BoringSSL (or OpenSSL) require that when SSL_write fails due to the
underlying socket being blocked, it's retried with the same parameters
until it succeeds. But we weren't doing this, and our socket
abstraction doesn't have an equivalent requirement. So when this was
occurring, we would just end up trying to send the next RTP or STUN
packet (instead of the packet that couldn't be sent), and BoringSSL
doesn't like that.
So, when this condition occurs now, we'll simply enter a "pending write"
mode and buffer the data that couldn't be completely sent. When the
underlying socket becomes writable again, or if Send is called again
before that happens, we retry sending the buffered data. Making both
BoringSSL and the upper layer of code that expects normal TCP socket
behavior happy.
Also adding some more logging, and fixing an issue with VirtualSocketServer
that made it behave slightly differently than PhysicalSocketServer when a
TCP packet is only partially read.
BUG=webrtc:7753
Review-Url: https://codereview.webrtc.org/2915243002
Cr-Commit-Position: refs/heads/master@{#18416}
This does not fix the myriad of other problems here, but at least
removes the dependency on CONF_VALUE.
BUG=526270
Review-Url: https://codereview.webrtc.org/2705603003
Cr-Commit-Position: refs/heads/master@{#16676}
Replaced by assigning value to a local variable, followed by a DCHECK.
Also deletes dead test code under the always false TEST_DIGEST define.
BUG=webrtc:6424
Review-Url: https://codereview.webrtc.org/2623473004
Cr-Commit-Position: refs/heads/master@{#16476}
Bulk of the changes were done using
git grep -l '#include "webrtc/base/common.h"' | \
xargs sed -i '\,^#include.*webrtc/base/common\.h,d'
followed by adding back the include in the few places where it is
still needed, and in one case (pseudotcp.cc) instead deleting its use
of RTC_UNUSED.
BUG=webrtc:6424
Review-Url: https://codereview.webrtc.org/2644103002
Cr-Commit-Position: refs/heads/master@{#16263}
These defines don't work any more, so they only cause confusion:
FEATURE_ENABLE_SSL
HAVE_OPENSSL_SSL_H
SSL_USE_OPENSSL
BUG=webrtc:7025
Review-Url: https://codereview.webrtc.org/2640513002
Cr-Commit-Position: refs/heads/master@{#16224}
Reason for revert:
Broke chromium build, due to a config being removed. Will add it back and remove the dependency in a chromium CL.
Original issue's description:
> Removing #defines previously used for building without BoringSSL/OpenSSL.
>
> These defines don't work any more, so they only cause confusion:
>
> FEATURE_ENABLE_SSL
> HAVE_OPENSSL_SSL_H
> SSL_USE_OPENSSL
>
> BUG=webrtc:7025
>
> Review-Url: https://codereview.webrtc.org/2640513002
> Cr-Commit-Position: refs/heads/master@{#16196}
> Committed: eaa826c2eeTBR=kjellander@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:7025
Review-Url: https://codereview.webrtc.org/2648003003
Cr-Commit-Position: refs/heads/master@{#16197}
These defines don't work any more, so they only cause confusion:
FEATURE_ENABLE_SSL
HAVE_OPENSSL_SSL_H
SSL_USE_OPENSSL
BUG=webrtc:7025
Review-Url: https://codereview.webrtc.org/2640513002
Cr-Commit-Position: refs/heads/master@{#16196}
Bulk of the changes were produced using
git grep -l ' ASSERT(' | grep -v test | grep -v 'common\.h' |\
xargs -n1 sed -i 's/ ASSERT(/ RTC_DCHECK(/'
followed by additional includes of base/checks.h in affected files,
and git cl format.
Also had to do some tweaks to #if !defined(NDEBUG) logic in the
taskrunner code (webrtc/base/task.cc, webrtc/base/taskparent.cc,
webrtc/base/taskparent.h, webrtc/base/taskrunner.cc), replaced to
consistently use RTC_DCHECK_IS_ON, and some of the checks needed
additional #if protection.
Test code was excluded, because it should probably use RTC_CHECK
rather than RTC_DCHECK.
BUG=webrtc:6424
Review-Url: https://codereview.webrtc.org/2620303003
Cr-Commit-Position: refs/heads/master@{#16030}
This change makes WebRTC no longer stop sending video when we receive an
EWOULDBLOCK error from the operating system. This was previously
causing calls on a slow link (where the first hop is slow) to rapidly
oscillate between starting and stopping video.
