Граф коммитов

1687 Коммитов

Автор SHA1 Сообщение Дата
Patrick McManus e364e4019d bug 623921: Out of an abundance of caution back out feature: 592284 - HTTP Syn Retry r=honzab a=backout 2011-01-07 11:11:22 -05:00
Patrick McManus 0ce3b79531 Bug 613977 - Intermittent invalid certificate error prompt in security tests causing timeouts r=honzab a=beta-N 2011-01-05 08:37:45 -05:00
Josh Matthews c32f7e8691 Bug 616861 - Replicate behaviour of nsHttpChannel in regards to failure status codes in OnStopRequest. r=jduell a=blocking-fennec 2010-12-06 18:28:45 -05:00
Mark Banner ebfbdfc948 Non-IPC Bustage fix from bug 613977/bug 614677/bug 614950 - add nsAutoPtr.h header to fix compilation bustage. a=Bustage fix 2010-12-16 14:35:12 +00:00
Patrick McManus 111d701adf Bug 613977 - Intermittent invalid certificate error prompt (partial) r=honzab a=blocking-beta9
Bug 614677 - Connection is reset message appears intermittently
Bug 614950 - Connections stall occasionally after 592284 landed

A couple of follow-on changes to 592284 rolled together to prevent
diff conflicts

1] Set the securitycallback information for unused speculative
connections in the connection manager to be the new cloned connection
rather than the one they originated on. (613977)

2] When adding unused speculative connections to the connection
manager, due so with a short timeout (<= 5 seconds) as some servers
get grumpy if they haven't seen a request in that time. Most will
close the connection, but some will just sit there quietly and RST
things when the connection is used - so if you don't use the
connection quickly don't use it at all. This is probably a L4 load
balancer issue, actually. Mozillazine illustrates the
problem. Connections are made in bursts anyhow, so the reuse optimization is
likely still quite useful. (614677 and 614950)

3] mark every connection in the connection manager persistent
conneciton pool as "reused". This allows the transaction to be
restarted if a RST is recvd upon sending the requests (see #2) - with
the conservative timeout this is now a rare event, but still possible
so recovery is the right thing to do. (614677 and 614950)

4] obtain an nshttpconnection object from the connection manager,
subject to the max connection constraints, at the same time as
starting the backup conneciton. If we defer that until recycling time
the exceeded limits of the SocketService can cause problems for other
connections.

also re-enables the syn retry feature by default.

r+ honzab
2010-12-16 08:50:36 -05:00
Boris Zbarsky f323857cf9 Bug 616591 followup to fix a comment., a=DONTBUILD 2010-12-08 21:01:17 -05:00
Honza Bambas 81b05b99bc Bug 616591. Fix crash due to http transactions trying to get connection info from the http pipeline while it's being closed. r=bzbarsky, a=blocker 2010-12-08 20:57:14 -05:00
Ehsan Akhgari 49d4a28c32 Bug 616591 - Firefox 4.0b8pre crash in [@ nsHttpPipeline::GetConnectionInfo(nsHttpConnectionInfo**]; r=bzbarsky a=blocking-beta8+ 2010-12-04 04:21:03 -05:00
Ehsan Akhgari 502ef553b8 Bug 614286 followup - Reorder the constructor initializer list to prevent compiler warnings
--HG--
extra : rebase_source : 3ab3d1371f8d8b396e665cffb5c57dc96a7187f3
2010-12-03 09:56:13 -05:00
Michal Novotny d56d1840a3 bug 363109 - body of HTTP 304 response is treated as a HTTP/0.9 response to subsequent HTTP request. r=biesi, sr=bz, a=blocker 2010-12-01 02:02:52 +02:00
Ehsan Akhgari d571c78f3a Bug 614286 - Don't initialize the socket transport service unless we know that we're online; r=bzbarsky a=blocking-beta8+ 2010-11-25 00:20:11 -05:00
Ehsan Akhgari ebb57cf670 Backed out changeset 630b08a7fe63 because of xpcshell test failures 2010-11-30 13:57:24 -05:00
Ehsan Akhgari c6b4773258 Bug 614286 - Don't initialize the socket transport service unless we know that we're online; r=bzbarsky a=blocking-beta8+ 2010-11-25 00:20:11 -05:00
Christian Biesinger a87f071eaa Backed out changeset dd249f45dc3d
Clearly I don't know hg well enough, didn't mean to commit all this. a=I-messed-up
2010-11-29 15:34:30 -08:00
Christian Biesinger e0273458ec Bug 614717 Make test_bug369787.js not hang when the test fails by ensuring
that do_test_finished is called even when an exception is thrown.
Also make the test work in the presence of PAC, where the channel changes
between asyncOpen and onStartRequest.

