suspicious, but can't tell for sure.

[alfredh]

hi @janakj how are you doing, are you still working on VoIP projects ?

The packet has 1 byte in the beginning:

        /* CMR value 15 indicates that no mode request is present */
        buf[0] = 15 << 4;

so the correct answer should be 1+ NB_SERIAL_MAX

is anyone willing to make a tested fix for this ?

Alfred

[janakj]

Hi Alfred,

hi @janakj how are you doing, are you still working on VoIP projects ?

Yeah, occasionally, but mostly on special-purpose VoIP setups, for example:
https://irtlab.gitlab.io/lmr-emulator/

-Jan