On 02/13/2015 10:29 PM, Luca Barbato wrote:
+ uint8_t *pkt_data = 0;
This should be NULL (otherwise some checker will complain).
Okay, applied.
+ for (pass = 0; pass < 2; pass++) {
+ const uint8_t *buf1 = buf;
+ int len1 = len;
+
+ while (len1 > RTP_HEVC_AP_NALU_LENGTH_FIELD_SIZE) {
+ uint16_t nalu_size = AV_RB16(buf1);
+
+ /* pass the NALU length field */
+ buf1 += RTP_HEVC_AP_NALU_LENGTH_FIELD_SIZE;
+ len1 -= RTP_HEVC_AP_NALU_LENGTH_FIELD_SIZE;
+
+ if (nalu_size > 0 && nalu_size <= len1) {
+ if (pass == 0) {
+ pkt_size += sizeof(start_sequence) +
nalu_size;
+ } else {
+ /* A/V packet: copy start sequence */
+ memcpy(pkt_data, start_sequence,
sizeof(start_sequence));
+ /* A/V packet: copy NAL unit data */
+ memcpy(pkt_data +
sizeof(start_sequence), buf1, nalu_size);
+ /* shift pointer beyond the current NAL
unit */
+ pkt_data += sizeof(start_sequence) +
nalu_size;
+ }
+ }
here I'd add an else branch with the failure path or even better check
directly impossible situations (nalu_size larger than the packet
lenght, 0-sized nalu) and fail there.
Okay, applied.
The code itself might be better if moved to a function.
The code is a bit similar to the one of the H.264 parser. But both
parser differs in the handling of "RTP_HEVC_DOND_FIELD_SIZE".
At the moment, I would rather use one switch-case block than splitting
it into functions.
Best regards,
Thomas.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
