gecko/netwerk/protocol/http/nsHttpConnection.cpp

1093 lines
37 KiB
C++
Raw Normal View History

/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
/* vim:set ts=4 sw=4 sts=4 et cin: */
/* ***** BEGIN LICENSE BLOCK *****
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
*
* The contents of this file are subject to the Mozilla Public License Version
* 1.1 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
* for the specific language governing rights and limitations under the
* License.
*
* The Original Code is Mozilla.
*
* The Initial Developer of the Original Code is
* Netscape Communications.
* Portions created by the Initial Developer are Copyright (C) 2001
* the Initial Developer. All Rights Reserved.
*
* Contributor(s):
* Darin Fisher <darin@netscape.com> (original author)
*
* Alternatively, the contents of this file may be used under the terms of
* either the GNU General Public License Version 2 or later (the "GPL"), or
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
* in which case the provisions of the GPL or the LGPL are applicable instead
* of those above. If you wish to allow use of your version of this file only
* under the terms of either the GPL or the LGPL, and not to allow others to
* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
* ***** END LICENSE BLOCK ***** */
#include "nsHttpConnection.h"
#include "nsHttpTransaction.h"
#include "nsHttpRequestHead.h"
#include "nsHttpResponseHead.h"
#include "nsHttpHandler.h"
#include "nsIOService.h"
#include "nsISocketTransportService.h"
#include "nsISocketTransport.h"
#include "nsIServiceManager.h"
#include "nsISSLSocketControl.h"
#include "nsStringStream.h"
#include "netCore.h"
#include "nsNetCID.h"
#include "nsAutoLock.h"
#include "prmem.h"
#ifdef DEBUG
// defined by the socket transport service while active
extern PRThread *gSocketThread;
#endif
static NS_DEFINE_CID(kSocketTransportServiceCID, NS_SOCKETTRANSPORTSERVICE_CID);
// Statistics - only update on gSocketThread
// currently uncollected
static PRUint32 sCreateTransport1 = 0;
static PRUint32 sCreateTransport2 = 0;
static PRUint32 sSuccessTransport1 = 0;
static PRUint32 sSuccessTransport2 = 0;
static PRUint32 sUnNecessaryTransport2 = 0;
static PRUint32 sWastedReuseCount = 0;
//-----------------------------------------------------------------------------
// nsHttpConnection <public>
//-----------------------------------------------------------------------------
nsHttpConnection::nsHttpConnection()
: mTransaction(nsnull)
, mConnInfo(nsnull)
, mLock(nsnull)
, mLastReadTime(0)
, mIdleTimeout(0)
, mKeepAlive(PR_TRUE) // assume to keep-alive by default
, mKeepAliveMask(PR_TRUE)
, mSupportsPipelining(PR_FALSE) // assume low-grade server
, mIsReused(PR_FALSE)
, mCompletedSSLConnect(PR_FALSE)
, mActivationCount(0)
{
LOG(("Creating nsHttpConnection @%x\n", this));
// grab a reference to the handler to ensure that it doesn't go away.
nsHttpHandler *handler = gHttpHandler;
NS_ADDREF(handler);
}
nsHttpConnection::~nsHttpConnection()
{
LOG(("Destroying nsHttpConnection @%x\n", this));
if (mIdleSynTimer) {
mIdleSynTimer->Cancel();
mIdleSynTimer = nsnull;
}
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 05:50:36 -08:00
if (mBackupConnection) {
gHttpHandler->ReclaimConnection(mBackupConnection);
mBackupConnection = nsnull;
}
NS_IF_RELEASE(mConnInfo);
NS_IF_RELEASE(mTransaction);
if (mLock) {
PR_DestroyLock(mLock);
mLock = nsnull;
}
// release our reference to the handler
nsHttpHandler *handler = gHttpHandler;
NS_RELEASE(handler);
}
nsresult
nsHttpConnection::Init(nsHttpConnectionInfo *info, PRUint16 maxHangTime)
{
LOG(("nsHttpConnection::Init [this=%x]\n", this));
NS_ENSURE_ARG_POINTER(info);
NS_ENSURE_TRUE(!mConnInfo, NS_ERROR_ALREADY_INITIALIZED);
mLock = PR_NewLock();
if (!mLock)
return NS_ERROR_OUT_OF_MEMORY;
mConnInfo = info;
NS_ADDREF(mConnInfo);
mMaxHangTime = maxHangTime;
mLastReadTime = NowInSeconds();
return NS_OK;
}
void
nsHttpConnection::IdleSynTimeout(nsITimer *timer, void *closure)
{
// nsITimer is guaranteed to execute timer on same thread it
// was initialized on
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
nsHttpConnection *self = (nsHttpConnection *)closure;
NS_ABORT_IF_FALSE(timer == self->mIdleSynTimer, "wrong timer");
