msg237609 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 09:08 |
The PEP 475 has been accepted and is partialy implemented. The socket module is not fully patched to handle EINTR. For example, socket.socket.connect() doesn't handle EINTR yet. Attached patch fixes socket.connect() to handle EINTR. It should fix issue #11266 and #20611 in Python 3.5 (Python 2.7 and 3.4 will need to be patched to handle explicitly InterruptedError/OSError(EINTR) in Python). By the way, some socket functions handle EINTR but don't recompute the timeout yet. _PyTime_monotonic() should be used. |
|
|
msg237610 - (view) |
Author: Charles-François Natali (neologix) * |
Date: 2015-03-09 09:10 |
If EINTR is received during connect, the socket is unusable, that's why i didn't implement it. |
|
|
msg237616 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 09:20 |
> If EINTR is received during connect, the socket is unusable, > that's why i didn't implement it. Can you elaborate? socket.connect() should be modified according to the PEP 475: https://www.python.org/dev/peps/pep-0475/#modified-functions What do you mean by "unusable"? Is it possible to retry connect()? Is it safe to call getsockopt(fd, SOL_SOCKET, SO_ERROR) until it returns EISCONN? For high-level functions for socket.create_connection(), how should we handle InterruptedError? See also this change in asyncio to handle InterruptedError in asyncio: https://hg.python.org/cpython/rev/ad67f66a5f3c (I wrote it and I didn't test my change, I didn't know how to test it.) Tulip issue: https://code.google.com/p/tulip/issues/detail?id=205 |
|
|
msg237619 - (view) |
Author: Antoine Pitrou (pitrou) * |
Date: 2015-03-09 09:29 |
About EINTR and connect(), I've found the following insightful page: http://www.madore.org/~david/computers/connect-intr.html Official POSIX wording is this: """If connect() is interrupted by a signal that is caught while blocked waiting to establish a connection, connect() shall fail and set errno to [EINTR], but the connection request shall not be aborted, and the connection shall be established asynchronously.""" |
|
|
msg237656 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 15:05 |
connect_eintr.py: script calling socket.connect() in a loop and sending SIGARLM signal every millisecond. |
|
|
msg237657 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 15:08 |
Oops, connect_eintr.py noticed me (thanks to my recent change of the issue #23571 !) that connect_eintr.patch is wrong: socket.connect() returned None with an exception sent, send connect() was interrupted by SIGINT (CTRL+c). Fixed patch. |
|
|
msg237666 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 15:43 |
> http://www.madore.org/~david/computers/connect-intr.html This article contains a program connect_test.c to test how connect() behaves on EINTR. Since it's in the public domain, I attached a copy. The program contains the comment: "All systems function as expected when TEST_TWO is set." If TEST_TWO is defined, poll() is used to wait until the socket is writable, and then getsockopt(SO_ERROR) is used to check if the error is now zero, when connect() fails with EINTR. |
|
|
msg237669 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-09 15:59 |
connect_eintr-3.patch: Different patch, don't retry connect() if it returns EINTR, but poll using poll/select. The patch changes also the asyncio module. |
|
|
msg238624 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-20 09:09 |
I tested polling+getsockopt(SO_ERROR) with connect_test.c on Linux, it works: haypo@smithers$ gcc -D TEST_TWO=1 connect_test.c -o connect_test haypo@smithers$ ./connect_test Will try to connect to 127.0.0.1 on port 80 (connect has been interrupted and now completed successfully) |
|
|
msg238720 - (view) |
Author: Charles-François Natali (neologix) * |
Date: 2015-03-20 20:49 |
Could you regenerate your latest patch? It doesn't show properly in the review tool. Also, what's with the assert? |
|
|
msg239444 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-28 00:29 |
New changeset f841d3bc30ee by Victor Stinner in branch 'default': Issue #23618, #22117: refactor socketmodule.c https://hg.python.org/cpython/rev/f841d3bc30ee |
|
|
msg239640 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-30 20:34 |
test_selectors.patch: Enhance test_selector to test the two kinds of signal handlers: one raises an exception, the other one doesn't. I wait until kqueue & devpoll retry on EINTR to push test_selectors.patch. |
|
|
msg239690 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-31 12:11 |
New changeset e8246baad0f6 by Victor Stinner in branch 'default': Issue #23618: Refactor the _socket module https://hg.python.org/cpython/rev/e8246baad0f6 New changeset fa5542660b17 by Victor Stinner in branch 'default': Issue #23618: Refactor internal_select() to prepare socket.connect() for EINTR https://hg.python.org/cpython/rev/fa5542660b17 New changeset c027d6468667 by Victor Stinner in branch 'default': Issue #23618: internal_connect_select() now waits also for error events https://hg.python.org/cpython/rev/c027d6468667 |
|
|
msg239693 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-31 12:26 |
New changeset 47b2d1ff9743 by Victor Stinner in branch 'default': Issue #23618: Fix internal_connect_select() https://hg.python.org/cpython/rev/47b2d1ff9743 |
|
|
msg239719 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-31 14:39 |
New changeset daf3d2a717e5 by Victor Stinner in branch 'default': Issue #23618: Refactor internal_connect() https://hg.python.org/cpython/rev/daf3d2a717e5 New changeset c59d81b802f8 by Victor Stinner in branch 'default': Issue #23618: Refactor internal_connect() https://hg.python.org/cpython/rev/c59d81b802f8 |
|
|
msg239741 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-31 17:39 |