r=bz a=test-only
2010-11-29 15:28:40 -08:00
bjarne@runitsoft.com 3e5b7cb784 Bug 468426 - improved support for "Vary: Cookie"-headers , r=biesi a=bsmedberg 2010-11-26 12:13:37 -05:00
Benjamin Smedberg dea04ce175 Bug 614229 - nsHttpHandler getting the private-browsing service can cause recursive initialization of layout, and I suspect can cause incorrect behavior in Firefox because we may end up creating HTTP URIs before we're finished with XPCOM registration, r=honza+ehsan a=blocking bug 581335 which blocks Fennec b3 2010-11-24 08:58:42 -05:00
Honza Bambas 069ee7e772 Bug 591707 - e10s: handle redirects from HTTP to a different protocol, r=jduell, a=blocking-fennec2.0b3+ 2010-11-23 23:56:06 +01:00
Michal Novotny 7ba89a94b4 bug 599127 - Wyciwyg channel writes end up doing synchronous disk IO. r=jduell, a=betaN 2010-11-23 14:01:30 +02:00
Patrick McManus 05c44b2d25 Bug 592284 - Accelerate TCP connection retries in HTTP. r=honzab a=blocking2.0
Losing a TCP SYN requires a long painful (typically 3 second) delay
before being retried. This patch creates a second parallel connection
attempt for any nsHttpConnection which has not become writable before
a timeout occurs.

If you assume .5% packet loss, this converts a full 3 second delay
from a 1 in 200 event into a 1 in 40,000 event.

Whichever connection establishes itself first is used. If another one
has been started and it does connect before the one being used is
closed then the extra one is handed to the connection manager for use
by a different transaction - essentially a persistent connection with
0 previous transactions on it. (Another way to think about is
pre-fetching a 3WHS on a high latency connection).

The pref network.http.connection-retry-timeout controls the amount of
time in ms to wait for success on the initial connection before beginning
the second one. Setting it to 0 disables the parallel connection, the
default is 250.
2010-11-21 09:50:36 +01:00
Michael Wu c54ad808c2 Backed out changeset 090046fc6891 bug 482935 due to a=orange 2010-11-10 19:29:51 -08:00
bjarne@runitsoft.com 4ba6ae28f4 Bug 482935 - Cancel() from OnStopRequest should not cause cache entry to be doomed;
(Av1f) Allow content to be cached if it was loaded successfully.
r=(dwitte, jduell) a2.0=benjamin.
2010-11-11 01:02:15 +01:00
Ehsan Akhgari 6596954638 Backed out changeset c85c2f1b0566 because of oranges on mochitest-1 2010-11-10 17:46:32 -05:00
bjarne@runitsoft.com ce18152004 Bug 482935 - Cancel() from OnStopRequest should not cause cache entry to be doomed;
(Av1e) Allow content to be cached if it was loaded successfully.
r=(dwitte, jduell) a2.0=benjamin.
2010-11-10 22:28:54 +01:00
Mark Finkle 28678340e7 Bug 608258 - Crash at visiting AMO [r=jdm a=blocking-fennecb2] 2010-10-29 11:38:26 -04:00
Kai Engert e4730af01f Bug 566478 - Invalid SSL cert error is not mobile-friendly [r=mayhemer a=blocking-fennecb2] 2010-10-28 11:46:00 -04:00
Patrick McManus 1e0ea7a95f Bug 606719 - Browser stalls with connections left hanging in TCP close wait state. Part 2: fix it. r=dwitte, a=b7+ 2010-10-28 10:10:03 -07:00
Mark Finkle b09ca98a42 Backout bug 566478 due to test orange a=orangehurts 2010-10-28 12:17:01 -04:00
Honza Bambas fe340c9a46 Bug 566478 - Invalid SSL cert error is not mobile-friendly [r=kaie a=blocking-fennecb2] 2010-10-27 14:35:00 -04:00
Honza Bambas 926c40f427 Bug 597374 - Downloading of a file from a password protected directory fails, r=dolske, sr=bzbarsky, a=blocking2.0:betaN+ 2010-10-25 18:47:23 +02:00
Mark Banner 8e65af0085 Fix non-IPC bustage from bug 536289. Patch as agreed with jdm over irc. a=bustage-fix. 2010-10-21 22:01:38 +01:00
Alon Zakai ba212d56e1 Bug 536289 - Part 2: Make FTP work for fennec. r=jduell a=blocking-fennec 2010-08-10 14:47:00 -04:00
Josh Matthews 33d17004e2 Bug 536289 - Part 1: Move IPC::InputStream to a more general location. r=jduell a=blocking-fennec 2010-10-14 14:21:48 -04:00
Jae-Seong Lee-Russo b8363c8da8 Bug 564553 - e10s HTTP: Serialize nsInputStreams to support large file uploads. r=dwitte, a=blocking-fennec2.0b2+ 2010-10-21 11:36:13 -07:00
Honza Bambas e3cf711385 Bug 536295 - e10s HTTP: offline application cache, r=dwitte, sr=cbiesinger, a=fennec-2.0b2+
--HG--
rename : uriloader/prefetch/nsOfflineCacheUpdate.cpp => uriloader/prefetch/OfflineCacheUpdateChild.cpp
rename : uriloader/prefetch/nsOfflineCacheUpdate.cpp => uriloader/prefetch/nsOfflineCacheUpdateService.cpp
2010-10-20 19:12:32 +02:00
Dan Witte de6d8ff419 Bounce input stream and FTP patches for Windows test failure. a=bustage 2010-10-19 19:46:54 -07:00
Alon Zakai 80f231dc57 Bug 536289 - Part 2: Make FTP work for fennec. r=jduell a=blocking-fennec 2010-08-10 14:47:00 -04:00
Josh Matthews 2836986309 Bug 536289 - Part 1: Move IPC::InputStream to a more general location. r=jduell a=blocking-fennec 2010-10-14 14:21:48 -04:00
Jae-Seong Lee-Russo e4d607cff8 Bug 564553 - e10s HTTP: Serialize nsInputStreams to support large file uploads. Part 1: serialize them. r=dwitte, a=blocking-fennec2.0b2+ 2010-10-19 17:24:52 -07:00
Josh Matthews 88588a8fdf Bug 536289 - Generalize IPDL event queue some more. r=jduell a=blocking-fennec 2010-10-16 01:26:14 -04:00
Patrick McManus ab5ab1fd83 Bug 593386 - HTTP Pipeline Blacklist annotations. r=jduell
Add EFAServer, Nestscape Enterprise 4/5/6, and Weblogic <=6 to
server pipeline blacklist, joining iis/4, iis/5, and nses/3.