self->mIdleSynTimer = nsnull;
if (!self->mSocketTransport) {
NS_ABORT_IF_FALSE(self->mSocketTransport1 && !self->mSocketTransport2,
"establishing backup tranport");
LOG(("SocketTransport hit idle timer - starting backup socket"));
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 05:50:36 -08:00
gHttpHandler->ConnMgr()->GetConnection(self->mConnInfo,
self->mSocketCaps,
getter_AddRefs(
self->mBackupConnection));
if (!self->mBackupConnection)
return;
nsresult rv =
self->CreateTransport(self->mSocketCaps,
getter_AddRefs(self->mSocketTransport2),
getter_AddRefs(self->mSocketIn2),
getter_AddRefs(self->mSocketOut2));
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 05:50:36 -08:00
if (NS_SUCCEEDED(rv)) {
sCreateTransport2++;
self->mSocketOut2->AsyncWait(self, 0, 0, nsnull);
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 05:50:36 -08:00
}
}
return;
}
// called on the socket thread
nsresult
nsHttpConnection::Activate(nsAHttpTransaction *trans, PRUint8 caps)
{
nsresult rv;
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
LOG(("nsHttpConnection::Activate [this=%x trans=%x caps=%x]\n",
this, trans, caps));
NS_ENSURE_ARG_POINTER(trans);
NS_ENSURE_TRUE(!mTransaction, NS_ERROR_IN_PROGRESS);
// take ownership of the transaction
mTransaction = trans;
NS_ADDREF(mTransaction);
mActivationCount++;
// set mKeepAlive according to what will be requested
mKeepAliveMask = mKeepAlive = (caps & NS_HTTP_ALLOW_KEEPALIVE);
// need to handle SSL proxy CONNECT if this is the first time.
if (mConnInfo->UsingSSL() && mConnInfo->UsingHttpProxy() && !mCompletedSSLConnect) {
rv = SetupSSLProxyConnect();
if (NS_FAILED(rv))
goto failed_activation;
}
// if we don't have a socket transport then create a new one
if (!mSocketTransport) {
rv = CreateTransport(caps);
}
else {
NS_ABORT_IF_FALSE(mSocketOut && mSocketIn,
"Socket Transport and SocketOut mismatch");
// If this is the first transaction on this connection, but
// we already have a socket that means the socket was created
// speculatively in the past, not used at that time, and
// given to the connection manager.
if (mActivationCount == 1)
sWastedReuseCount++;
rv = mSocketOut->AsyncWait(this, 0, 0, nsnull);
}
failed_activation:
if (NS_FAILED(rv))
NS_RELEASE(mTransaction);
return rv;
}
void
nsHttpConnection::Close(nsresult reason)
{
LOG(("nsHttpConnection::Close [this=%x reason=%x]\n", this, reason));
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
if (NS_FAILED(reason)) {
if (mSocketTransport) {
mSocketTransport->SetSecurityCallbacks(nsnull);
mSocketTransport->SetEventSink(nsnull, nsnull);
mSocketTransport->Close(reason);
}
if (mSocketTransport1) {
mSocketTransport1->SetSecurityCallbacks(nsnull);
mSocketTransport1->SetEventSink(nsnull, nsnull);
mSocketTransport1->Close(reason);
}
if (mSocketTransport2) {
mSocketTransport2->SetSecurityCallbacks(nsnull);
mSocketTransport2->SetEventSink(nsnull, nsnull);
mSocketTransport2->Close(reason);
}
mKeepAlive = PR_FALSE;
}
}
// called on the socket thread
nsresult
nsHttpConnection::ProxyStartSSL()
{
LOG(("nsHttpConnection::ProxyStartSSL [this=%x]\n", this));
#ifdef DEBUG
NS_PRECONDITION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
#endif
nsCOMPtr<nsISupports> securityInfo;
nsresult rv = mSocketTransport->GetSecurityInfo(getter_AddRefs(securityInfo));
if (NS_FAILED(rv)) return rv;
nsCOMPtr<nsISSLSocketControl> ssl = do_QueryInterface(securityInfo, &rv);
if (NS_FAILED(rv)) return rv;
return ssl->ProxyStartSSL();
}
PRBool
nsHttpConnection::CanReuse()
{
return IsKeepAlive() && (NowInSeconds() - mLastReadTime < mIdleTimeout)
&& IsAlive();
}
PRUint32 nsHttpConnection::TimeToLive()
{
PRInt32 tmp = mIdleTimeout - (NowInSeconds() - mLastReadTime);
if (0 > tmp)
tmp = 0;
return tmp;
}
PRBool
nsHttpConnection::IsAlive()
{
if (!mSocketTransport)
return PR_FALSE;
PRBool alive;
nsresult rv = mSocketTransport->IsAlive(&alive);
if (NS_FAILED(rv))
alive = PR_FALSE;
//#define TEST_RESTART_LOGIC
#ifdef TEST_RESTART_LOGIC
if (!alive) {
LOG(("pretending socket is still alive to test restart logic\n"));
alive = PR_TRUE;
}
#endif
return alive;
}
PRBool
nsHttpConnection::SupportsPipelining(nsHttpResponseHead *responseHead)
{
// XXX there should be a strict mode available that disables this
// blacklisting.
