admin/webrtc_m130
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Changing from hard to soft error when temporary DTLS buffer is full.

Browse Source 
We thought we had resolved this issue earlier, by reading DTLS
records in a loop. But this condition may be triggered in other cases,
such as when an internal DTLS error occurs and more DTLS records
continue to be received afterwords.

Changing this from a hard to soft error will avoid a crash (which
is happening more frequently for whatever reason) and hopefully
enable us to collect logs to debug the issue further.

Bug: chromium:1063834
Change-Id: I22c01a9e064a9db65bab38d00c62a424b5a27437
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/172560
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Reviewed-by: Jonas Oreland <jonaso@webrtc.org>
Commit-Queue: Jonas Oreland <jonaso@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30955}
This commit is contained in:
Taylor Brandstetter 2020-03-31 13:32:42 -07:00 committed by Commit Bot
parent 7ee8a88064
commit 784630f0e6
1 changed files with 7 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
p2p/base/dtls_transport.cc
View File

@ -99,12 +99,15 @@ rtc::StreamResult StreamInterfaceChannel::Write(const void* data,
}
bool StreamInterfaceChannel::OnPacketReceived(const char* data, size_t size) {
// We force a read event here to ensure that we don't overflow our queue.
bool ret = packets_.WriteBack(data, size, NULL);
RTC_CHECK(ret) << "Failed to write packet to queue.";
if (ret) {
SignalEvent(this, rtc::SE_READ, 0);
if (!ret) {
// Somehow we received another packet before the SSLStreamAdapter read the
// previous one out of our temporary buffer. In this case, we'll log an
// error and still signal the read event, hoping that it will read the
// packet currently in packets_.
RTC_LOG(LS_ERROR) << "Failed to write packet to queue.";
}
SignalEvent(this, rtc::SE_READ, 0);
return ret;
}