http://bugs.winehq.org/show_bug.cgi?id=16453
Summary: Chrome outer installer can't download inner installer Product: Wine Version: 1.1.10 Platform: Other URL: http://chrome.google.com OS/Version: other Status: NEW Keywords: download, Installer, patch Severity: normal Priority: P2 Component: winhttp AssignedTo: wine-bugs@winehq.org ReportedBy: dank@kegel.com
Tried Chrome today by going to chrome.google.com in firefox under wine and downloading what they offered. The small (750KB) outer installer starts, but fails because WinHttpCrackUrl is too picky about its arguments. I removed the extra checks (see attached patch) and it got a fair bit further; the failure seems to be associated with the log message
err:winhttp:netconn_secure_connect SSL_connect failed: error:140770FC:SSL routines:SSL23_GET_SERVER_HELLO:unknown protocol
(which might mean all sorts of things, including we tried to connect to an http server with https...)
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #1 from Dan Kegel dank@kegel.com 2008-12-10 10:51:13 --- Created an attachment (id=17802) --> (http://bugs.winehq.org/attachment.cgi?id=17802) Patch to remove overly-strict checks on WinHttpCrackUrl
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #2 from Dan Kegel dank@kegel.com 2008-12-10 10:52:28 --- Oh, all this was after doing winetricks msxml3 firefox corefonts flash winxp as recommended by the appdb.
http://bugs.winehq.org/show_bug.cgi?id=16453
Hans Leidekker hans@meelstraat.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #17802|0 |1 is obsolete| |
--- Comment #3 from Hans Leidekker hans@meelstraat.net 2008-12-10 15:32:24 --- Created an attachment (id=17811) --> (http://bugs.winehq.org/attachment.cgi?id=17811) winhttp: Validate parameters per component in WinHttpCrackUrl.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #4 from Hans Leidekker hans@meelstraat.net 2008-12-10 15:33:21 --- Created an attachment (id=17812) --> (http://bugs.winehq.org/attachment.cgi?id=17812) winhttp: Resolve host port at the latest possible moment.
http://bugs.winehq.org/show_bug.cgi?id=16453
Hans Leidekker hans@meelstraat.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |hans@meelstraat.net
--- Comment #5 from Hans Leidekker hans@meelstraat.net 2008-12-10 15:36:45 --- These patches fix both reported issues but the installer still fails unfortunately.
http://bugs.winehq.org/show_bug.cgi?id=16453
Hans Leidekker hans@meelstraat.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Component|winhttp |msxml3
--- Comment #6 from Hans Leidekker hans@meelstraat.net 2008-12-11 07:36:14 --- Actually the failure is the familiar msxml3 bug; we're sending this:
<?xml version="1.0" encoding="UTF-8"?><?xml version="1.0"?>...
where we should send this:
<?xml version="1.0" encoding="UTF-8"?>...
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #7 from Hans Leidekker hans@meelstraat.net 2008-12-11 09:44:09 --- Another difference between native:
<o:gupdate xmlns:o="http://www.google.com/update2/request" protocol="2.0"...
and builtin:
<o:gupdate protocol="2.0"...
So the xmlns attribute is missing from builtin generated xml.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #8 from Hans Leidekker hans@meelstraat.net 2008-12-11 09:52:20 --- Created an attachment (id=17833) --> (http://bugs.winehq.org/attachment.cgi?id=17833) msxml3 hack: Remove xml preamble.
Patch to chop off the xml preamble that libxml2 automatically adds.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #9 from Dan Kegel dank@kegel.com 2008-12-12 08:56:47 --- FWIW, that patch isn't enough, I still get err:eventlog:ReportEventW L"Network Request Error.\r\nError: 0x80070057. Http st atus code: 0.\r\nUrl=https://tools.google.com/service/update2 .. (And 'make test' fails.)
I guess we also need a patch for the xmlns problem you spotted?
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #10 from Hans Leidekker hans@meelstraat.net 2008-12-12 09:29:07 ---
(And 'make test' fails.)
That's why I called it a hack, we can't unconditionally remove the preamble because it may have been added by the user or read in via loadXML.
I guess we also need a patch for the xmlns problem you spotted?
I took a shot at that too but this is getting pretty hairy. If you set the namespace attribute in libxml2 it adds it to all element nodes, whereas msxml adds it to the root element only. And there's difference in handling of namespace prefixes too.