// assuming connection is HTTP/1.1 with keep-alive enabled
if (mConnInfo->UsingHttpProxy() && !mConnInfo->UsingSSL()) {
// XXX check for bad proxy servers...
return PR_TRUE;
}
// XXX what about checking for a Via header? (transparent proxies)
// check for bad origin servers
const char *val = responseHead->PeekHeader(nsHttp::Server);
if (!val)
return PR_FALSE; // no header, no love
// The blacklist is indexed by the first character. All of these servers are
// known to return their identifier as the first thing in the server string,
// so we can do a leading match.
static const char *bad_servers[26][5] = {
{ nsnull }, { nsnull }, { nsnull }, { nsnull }, // a - d
{ "EFAServer/", nsnull }, // e
{ nsnull }, { nsnull }, { nsnull }, { nsnull }, // f - i
{ nsnull }, { nsnull }, { nsnull }, // j - l
{ "Microsoft-IIS/4.", "Microsoft-IIS/5.", nsnull }, // m
{ "Netscape-Enterprise/3.", "Netscape-Enterprise/4.",
"Netscape-Enterprise/5.", "Netscape-Enterprise/6.", nsnull }, // n
{ nsnull }, { nsnull }, { nsnull }, { nsnull }, // o - r
{ nsnull }, { nsnull }, { nsnull }, { nsnull }, // s - v
{ "WebLogic 3.", "WebLogic 4.","WebLogic 5.", "WebLogic 6.", nsnull }, // w
{ nsnull }, { nsnull }, { nsnull } // x - z
};
int index = val[0] - 'A'; // the whole table begins with capital letters
if ((index >= 0) && (index <= 25))
{
for (int i = 0; bad_servers[index][i] != nsnull; i++) {
if (!PL_strncmp (val, bad_servers[index][i], strlen (bad_servers[index][i]))) {
LOG(("looks like this server does not support pipelining"));
return PR_FALSE;
}
}
}
// ok, let's allow pipelining to this server
return PR_TRUE;
}
//----------------------------------------------------------------------------
// nsHttpConnection::nsAHttpConnection compatible methods
//----------------------------------------------------------------------------
nsresult
nsHttpConnection::OnHeadersAvailable(nsAHttpTransaction *trans,
nsHttpRequestHead *requestHead,
nsHttpResponseHead *responseHead,
PRBool *reset)
{
LOG(("nsHttpConnection::OnHeadersAvailable [this=%p trans=%p response-head=%p]\n",
this, trans, responseHead));
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
NS_ENSURE_ARG_POINTER(trans);
NS_ASSERTION(responseHead, "No response head?");
// If the server issued an explicit timeout, then we need to close down the
// socket transport. We pass an error code of NS_ERROR_NET_RESET to
// trigger the transactions 'restart' mechanism. We tell it to reset its
// response headers so that it will be ready to receive the new response.
if (responseHead->Status() == 408) {
Close(NS_ERROR_NET_RESET);
*reset = PR_TRUE;
return NS_OK;
}
// we won't change our keep-alive policy unless the server has explicitly
// told us to do so.