We still do need to stop sending packets if there is no known good
connection we can use for that. We used to generate a synthetic
EWOULDBLOCK error in that case. This CL replaces it with a different
code (ENOTCONN); EWOULDBLOCK no longer stops the stream but ENOTCONN
does.
I've updated all the places where we seemed to be generating EWOULDBLOCK
for reasons other than some buffer been full; please give it a thorough
look in case I missed something.
R=pthatcher@webrtc.org
Review URL: https://codereview.webrtc.org/2192963002 .
Cr-Commit-Position: refs/heads/master@{#13566}
Every message will now be traced with the location from which it was
posted, including function name, file and line number.
This CL also writes a normal LOG message when the dispatch took more
than a certain amount of time (currently 50ms).
This logging should help us identify messages that are taking
longer than expected to be dispatched.
R=pthatcher@webrtc.org, tommi@webrtc.org
Review URL: https://codereview.webrtc.org/2019423006 .
Cr-Commit-Position: refs/heads/master@{#13104}
This helps a lot on Android devices where the user threads can be scheduled with low priority when the app is in the background, causing spurious significantly delayed before a packet can be read from the socket. With this patch the timestamp is taken by the kernel when the packet actually arrives.
R=juberti@chromium.orgTBR=juberti@webrtc.org
BUG=webrtc:5773
Review URL: https://codereview.webrtc.org/1944683002 .
Cr-Commit-Position: refs/heads/master@{#12850}
Reason for revert:
We will make it possible to link to BoringSSL for WebRTC's usages of the crypto APIs and OpenSSL for other usages in the same binary. Once that is completed, we will reland this.
Original issue's description:
> Remove code interfacing legacy openssl.
>
> BUG=webrtc:5664
>
> Committed: https://crrev.com/4cd331beade6de16c073dcdaf89c4e038bdbf73f
> Cr-Commit-Position: refs/heads/master@{#12041}
TBR=tommi@webrtc.org,davidben@webrtc.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=webrtc:5664
Review URL: https://codereview.webrtc.org/1828773003 .
Cr-Commit-Position: refs/heads/master@{#12117}
This copies the contents (unittest excluded) of base/numerics in
chromium to base/numerics in webrtc. Files added:
- safe_conversions.h
- safe_conversions_impl.h
- safe_math.h
- safe_math_impl.h
A really old version of safe_conversions[_impl].h previously existed in
base/, this has been deleted and sources using it have been updated
to include the new base/numerics/safe_converions.h.
This CL also adds a DEPS file to webrtc/base.
NOPRESUBMIT=True
BUG=webrtc:5548, webrtc:5623
Review URL: https://codereview.webrtc.org/1753293002
Cr-Commit-Position: refs/heads/master@{#11907}
socket_ in OpenSSLAdapter should be (and is in tests) an AsyncSocket but
doesn't have to be an AsyncSocketAdapter. In tests this is
rtc::VirtualSocket which is an rtc::AsyncSocket. This also matches the
BIO_new_socket type signature.
This fixes the remaining UBSan vptr bot errors.
BUG=webrtc:5124, webrtc:5226
R=tommi@webrtc.org, torbjorng@webrtc.org
Review URL: https://codereview.webrtc.org/1639883002 .
Cr-Commit-Position: refs/heads/master@{#11391}
ARRAY_SIZE is the old version of arraysize and does not cover
all the cases in C++, arraysize is a copy of Chromium's
version and thus have wider coverage.
BUG=None
R=tommi@webrtc.org
Review URL: https://codereview.webrtc.org/1405023016
Cr-Commit-Position: refs/heads/master@{#10594}
1) Added SetMode() to SSLAdapter and OpenSSLAdapter so the mode can be set to
SSL_MODE_DTLS
2) OpenSSLAdapter overrides SendTo() and RecvFrom() to handle calls from
TurnPort via AsyncUdpSocket
3) OpenSSLAdapter derives from MessageHandler to implement an internal DTLS
timer
4) Updated SSLAdapter unit tests
BUG=
R=juberti@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/19059004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7981 4adac7df-926f-26a2-2b94-8c16560cd09d