We may be able to work around this by inserting the xmlns attribute ourselves, I'm not sure.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #11 from Hans Leidekker hans@meelstraat.net 2008-12-22 04:29:19 --- I'm starting to think we should write our own serialization code instead of fixing up the libxml2 generated stream.
http://bugs.winehq.org/show_bug.cgi?id=16453
Jon Dufresne jon.dufresne@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |jon.dufresne@gmail.com
http://bugs.winehq.org/show_bug.cgi?id=16453
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Chrome outer installer |Google Chrome outer |can't download inner |installer can't download |installer |inner installer
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #12 from Nikolay Sivov bunglehead@gmail.com 2010-01-18 18:13:32 --- (In reply to comment #11)
I'm starting to think we should write our own serialization code instead of fixing up the libxml2 generated stream.
Hi, Hans.
Could you summarize here all problems you spotted with saving tree to xml stream with libxml2? I suppose a duplicated but fixed xmlNodeDump logic needed here?
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #13 from Hans Leidekker hans@meelstraat.net 2010-01-19 02:08:55 ---
Could you summarize here all problems you spotted with saving tree to xml stream with libxml2?
See comment 6, 7 and 10 in this bug. In addition to that we're already fixing up white space, line endings (EnsureCorrectEOL) and encoding attribute (EnsureNoEncoding).
I suppose a duplicated but fixed xmlNodeDump logic needed here?
Something like that, yes.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #14 from Nikolay Sivov bunglehead@gmail.com 2010-02-16 15:35:06 --- Hi, Hans.
Regarding comment 6 it's not clear where second declaration comes from. Was another one added as PI node? libxml2 has a switch for saving context to skip declaration on file start. I agree that it's better to rewrite saving functionality, but at this point I need to know where another '<?xml..' comes from...
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #15 from Hans Leidekker hans@meelstraat.net 2010-02-16 15:46:21 ---
Regarding comment 6 it's not clear where second declaration comes from. Was
I believe it's added by the installer.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #16 from Nikolay Sivov bunglehead@gmail.com 2010-02-16 16:37:48 --- (In reply to comment #15)
Regarding comment 6 it's not clear where second declaration comes from. Was
I believe it's added by the installer.
So there's two problems with that: - builtin doesn't create PI node for xml declaration; - xml declaration always added by libxml2 saving logic.
As a solution we should always disable implicit declaration from libxml2 (which is trivial) and manually create additional node on document creation (which I'm not sure how to do in a clean way).
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #17 from Nikolay Sivov bunglehead@gmail.com 2010-02-16 17:07:14 --- It seems I found a clean enough workaround for that - we could reliably know that declaration was present in original document checking 'doc->standalone' or similar field for input parser context. If it is -1 then no declaration was present in xml stream. So if it's not -1 we could construct additional PI node with data already present in xmlDoc structure.
http://bugs.winehq.org/show_bug.cgi?id=16453
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |15190
http://bugs.winehq.org/show_bug.cgi?id=16453
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |15255
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #18 from Nikolay Sivov bunglehead@gmail.com 2010-09-15 15:18:06 CDT --- Please try this another time with current git. Extra <?xml node in output stream should be fixed for months now, and next bug about missed xmlns namespace attribute is probably fixed with 8d055c78547cc3cb14df485b6c3bd4771c504445.
If it still doesn't work attach +msxml running today git version.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #19 from Austin English austinenglish@gmail.com 2010-09-15 18:14:42 CDT --- (In reply to comment #18)
Please try this another time with current git. Extra <?xml node in output stream should be fixed for months now, and next bug about missed xmlns namespace attribute is probably fixed with 8d055c78547cc3cb14df485b6c3bd4771c504445.
If it still doesn't work attach +msxml running today git version.
Still fails. With native msxml3, however, it crashes for me:
[61:60:91699:FATAL:network_change_notifier_win.cc(36)] Check failed: ret == ERROR_IO_PENDING. wine: Unhandled exception 0x80000003 at address 0x1cf97c3 (thread 003c), starting debugger...
I'll attach a +msxml trace.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #20 from Austin English austinenglish@gmail.com 2010-09-15 18:16:17 CDT --- Created an attachment (id=30773) --> (http://bugs.winehq.org/attachment.cgi?id=30773) +msxml
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #21 from Juan Lang juan_lang@yahoo.com 2010-09-15 18:32:57 CDT --- (In reply to comment #19)
Still fails. With native msxml3, however, it crashes for me:
[61:60:91699:FATAL:network_change_notifier_win.cc(36)] Check failed: ret == ERROR_IO_PENDING. wine: Unhandled exception 0x80000003 at address 0x1cf97c3 (thread 003c), starting debugger...