// inspect the connection headers for keep-alive info provided the
// transaction completed successfully.
const char *val = responseHead->PeekHeader(nsHttp::Connection);
if (!val)
val = responseHead->PeekHeader(nsHttp::Proxy_Connection);
// reset to default (the server may have changed since we last checked)
mSupportsPipelining = PR_FALSE;
// Ignore response to CONNECT from SSL proxy, we need
// version of the target server.
if (!mSSLProxyConnectStream) {
if ((responseHead->Version() > NS_HTTP_VERSION_0_9) &&
(requestHead->Version() > NS_HTTP_VERSION_0_9))
{
mConnInfo->DisallowHttp09();
}
}
if ((responseHead->Version() < NS_HTTP_VERSION_1_1) ||
(requestHead->Version() < NS_HTTP_VERSION_1_1)) {
// HTTP/1.0 connections are by default NOT persistent
if (val && !PL_strcasecmp(val, "keep-alive"))
mKeepAlive = PR_TRUE;
else
mKeepAlive = PR_FALSE;
}
else {
// HTTP/1.1 connections are by default persistent
if (val && !PL_strcasecmp(val, "close"))
mKeepAlive = PR_FALSE;
else {
mKeepAlive = PR_TRUE;
// Do not support pipelining when we are establishing
// an SSL tunnel though an HTTP proxy. Pipelining support
// determination must be based on comunication with the
// target server in this case. See bug 422016 for futher
// details.
if (!mSSLProxyConnectStream)
mSupportsPipelining = SupportsPipelining(responseHead);
}
}
mKeepAliveMask = mKeepAlive;
// if this connection is persistent, then the server may send a "Keep-Alive"
// header specifying the maximum number of times the connection can be
// reused as well as the maximum amount of time the connection can be idle
// before the server will close it. we ignore the max reuse count, because
// a "keep-alive" connection is by definition capable of being reused, and
// we only care about being able to reuse it once. if a timeout is not
// specified then we use our advertized timeout value.
if (mKeepAlive) {
val = responseHead->PeekHeader(nsHttp::Keep_Alive);
const char *cp = PL_strcasestr(val, "timeout=");
if (cp)
mIdleTimeout = (PRUint32) atoi(cp + 8);
else
mIdleTimeout = gHttpHandler->IdleTimeout();
LOG(("Connection can be reused [this=%x idle-timeout=%u]\n", this, mIdleTimeout));
}
// if we're doing an SSL proxy connect, then we need to check whether or not
// the connect was successful. if so, then we have to reset the transaction
// and step-up the socket connection to SSL. finally, we have to wake up the
// socket write request.
if (mSSLProxyConnectStream) {
mSSLProxyConnectStream = 0;
if (responseHead->Status() == 200) {
LOG(("SSL proxy CONNECT succeeded!\n"));
*reset = PR_TRUE;
nsresult rv = ProxyStartSSL();
if (NS_FAILED(rv)) // XXX need to handle this for real
LOG(("ProxyStartSSL failed [rv=%x]\n", rv));
mCompletedSSLConnect = PR_TRUE;
rv = mSocketOut->AsyncWait(this, 0, 0, nsnull);
// XXX what if this fails -- need to handle this error
NS_ASSERTION(NS_SUCCEEDED(rv), "mSocketOut->AsyncWait failed");
}
else {
LOG(("SSL proxy CONNECT failed!\n"));
// NOTE: this cast is valid since this connection cannot be
// processing a transaction pipeline until after the first HTTP/1.1
// response.
nsHttpTransaction *trans =
static_cast<nsHttpTransaction *>(mTransaction);
trans->SetSSLConnectFailed();
}
}
return NS_OK;
}
void
nsHttpConnection::GetSecurityInfo(nsISupports **secinfo)
{
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
if (mSocketTransport) {
if (NS_FAILED(mSocketTransport->GetSecurityInfo(secinfo)))
*secinfo = nsnull;
}
}
nsresult
nsHttpConnection::ResumeSend()
{
LOG(("nsHttpConnection::ResumeSend [this=%p]\n", this));
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
if (mSocketOut)
return mSocketOut->AsyncWait(this, 0, 0, nsnull);
NS_NOTREACHED("no socket output stream");
return NS_ERROR_UNEXPECTED;
}
nsresult
nsHttpConnection::ResumeRecv()
{
LOG(("nsHttpConnection::ResumeRecv [this=%p]\n", this));
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
if (mSocketIn)
return mSocketIn->AsyncWait(this, 0, 0, nsnull);
NS_NOTREACHED("no socket input stream");
return NS_ERROR_UNEXPECTED;
}
//-----------------------------------------------------------------------------
// nsHttpConnection <private>
//-----------------------------------------------------------------------------
nsresult
nsHttpConnection::CreateTransport(PRUint8 caps)
{
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
NS_ABORT_IF_FALSE(!mSocketTransport, "unexpected");
nsresult rv;
mSocketCaps = caps;
sCreateTransport1++;
PRUint16 timeout = gHttpHandler->GetIdleSynTimeout();
if (timeout) {
// Setup the timer that will establish a backup socket
// if we do not get a writable event on the main one.
// We do this because a lost SYN takes a very long time
// to repair at the TCP level.
