liblog: longer timeout for tail operations
Tail operation can take longer than 2s on CPU bounded devices.
In our tests, stuffing the log buffer and running logcat -t needed
slightly more than 2s on a 2 core CPU device. With the current 2s
timeout in place recv() times out with no logs received. Increasing the
timeout to 5s gives extra buffer for scenarios of high CPU usage.
Bug: 303389424
Test: Following test was performed before/after
1. increase ring size to 64MB (-G)
2. read back size and usage (-g)
3. run while loop to fill in logs
4. wait 1h30
5. read back size and usage (-g) now logs have 21/16/21 MBs of data
6. tail last 10 entries => logcat times out after 2s
7. tail last 500 logs logcat successful (hot ram?)
8. tail last 100 entries works
9. tail last 10 again works this time
With the 5s timeout logcat does not time out in step 6.
Change-Id: I2a18d45a3c70f2e463a4fdb5498ebb1606757945
Signed-off-by: Andrei Diea <adiea@google.com>
1 file changed