The previous code did 3 strcasestr()s on every http transaction to check
the old blacklist - it did that even if pipelining was configured off.

The new code creates indexes to the lookups according to the first
char and changes the comparison to straight strncmp()s which are
totally fine for finding these signatures... Servers that begin with A
don't need any string operations now, and even servers that begin with
M have gone from 3 strcasestr()s to 2 strcncmp()s which is definitely
better.
2010-09-03 21:35:49 -04:00
Dão Gottwald dfd3ee36d8 Bug 591573 - remove support for general.useragent.vendor & vendorSub. r=dwitte sr=dveditz a=jst 2010-10-15 11:11:19 +02:00
Alon Zakai 1ed5f2a9ec Bug 517584 - Fix orange in test for bug 480619. r=dwitte a=blocking-fennec 2010-10-14 17:28:45 -07:00
Jon Hemming 8c32c3ad25 Bug 591881 - Suspend pruning of idle connections when possible. r=dwitte a=blocking-fennec 2010-10-14 17:28:43 -07:00
Honza Bambas 6fdab9c0bc Bug 586766 - e10s http: fix redirect/cancel race. r=jduell, a=blocking-fennec-b2 2010-10-08 22:07:58 -07:00
Honza Bambas 8d3a261e4f Bug 586738 - e10s http: check HttpChannelChild::Redirect1Begin cancel logic. r=jduell, a=blocking-fennec-b2 2010-10-08 22:07:49 -07:00
Josh Matthews 12e271e689 Bug 587171 - Pass SetPriority IPC messages to redirected channels. r=honzab a=blocking-fennec
--HG--
extra : rebase_source : ceffe93f7f98434db5b5c82350cfc7963bad7640
2010-09-15 13:12:38 -07:00
Josh Matthews 4b4cdaeec2 Bug 591708 - Initiate all PHttpChannel protocol deletion from the child to avoid races. r=honzab a=blocking-fennec
--HG--
extra : rebase_source : 76568dbe568345aaf276ef166407c9338bf8f477
2010-09-15 12:26:15 -07:00
Josh Matthews 95c514d87f Bug 598076 - Update HTTPChannelChild to use new general event queueing infrastructure. r=jduell a=blocking-fennec
--HG--
extra : rebase_source : a6e12cbcf6a98139620f19590ada2cde583d70bd
2010-09-20 14:37:13 -04:00
Doug Turner 66fb6fa80b Bug 558624 - e10s Cookies: Optimize implementation. Part 5. Mop up mRemoteChannel and its uses. r=dwitte. a=blocking-fennec
--HG--
extra : rebase_source : 1a9ca4b7e50588c556d1b290cca1c3c4546902d1
2010-10-01 21:17:27 -07:00