Lock-free sharing of live-recorded circular buffer resources
View Patent ↗Novel techniques are described for lock-free sharing of a circular buffer. Embodiments can provide shared, lock-free, constant-bitrate access by multiple consumer systems to a live stream of audiovisual information being recorded to a circular buffer by a producer. For example, when a producer system writes a data stream to the circular buffer, the producer system records shared metadata. When a consumer system desires to begin reading from the shared buffer at a particular time, the shared metadata is used to compute a predicted write pointer location and corresponding dirty region around the write pointer at the desired read time. A read pointer of the consumer system can be set to avoid the dirty region, thereby permitting read access to a stable region of the circular buffer without relying on a buffer lock.
1. A method for lock-free sharing of a circular buffer, the method comprising:
computing a predicted write pointer location, in association with a request to read data from a circular buffer at a request time, based on a start time and a bitrate of a data stream being written to the circular buffer;
computing a dirty region around the write pointer location that begins at a dirty region start bit of the circular buffer prior to the write pointer and ends at a dirty region end bit of the circular buffer subsequent to the write pointer; and
initiating reading of data of the data stream, responsive to the request, at a read pointer location in the circular buffer set to be outside the dirty region.
2. The method of claim 1 , wherein the computing the dirty region comprises:
setting the dirty region start bit of the circular buffer to be prior to the write pointer by a number of bits corresponding to a real-time read start lag time; and
setting the dirty region end bit of the circular buffer to be subsequent to the write pointer by the number of bits corresponding to the real-time read start lag time.
3. The method of claim 2 , wherein:
the request is received from a consumer system determined to have the real-time read start lag time; and
the number of bits is computed to correspond to the real-time read start lag time as a function of the bitrate of the data stream being written to the circular buffer.
4. The method of claim 1 , wherein the request to read data comprises an indication of one of a real-time reading request, or an archival recording request.
5. The method of claim 4 , wherein, responsive to the request to read data comprising the indication of the real-time reading request, the initiating reading comprises setting the read pointer location to a bit location of the circular buffer corresponding to the dirty region start bit.
6. The method of claim 4 , wherein, responsive to the request to read data comprising the indication of the archival recording request, the initiating reading comprises setting the read pointer location to a bit location of the circular buffer corresponding to the dirty region end bit.
7. The method of claim 1 , wherein:
the request time is a current time; and
the computing the predicted write pointer location comprises computing a present write pointer location based on the start time, the bitrate, a circular buffer size, and the current time.
8. The method of claim 1 , wherein:
the request time indicates a subsequent read time; and
the computing the predicted write pointer location is based on the start time, the bitrate, a circular buffer size, and the subsequent read time.
9. The method of claim 1 , wherein:
the data stream is being written to the circular buffer by a producer system along with the producer system writing shared metadata indicating one or more of the start time, the bitrate, or the buffer size; and
the computing the predicted write pointer location is by a consumer system in communication with the producer system and is based on the shared metadata.
10. The method of claim 1 , wherein the data stream is being written to the circular buffer by a producer system configured so that the bitrate is a hard-coded bitrate.
11. A lock-free circular buffer sharing system comprising:
a write pointer processor to compute a predicted write pointer location, in association with a request to read data from a circular buffer at a request time, based on a start time and a bitrate of a data stream being written to the circular buffer;
a dirty region processor to compute a dirty region around the write pointer location that begins at a dirty region start bit of the circular buffer prior to the write pointer and ends at a dirty region end bit of the circular buffer subsequent to the write pointer; and
a read module to initiate reading of data of the data stream, responsive to the request, at a read pointer location in the circular buffer set to be outside the dirty region.
12. The lock-free circular buffer sharing system of claim 11 , wherein the dirty region processor is to compute the dirty region by:
setting the dirty region start bit of the circular buffer to be prior to the write pointer by a number of bits corresponding to a real-time read start lag time; and
setting the dirty region end bit of the circular buffer to be subsequent to the write pointer by the number of bits corresponding to the real-time read start lag time.
13. The lock-free circular buffer sharing system of claim 12 , wherein:
the request is received from a consumer system determined to have the real-time read start lag time; and
the number of bits is computed to correspond to the real-time read start lag time as a function of the bitrate of the data stream being written to the circular buffer.
14. The lock-free circular buffer sharing system of claim 11 , wherein the request to read data comprises an indication of one of a real-time reading request, or an archival recording request.
15. The lock-free circular buffer sharing system of claim 14 , wherein,
responsive to the request to read data comprising the indication of the real-time reading request, the read module is to initiate reading by setting the read pointer location to a bit location of the circular buffer corresponding to the dirty region start bit.
16. The lock-free circular buffer sharing system of claim 14 , wherein, responsive to the request to read data comprising the indication of the archival recording request, the read module is to initiate reading comprises setting the read pointer location to a bit location of the circular buffer corresponding to the dirty region end bit.
17. The lock-free circular buffer sharing system of claim 11 , wherein:
the request time is a current time; and
the write pointer processor is to compute the predicted write pointer location by computing a present write pointer location based on the start time, the bitrate, a circular buffer size, and the current time.
18. The lock-free circular buffer sharing system of claim 11 , wherein:
the request time indicates a subsequent read time; and
the write pointer processor is to compute the predicted write pointer location based on the start time, the bitrate, a circular buffer size, and the subsequent read time.
19. The lock-free circular buffer sharing system of claim 11 , further comprising:
a memory system comprising the circular buffer;
a capture module to capture audiovisual data from an audiovisual source; and
a write module, coupled with the capture module and the memory system, to write the audio visual data as the data stream to the circular buffer.
20. A system for lock-free sharing of a circular buffer, the system comprising:
a set of processors;
a non-transient memory having instructions stored thereon, which, when executed, cause the set of processors to perform steps comprising:
computing a predicted write pointer location, in association with a request to read data from a circular buffer at a request time, based on a start time and a bitrate of a data stream being written to the circular buffer;
computing a dirty region around the write pointer location that begins at a dirty region start bit of the circular buffer prior to the write pointer and ends at a dirty region end bit of the circular buffer subsequent to the write pointer; and
initiating reading of data of the data stream, responsive to the request, at a read pointer location in the circular buffer set to be outside the dirty region.