http://bugs.winehq.org/show_bug.cgi?id=20622
Summary: chromium's net_unittests.exe hangs in SSLClientSocketTest.Read Product: Wine Version: 1.1.32 Platform: PC OS/Version: Linux Status: NEW Keywords: download, source Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs@winehq.org ReportedBy: dank@kegel.com
Chromium's SSLClientSocketTest.Read test outputs (ignoring repeated fixmes)
[ RUN ] SSLClientSocketTest.Read fixme:crypt:CRYPT_CriticalExtensionsSupported unsupported critical extension "2.5.29.32" fixme:threadpool:RtlQueueWorkItem Flags 0x4 not supported fixme:secur32:schan_InitializeSecurityContextW Using hardcoded "NORMAL" priority
and then seems to hang. Five other SSL-related tests behave similarly: SSLClientSocketTest.Read_SmallChunks SSLClientSocketTest.Read_Interrupted HTTPSRequestTest.HTTPSExpiredTest HTTPSRequestTest.HTTPSGetTest HTTPSRequestTest.HTTPSMismatchedTest
To repeat: $ wget http://kegel.com/wine/chromium/chromium-tests.tar.bz2 $ tar -xjvf chromium-tests.tar.bz2 $ sudo cp src/net/data/ssl/certificates/root_ca_cert.crt /usr/share/ca-certificates/ $ sudo vi /etc/ca-certificates.conf (and add the line root_ca_cert.crt) $ sudo update-ca-certificates $ # (or run juan's importer, http://bugs.winehq.org/show_bug.cgi?id=20370#c4 ) $ wine src/chrome/Debug/net_unittests.exe --gtest_filter=SSLClientSocketTest.Read
I'll attach a +relay,+secur32,+crypt,+seh log.
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #1 from Dan Kegel dank@kegel.com 2009-11-08 12:41:12 --- Created an attachment (id=24596) --> (http://bugs.winehq.org/attachment.cgi?id=24596) Unedited, rzipped log of test run
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #2 from Juan Lang juan_lang@yahoo.com 2009-11-09 11:42:44 --- At first glance, this doesn't appear to have anything to do with crypto. secur32 doesn't really appear in the log. The crypt fixme appears when the app is reading the root stores, basically at startup. A winsock log might be a little more informative.
The test creates a socket to localhost:9443, then should be using schannel to create buffers to be sent and received. It never gets there, though. So I expect the error has more to do with either winsock, or with synchronization functions in general.
A more basic question: where does the server come from? I see a bunch of client code in the tests, but I don't see any code for the server. Is this expected to be started externally?
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #3 from Dan Kegel dank@kegel.com 2009-11-09 11:54:33 --- The server is written in python, is included in the test tarball, and is started by the test.
You could probably dummy that out and start the original python script with a linux python interpreter ahead of time.
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #4 from Juan Lang juan_lang@yahoo.com 2009-11-09 12:37:14 --- Ah, I didn't find that till you mentioned it. Thanks. For other readers: it's in src/third_party/tlslite/test/httpsserver.py.
I can't run the tests easily here, as the gnutls on this machine (FC10) doesn't work with schannel, so it fails for me with: [8:9:1109/103044:1032:ERROR:ssl_client_socket_win.cc(216)] AcquireCredentialsHandle failed: -2146893051 [8:9:1109/103044:1033:ERROR:ssl_client_socket_win.cc(475)] InitializeSecurityContext failed: -2146893055 .\socket\ssl_client_socket_unittest.cc(200): error: Value of: rv Actual: -9 Expected: net::ERR_IO_PENDING Which is: -1
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #5 from Henri Verbeet hverbeet@gmail.com 2009-11-09 12:44:58 --- I guess this might be a gnutls/schannel issue. I don't personally have time to look at this right now though.
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #6 from Juan Lang juan_lang@yahoo.com 2009-11-09 13:12:56 --- Mine's a gnutls/schannel, but Dan's might not be.
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #7 from Hans Leidekker hans@meelstraat.net 2009-11-18 07:04:46 --- Created an attachment (id=24819) --> (http://bugs.winehq.org/attachment.cgi?id=24819) secur32: Handle EOF from gnutls_record_recv.
This patch fixes the hang here.
http://bugs.winehq.org/show_bug.cgi?id=20622
--- Comment #8 from Dan Kegel dank@kegel.com 2009-11-18 07:33:45 --- Here, too. Thanks!
I'll file more bugs for the remaining ssl problems.
http://bugs.winehq.org/show_bug.cgi?id=20622
Hans Leidekker hans@meelstraat.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED CC| |hans@meelstraat.net Resolution| |FIXED
--- Comment #9 from Hans Leidekker hans@meelstraat.net 2009-11-19 13:25:15 --- Fixed by commit aeb161f13f5119d5647a5cb8775cf13719444e7e.
http://bugs.winehq.org/show_bug.cgi?id=20622
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #10 from Alexandre Julliard julliard@winehq.org 2009-12-04 12:16:25 --- Closing bugs fixed in 1.1.34.