That's probably this code from: http://src.chromium.org/viewvc/chrome/trunk/src/net/base/network_change_noti...
DWORD ret = NotifyAddrChange(&handle, &addr_overlapped_); CHECK(ret == ERROR_IO_PENDING);
A quick patch to iphlpapi should fix that, I'll attach in a sec.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #22 from Juan Lang juan_lang@yahoo.com 2010-09-15 18:34:22 CDT --- Created an attachment (id=30775) --> (http://bugs.winehq.org/attachment.cgi?id=30775) Patch: Return ERROR_IO_PENDING from NotifyAddrChange
This should apply with a little fuzz, I've been messing about with iphlpapi in my tree so the line numbers will be different with a clean tree.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #23 from Nikolay Sivov bunglehead@gmail.com 2010-10-02 10:55:04 CDT --- If it's still a msxml problem please attach current +msxml.
http://bugs.winehq.org/show_bug.cgi?id=16453
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #30773|0 |1 is obsolete| |
--- Comment #24 from Austin English austinenglish@gmail.com 2010-10-02 15:01:06 CDT --- Created an attachment (id=31073) --> (http://bugs.winehq.org/attachment.cgi?id=31073) +msxml
Still in 1.3.4.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #25 from Austin English austinenglish@gmail.com 2010-10-02 15:22:46 CDT --- (In reply to comment #22)
Created an attachment (id=30775)
--> (http://bugs.winehq.org/attachment.cgi?id=30775) [details]
Patch: Return ERROR_IO_PENDING from NotifyAddrChange
This should apply with a little fuzz, I've been messing about with iphlpapi in my tree so the line numbers will be different with a clean tree.
Gets rid of that crash, but still fails later (after native msxml3 to workaround the first bug). I'll attach a new log.
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #26 from Austin English austinenglish@gmail.com 2010-10-02 15:23:50 CDT --- Created an attachment (id=31075) --> (http://bugs.winehq.org/attachment.cgi?id=31075) output
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #27 from Nikolay Sivov bunglehead@gmail.com 2010-10-02 15:31:12 CDT --- Austin, could you split this? I mean move iphlpapi to a separate report and link them.
http://bugs.winehq.org/show_bug.cgi?id=16453
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |24607
http://bugs.winehq.org/show_bug.cgi?id=16453
--- Comment #28 from Austin English austinenglish@gmail.com 2010-10-02 23:09:35 CDT --- (In reply to comment #27)
Austin, could you split this? I mean move iphlpapi to a separate report and link them.
Wasn't my bug, but sure :-).
See bug 24607 for the install failure.
http://bugs.winehq.org/show_bug.cgi?id=16453
Luke Bratch l_bratch@yahoo.co.uk changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |l_bratch@yahoo.co.uk
http://bugs.winehq.org/show_bug.cgi?id=16453
Juan Lang juan_lang@yahoo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #30775|0 |1 is obsolete| |
--- Comment #29 from Juan Lang juan_lang@yahoo.com 2011-02-16 09:45:31 CST --- (From update of attachment 30775) Commit 82a2ec13f01b48e78d28d630914db311b55f4948 addresses this, so this patch is no longer needed.
http://bugs.winehq.org/show_bug.cgi?id=16453
fracting fracting@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |fracting@gmail.com
--- Comment #30 from fracting fracting@gmail.com 2011-04-06 23:54:15 CDT --- Add me to CC list
http://bugs.winehq.org/show_bug.cgi?id=16453
Adam Martinson amartinson@codeweavers.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |amartinson@codeweavers.com
--- Comment #31 from Adam Martinson amartinson@codeweavers.com 2011-05-20 17:07:22 CDT --- Should be fixed by f730d13c84507023cc18c97d391c8f84a4f8e431.
http://bugs.winehq.org/show_bug.cgi?id=16453
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED
--- Comment #32 from Austin English austinenglish@gmail.com 2011-05-21 11:53:58 CDT --- (In reply to comment #31)
Should be fixed by f730d13c84507023cc18c97d391c8f84a4f8e431.
http://www.google.com/chrome/index.html?platform=win&hl=en 6bed5e4b54000344da55a9b21be3db88f03913ee ChromeSetup.exe wine-1.3.20-230-g456e48e
works great!
http://bugs.winehq.org/show_bug.cgi?id=16453
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #33 from Alexandre Julliard julliard@winehq.org 2011-05-27 13:38:23 CDT --- Closing bugs fixed in 1.3.21.
-
wine-bugs＠winehq.org