Opened 7 years ago
Closed 4 years ago
Last modified 4 years ago
#11564 closed Patch - Feature (fixed)
Feature patch to increase read size in hdhrstreamhandler.cpp
| Reported by: | Owned by: | ||
|---|---|---|---|
| Priority: | minor | Milestone: | 29.0 |
| Component: | MythTV - HDHomeRun | Version: | Master Head |
| Severity: | medium | Keywords: | |
| Cc: | Ticket locked: | no |
Description
In ticket #10414 ( patch http://code.mythtv.org/trac/changeset/2be3dd91fba8d438cb25d81a1ba9d0d5e8657fb6/mythtv ) the patch increased the buffersize for the hdhr. This was ineffective in that patch, since that particular variable was no longer used. This patch increases the maximum returned read size to be 1/8 of a second of potential buffer (implementing, partially, what I believe the original ticket was intended to do). With the presumption that the loop *is* executed around 20 ms (the usleep counter) the current 64KB read limit, with a high bitrate stream (which can produce ~100KB in 20ms), can result in 3 or 4 calls through the hdhr library to read all the buffered data. Increasing the amount of data to examine to the 1/8 second value (~300KB) results in reducing the number of calls to 2 or 3 potentially reducing the lock overheads (whether it is the higher value will depend on whether a new packet arrived while processing the existing set of packets, and lower bitrate content (a more common case) will show no improvement).
Patch:
https://github.com/garybuhrmaster/mythtv/commit/e25720a96f35d794209f5b0f3369086e187756ff
Change History (6)
comment:1 Changed 7 years ago by
| Component: | MythTV - DVB → MythTV - HDHomeRun |
|---|---|
| Owner: | changed from danielk to Silicondust |
comment:2 Changed 7 years ago by
comment:3 Changed 7 years ago by
| Owner: | changed from Silicondust to danielk |
|---|---|
| Status: | new → assigned |
The reassignment was automatic with the change of component.
comment:4 Changed 4 years ago by
| Owner: | danielk deleted |
|---|---|
| Status: | assigned → new |
comment:5 Changed 4 years ago by
| Owner: | set to Gary Buhrmaster <gary.buhrmaster@…> |
|---|---|
| Resolution: | → fixed |
| Status: | new → closed |
comment:6 Changed 4 years ago by
| Milestone: | unknown → 29.0 |
|---|
Should the assignment not remain with danielk, since it is his previous code that added and then deleted the buffer adjustments (that were ineffective for other reasons) that this patch is attempting to (re)implement, so he is probably the most knowledgeable about this code fragment?