UDPSOCKET_ECHOTEST_NONBLOCK performance improvement. #10776
Conversation
|
@tymoteuszblochmobica, thank you for your changes. |
| if (tc_exec_time.read() >= time_allotted) { | ||
| break; | ||
| } | ||
| signals.wait_all(SIGNAL_SIGIO_RX, SIGIO_TIMEOUT); |
There was a problem hiding this comment.
The previous loop uses osSignalWait(SIGNAL_SIGIO_TX, SIGIO_TIMEOUT).status == osEventTimeout) which will wait that signal send to current thread.
However, I don't see anyone sending that signal.
The _sigio_handler() seems to set flags only on this EventFlags signals variable.
Please fix the previous loop to use the same eventflag variable.
There was a problem hiding this comment.
Changed
| } | ||
| break; | ||
| drop_bad_packets(*sock, 0); // timeout equivalent to set_blocking(false) |
There was a problem hiding this comment.
What is this drop_bad_packets()?
I don't see any need for it.. If we did not receive all packets, then what is there to receive?
There was a problem hiding this comment.
Removed
tymoteuszblochmobica
force-pushed
the
echotest
branch
from
451483c to
eab38bd
Compare
|
Updated |
|
There is conflict, so please rebase your work on top of master and push into this branch again. |
tymoteuszblochmobica
force-pushed
the
echotest
branch
from
eab38bd to
c223505
Compare
|
Rebased |
|
CI started |
Test run: FAILEDSummary: 1 of 3 test jobs failed Failed test jobs:
|
|
There's fix proposed in #10833, will restart CI after |
|
CI started |
Test run: FAILEDSummary: 1 of 4 test jobs failed Failed test jobs:
|
Test run: SUCCESSSummary: 4 of 4 test jobs passed |
Description
Refactoring UDPSOCKET_ECHOTEST_NONBLOCK test according to request from:
"ONME-4244 UDPSOCKET_ECHOTEST_NONBLOCK performance is not optimal".
Removed unnecessary thread creation for nonblocking socket read.
It will not improve latency for WISE-1570 but will keep code more readable.
Pull request type
Reviewers
@SeppoTakalo
Release Notes