//
// Failure to setup the timer is something we can live with,
// so don't return an error in that case.
mIdleSynTimer = do_CreateInstance(NS_TIMER_CONTRACTID, &rv);
if (NS_SUCCEEDED(rv))
mIdleSynTimer->InitWithFuncCallback(IdleSynTimeout, this,
timeout,
nsITimer::TYPE_ONE_SHOT);
}
rv = CreateTransport(mSocketCaps,
getter_AddRefs(mSocketTransport1),
getter_AddRefs(mSocketIn1),
getter_AddRefs(mSocketOut1));
if (NS_FAILED(rv))
return rv;
// wait for the output stream to be readable or timeout to occur
return mSocketOut1->AsyncWait(this, 0, 0, nsnull);
}
nsresult
nsHttpConnection::CreateTransport(PRUint8 caps,
nsISocketTransport **sock,
nsIAsyncInputStream **instream,
nsIAsyncOutputStream **outstream)
{
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
nsresult rv;
nsCOMPtr<nsISocketTransportService> sts =
do_GetService(NS_SOCKETTRANSPORTSERVICE_CONTRACTID, &rv);
if (NS_FAILED(rv)) return rv;
// configure the socket type based on the connection type requested.
const char* types[1];
if (mConnInfo->UsingSSL())
types[0] = "ssl";
else
types[0] = gHttpHandler->DefaultSocketType();
nsCOMPtr<nsISocketTransport> strans;
PRUint32 typeCount = (types[0] != nsnull);
rv = sts->CreateTransport(types, typeCount,
nsDependentCString(mConnInfo->Host()),
mConnInfo->Port(),
mConnInfo->ProxyInfo(),
getter_AddRefs(strans));
if (NS_FAILED(rv)) return rv;
PRUint32 tmpFlags = 0;
if (caps & NS_HTTP_REFRESH_DNS)
tmpFlags = nsISocketTransport::BYPASS_CACHE;
if (caps & NS_HTTP_LOAD_ANONYMOUS)
tmpFlags |= nsISocketTransport::ANONYMOUS_CONNECT;
strans->SetConnectionFlags(tmpFlags);
strans->SetQoSBits(gHttpHandler->GetQoSBits());
// NOTE: these create cyclical references, which we break inside
// nsHttpConnection::Close
rv = strans->SetEventSink(this, nsnull);
if (NS_FAILED(rv)) return rv;
rv = strans->SetSecurityCallbacks(this);
if (NS_FAILED(rv)) return rv;
// next open the socket streams
nsCOMPtr<nsIOutputStream> sout;
rv = strans->OpenOutputStream(nsITransport::OPEN_UNBUFFERED, 0, 0,
getter_AddRefs(sout));
if (NS_FAILED(rv)) return rv;
nsCOMPtr<nsIInputStream> sin;
rv = strans->OpenInputStream(nsITransport::OPEN_UNBUFFERED, 0, 0,
getter_AddRefs(sin));
if (NS_FAILED(rv)) return rv;
strans.forget(sock);
CallQueryInterface(sin, instream);
CallQueryInterface(sout, outstream);
return NS_OK;
}
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 05:50:36 -08:00
nsresult
nsHttpConnection::AssignTransport(nsISocketTransport *sock,
nsIAsyncOutputStream *outs,
nsIAsyncInputStream *ins)
{
nsresult rv = sock->SetEventSink(this, nsnull);
NS_ENSURE_SUCCESS(rv, rv);
rv =sock->SetSecurityCallbacks(this);
NS_ENSURE_SUCCESS(rv, rv);
mSocketTransport = sock;
mSocketOut = outs;
mSocketIn = ins;
return NS_OK;
}
void
nsHttpConnection::CloseTransaction(nsAHttpTransaction *trans, nsresult reason)
{
LOG(("nsHttpConnection::CloseTransaction[this=%x trans=%x reason=%x]\n",
this, trans, reason));
NS_ASSERTION(trans == mTransaction, "wrong transaction");
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
// mask this error code because its not a real error.
if (reason == NS_BASE_STREAM_CLOSED)
reason = NS_OK;
mTransaction->Close(reason);
NS_RELEASE(mTransaction);
mTransaction = 0;
if (NS_FAILED(reason))
Close(reason);
// flag the connection as reused here for convenience sake. certainly
// it might be going away instead ;-)
mIsReused = PR_TRUE;
}
NS_METHOD
nsHttpConnection::ReadFromStream(nsIInputStream *input,
void *closure,
const char *buf,
PRUint32 offset,
PRUint32 count,
PRUint32 *countRead)
{
// thunk for nsIInputStream instance
nsHttpConnection *conn = (nsHttpConnection *) closure;
return conn->OnReadSegment(buf, count, countRead);
}
nsresult
nsHttpConnection::OnReadSegment(const char *buf,
PRUint32 count,
PRUint32 *countRead)
{
if (count == 0) {
// some ReadSegments implementations will erroneously call the writer
// to consume 0 bytes worth of data. we must protect against this case
// or else we'd end up closing the socket prematurely.
NS_ERROR("bad ReadSegments implementation");
return NS_ERROR_FAILURE; // stop iterating
}
nsresult rv = mSocketOut->Write(buf, count, countRead);
if (NS_FAILED(rv))
mSocketOutCondition = rv;
else if (*countRead == 0)
mSocketOutCondition = NS_BASE_STREAM_CLOSED;
else
mSocketOutCondition = NS_OK; // reset condition
return mSocketOutCondition;
}
nsresult
nsHttpConnection::OnSocketWritable()
{
LOG(("nsHttpConnection::OnSocketWritable [this=%x]\n", this));
nsresult rv;
PRUint32 n;
PRBool again = PR_TRUE;
do {
// if we're doing an SSL proxy connect, then we need to bypass calling
// into the transaction.
//
// NOTE: this code path can't be shared since the transaction doesn't
// implement nsIInputStream. doing so is not worth the added cost of
// extra indirections during normal reading.
//
if (mSSLProxyConnectStream) {
LOG((" writing CONNECT request stream\n"));
rv = mSSLProxyConnectStream->ReadSegments(ReadFromStream, this,
nsIOService::gDefaultSegmentSize,
&n);
}
else {
LOG((" writing transaction request stream\n"));
rv = mTransaction->ReadSegments(this, nsIOService::gDefaultSegmentSize, &n);
}
LOG((" ReadSegments returned [rv=%x read=%u sock-cond=%x]\n",
rv, n, mSocketOutCondition));
// XXX some streams return NS_BASE_STREAM_CLOSED to indicate EOF.
if (rv == NS_BASE_STREAM_CLOSED) {
rv = NS_OK;
n = 0;
}
if (NS_FAILED(rv)) {
// if the transaction didn't want to write any more data, then
// wait for the transaction to call ResumeSend.
if (rv == NS_BASE_STREAM_WOULD_BLOCK)
rv = NS_OK;
again = PR_FALSE;
}
else if (NS_FAILED(mSocketOutCondition)) {
if (mSocketOutCondition == NS_BASE_STREAM_WOULD_BLOCK)
rv = mSocketOut->AsyncWait(this, 0, 0, nsnull); // continue writing
else
rv = mSocketOutCondition;
again = PR_FALSE;
}
else if (n == 0) {
//
// at this point we've written out the entire transaction, and now we
// must wait for the server's response. we manufacture a status message
// here to reflect the fact that we are waiting. this message will be
// trumped (overwritten) if the server responds quickly.
//
mTransaction->OnTransportStatus(nsISocketTransport::STATUS_WAITING_FOR,
LL_ZERO);
rv = mSocketIn->AsyncWait(this, 0, 0, nsnull); // start reading
again = PR_FALSE;
}
// write more to the socket until error or end-of-request...
} while (again);
return rv;
}
nsresult
nsHttpConnection::OnWriteSegment(char *buf,
PRUint32 count,
PRUint32 *countWritten)
{
if (count == 0) {
// some WriteSegments implementations will erroneously call the reader
// to provide 0 bytes worth of data. we must protect against this case
// or else we'd end up closing the socket prematurely.
NS_ERROR("bad WriteSegments implementation");
return NS_ERROR_FAILURE; // stop iterating
}
nsresult rv = mSocketIn->Read(buf, count, countWritten);
if (NS_FAILED(rv))
mSocketInCondition = rv;
else if (*countWritten == 0)
mSocketInCondition = NS_BASE_STREAM_CLOSED;
else
mSocketInCondition = NS_OK; // reset condition
return mSocketInCondition;
}
nsresult
nsHttpConnection::OnSocketReadable()
{
LOG(("nsHttpConnection::OnSocketReadable [this=%x]\n", this));
PRUint32 now = NowInSeconds();
if (mKeepAliveMask && (now - mLastReadTime >= PRUint32(mMaxHangTime))) {
LOG(("max hang time exceeded!\n"));
// give the handler a chance to create a new persistent connection to
// this host if we've been busy for too long.
mKeepAliveMask = PR_FALSE;
gHttpHandler->ProcessPendingQ(mConnInfo);
}
mLastReadTime = now;
nsresult rv;
PRUint32 n;
PRBool again = PR_TRUE;
do {
rv = mTransaction->WriteSegments(this, nsIOService::gDefaultSegmentSize, &n);
if (NS_FAILED(rv)) {
// if the transaction didn't want to take any more data, then
// wait for the transaction to call ResumeRecv.
if (rv == NS_BASE_STREAM_WOULD_BLOCK)
rv = NS_OK;
again = PR_FALSE;
}
else if (NS_FAILED(mSocketInCondition)) {
// continue waiting for the socket if necessary...
if (mSocketInCondition == NS_BASE_STREAM_WOULD_BLOCK)
rv = mSocketIn->AsyncWait(this, 0, 0, nsnull);
else
rv = mSocketInCondition;
again = PR_FALSE;
}
// read more from the socket until error...
} while (again);
return rv;
}
nsresult
nsHttpConnection::SetupSSLProxyConnect()
{
const char *val;
LOG(("nsHttpConnection::SetupSSLProxyConnect [this=%x]\n", this));
NS_ENSURE_TRUE(!mSSLProxyConnectStream, NS_ERROR_ALREADY_INITIALIZED);
nsCAutoString buf;
nsresult rv = nsHttpHandler::GenerateHostPort(
nsDependentCString(mConnInfo->Host()), mConnInfo->Port(), buf);
if (NS_FAILED(rv))
return rv;
// CONNECT host:port HTTP/1.1
nsHttpRequestHead request;
request.SetMethod(nsHttp::Connect);
request.SetVersion(gHttpHandler->HttpVersion());
request.SetRequestURI(buf);
request.SetHeader(nsHttp::User_Agent, gHttpHandler->UserAgent());
// send this header for backwards compatibility.
request.SetHeader(nsHttp::Proxy_Connection, NS_LITERAL_CSTRING("keep-alive"));
// NOTE: this cast is valid since this connection cannot be processing a
// transaction pipeline until after the first HTTP/1.1 response.
nsHttpTransaction *trans = static_cast<nsHttpTransaction *>(mTransaction);
val = trans->RequestHead()->PeekHeader(nsHttp::Host);
if (val) {
// all HTTP/1.1 requests must include a Host header (even though it
// may seem redundant in this case; see bug 82388).
request.SetHeader(nsHttp::Host, nsDependentCString(val));
}
val = trans->RequestHead()->PeekHeader(nsHttp::Proxy_Authorization);
if (val) {
// we don't know for sure if this authorization is intended for the
// SSL proxy, so we add it just in case.
request.SetHeader(nsHttp::Proxy_Authorization, nsDependentCString(val));
}
buf.Truncate();
request.Flatten(buf, PR_FALSE);
buf.AppendLiteral("\r\n");
return NS_NewCStringInputStream(getter_AddRefs(mSSLProxyConnectStream), buf);
}
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 05:50:36 -08:00
void
nsHttpConnection::ReleaseBackupTransport(nsISocketTransport *sock,
nsIAsyncOutputStream *outs,
nsIAsyncInputStream *ins)
{
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 05:50:36 -08:00
// We need to establish a small non-zero idle timeout so the connection
// mgr perceives this socket as suitable for persistent connection reuse
NS_ABORT_IF_FALSE(sock && outs && ins, "release Backup precond");
mBackupConnection->mIdleTimeout = NS_MIN((PRUint16) 5,
gHttpHandler->IdleTimeout());
mBackupConnection->mIsReused = PR_TRUE;
nsresult rv = mBackupConnection->AssignTransport(sock, outs, ins);
if (NS_SUCCEEDED(rv))
rv = gHttpHandler->ReclaimConnection(mBackupConnection);
if (NS_FAILED(rv))
NS_WARNING("Backup nsHttpConnection could not be reclaimed");
mBackupConnection = nsnull;
}
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 05:50:36 -08:00
void
nsHttpConnection::SelectPrimaryTransport(nsIAsyncOutputStream *out)
{
if (!mSocketOut) {
// Setup the Main Socket
if (mIdleSynTimer) {
mIdleSynTimer->Cancel();
mIdleSynTimer = nsnull;
}
if (out == mSocketOut1) {
sSuccessTransport1++;
mSocketTransport.swap(mSocketTransport1);
mSocketOut.swap(mSocketOut1);
mSocketIn.swap(mSocketIn1);
if (mSocketTransport2)
sUnNecessaryTransport2++;
}
else if (out == mSocketOut2) {
NS_ABORT_IF_FALSE(mSocketOut1,
"backup socket without primary being tested");
sSuccessTransport2++;
mSocketTransport.swap(mSocketTransport2);
mSocketOut.swap(mSocketOut2);
mSocketIn.swap(mSocketIn2);
}
else {
NS_ABORT_IF_FALSE(0, "setup on unexpected socket");
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 05:50:36 -08:00
return;
}
}
else if (out == mSocketOut1) {
// Socket2 became the primary socket but Socket1 is now valid - give it
// to the connection manager
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 05:50:36 -08:00
ReleaseBackupTransport(mSocketTransport1,
mSocketOut1,
mSocketIn1);
sSuccessTransport1++;
mSocketTransport1 = nsnull;
mSocketOut1 = nsnull;
mSocketIn1 = nsnull;
}
else if (out == mSocketOut2) {
// Socket1 became the primary socket but Socket2 is now valid - give it
// to the connectionmanager
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 05:50:36 -08:00
ReleaseBackupTransport(mSocketTransport2,
mSocketOut2,
mSocketIn2);
sSuccessTransport2++;
mSocketTransport2 = nsnull;
mSocketOut2 = nsnull;
mSocketIn2 = nsnull;
}
}
//-----------------------------------------------------------------------------
// nsHttpConnection::nsISupports
//-----------------------------------------------------------------------------
NS_IMPL_THREADSAFE_ISUPPORTS4(nsHttpConnection,
nsIInputStreamCallback,
nsIOutputStreamCallback,
nsITransportEventSink,
nsIInterfaceRequestor)
//-----------------------------------------------------------------------------
// nsHttpConnection::nsIInputStreamCallback
//-----------------------------------------------------------------------------
// called on the socket transport thread
NS_IMETHODIMP
nsHttpConnection::OnInputStreamReady(nsIAsyncInputStream *in)
{
NS_ASSERTION(in == mSocketIn, "unexpected stream");
NS_ASSERTION(PR_GetCurrentThread() == gSocketThread, "wrong thread");
// if the transaction was dropped...
if (!mTransaction) {
LOG((" no transaction; ignoring event\n"));
return NS_OK;
}
nsresult rv = OnSocketReadable();
if (NS_FAILED(rv))
CloseTransaction(mTransaction, rv);
return NS_OK;
}
//-----------------------------------------------------------------------------
// nsHttpConnection::nsIOutputStreamCallback
//-----------------------------------------------------------------------------
NS_IMETHODIMP
nsHttpConnection::OnOutputStreamReady(nsIAsyncOutputStream *out)
{
NS_ABORT_IF_FALSE(PR_GetCurrentThread() == gSocketThread, "wrong thread");
// if the transaction was dropped...
if (!mTransaction) {
LOG((" no transaction; ignoring event\n"));
return NS_OK;
}
NS_ABORT_IF_FALSE(out == mSocketOut ||
out == mSocketOut1 ||
out == mSocketOut2 , "unexpected socket");
if (out != mSocketOut)
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 05:50:36 -08:00
SelectPrimaryTransport(out);
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 05:50:36 -08:00
if (mSocketOut == out) {
NS_ABORT_IF_FALSE(!mIdleSynTimer,"IdleSynTimer should not be set");
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 05:50:36 -08:00
nsresult rv = OnSocketWritable();
if (NS_FAILED(rv))
CloseTransaction(mTransaction, rv);
}
return NS_OK;
}
//-----------------------------------------------------------------------------
// nsHttpConnection::nsITransportEventSink
//-----------------------------------------------------------------------------
NS_IMETHODIMP
nsHttpConnection::OnTransportStatus(nsITransport *trans,
nsresult status,
PRUint64 progress,
PRUint64 progressMax)
{
if (mTransaction)
mTransaction->OnTransportStatus(status, progress);
return NS_OK;
}
//-----------------------------------------------------------------------------
// nsHttpConnection::nsIInterfaceRequestor
//-----------------------------------------------------------------------------
// not called on the socket transport thread
NS_IMETHODIMP
nsHttpConnection::GetInterface(const nsIID &iid, void **result)
{
// NOTE: This function is only called on the UI thread via sync proxy from
// the socket transport thread. If that weren't the case, then we'd
// have to worry about the possibility of mTransaction going away
// part-way through this function call. See CloseTransaction.
NS_ASSERTION(PR_GetCurrentThread() != gSocketThread, "wrong thread");
if (mTransaction) {
nsCOMPtr<nsIInterfaceRequestor> callbacks;
mTransaction->GetSecurityCallbacks(getter_AddRefs(callbacks));
if (callbacks)
return callbacks->GetInterface(iid, result);
}
return NS_ERROR_NO_INTERFACE;
}