Le mardi 31 mars 2015, Roundup Robot <report@bugs.python.org> a écrit : > > New changeset c59d81b802f8 by Victor Stinner in branch 'default': > Issue #23618: Refactor internal_connect() > https://hg.python.org/cpython/rev/c59d81b802f8 > Oh I forgot to add an #ifdef for socklen_t. I didn't get a warning, so socklen_t is probably an int. Getsockopt() uses a int* for the value length. I also forgot to drop the comment before SetLastError(), it's no more revelant. |
|
|
msg239744 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-31 19:32 |
New changeset d9374864d4a9 by Victor Stinner in branch 'default': Issue #23618: Cleanup internal_connect() in socketmodule.c https://hg.python.org/cpython/rev/d9374864d4a9 New changeset 4fad2b9fc4e6 by Victor Stinner in branch 'default': Issue #23618: Fix EINTR handling in socket.connect() https://hg.python.org/cpython/rev/4fad2b9fc4e6 |
|
|
msg239746 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-03-31 20:15 |
connect_eintr-4.patch: Updated patch after all my changes to refactor socket.connect() (which is now almost a rewrite from scratch!). |
|
|
msg239748 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-03-31 20:24 |
New changeset 7ed567ad8b4c by Victor Stinner in branch 'default': Issue #23618: Enhance EINTR handling in socket.connect() https://hg.python.org/cpython/rev/7ed567ad8b4c |
|
|
msg239775 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-01 09:10 |
New changeset 8ec4acfdb851 by Victor Stinner in branch 'default': Issue #23618: Fix EINTR handling on Windows https://hg.python.org/cpython/rev/8ec4acfdb851 |
|
|
msg239896 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 10:33 |
test_connnect_eintr.py: program to interrupt socket.connect() with signals. It looks like socket.connect() cannot be interrupted by signals: connect() only fails with WSAEINTR when WSACancelBlockingCall() is called, but WSACancelBlockingCall() "has been removed in compliance with the Windows Sockets 2 specification, revision 2.2.0": https://msdn.microsoft.com/en-us/library/windows/desktop/ms741547%28v=vs.85%29.aspx "Blocking hooks are generally used to keep a single-threaded GUI application responsive during calls to blocking functions. Instead of using blocking hooks, an applications should use a separate thread (separate from the main GUI thread) for network activity." |
|
|
msg239899 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-02 10:57 |
New changeset aad52bfc816f by Victor Stinner in branch 'default': Issue #23618: Don't declare recvmsg/sendmsg helper functions on Windows https://hg.python.org/cpython/rev/aad52bfc816f New changeset f22188acc77d by Victor Stinner in branch 'default': Issue #23618: Document EINTR changes in socket documentation https://hg.python.org/cpython/rev/f22188acc77d |
|
|
msg239902 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 11:19 |
Hum, connect() does not always block with test_connect_eintr.py, and this program sometimes fail with ConnectionResetError on connect() on FreeBSD. New program which works on Linux and FreeBSD. It now ensures that connect() will block. |
|
|
msg239903 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 11:31 |
test_connect_eintr3.py: even better: - block signals in the server thread - count signals during connect() - display progress: "*" for signal received during connect(), "_" for signal received before/after connect(), "[" and "]" for the beginning and end of a connection, "#" for client connection reset Example of output on FreeBSD: Register SIGINT Register SIGALRM Register SIGWINCH Register SIGTERM Register SIGCHLD Send SIGALRM every 200.0 ms Run func() during 5.0 seconds Type CTRL+c, resize the window, etc. ___[]______[]_____[_*#]_____[#]__[#]________[#]____[#]_____[**]____[*#]______[__#]___[#]_____[#]____[*#]______[#]______[#]_____[*#]_____[#]_____[#]______[#]______[#]______[#]______[#]______[#]____[#]_______[#]_____[#]_______[#]_____[#]_____[#]______[#]_______[#]____[#]______[#]______[*#]_____[#]________[#]__ Test completed in 5.1 sec func() has been called 36 times Got 204 signals Got 7 signals during connect() |
|
|
msg239904 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 11:33 |
Example of test_connect_eintr3.py output on Linux (3.18): Register SIGINT Register SIGALRM Register SIGWINCH Register SIGTERM Register SIGCHLD Send SIGALRM every 200.0 ms Run func() during 5.0 seconds Type CTRL+c, resize the window, etc. []_____[]_____[]_____[]_____[**********][**********][]______[**********][]_____[]____[**********][**********] Test completed in 5.4 sec func() has been called 12 times Got 85 signals Got 50 signals during connect() Example of test_connect_eintr3.py output on Mac OS X Yosemite (10.10): Register SIGINT Register SIGALRM Register SIGWINCH Register SIGTERM Register SIGCHLD Send SIGALRM every 200.0 ms Run func() during 5.0 seconds Type CTRL+c, resize the window, etc. []______[*********][***********][**********]_[********][***********] Test completed in 5.3 sec func() has been called 6 times Got 55 signals Got 49 signals during connect() |
|
|
msg239905 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 11:37 |
Example of test_connect_eintr3.py output on OpenIndiana: Register SIGINT Register SIGALRM Register SIGWINCH Register SIGTERM Register SIGCHLD Send SIGALRM every 200.0 ms Run func() during 5.0 seconds Type CTRL+c, resize the window, etc. ______[]____[]_____[******]______[]_______[*****]______[*****]______[] Test completed in 5.2 sec func() has been called 7 times Got 56 signals Got 16 signals during connect() Oh, and obviously, test_connect_eintr3.py fails with InterruptedError without the patch. Example on Linux: Register SIGINT Register SIGALRM Register SIGWINCH Register SIGTERM Register SIGCHLD Send SIGALRM every 200.0 ms Run func() during 5.0 seconds Type CTRL+c, resize the window, etc. []____[]_____[]_____[]____[******Traceback (most recent call last): File "test_connect_eintr.py", line 97, in func() File "test_connect_eintr.py", line 51, in func client.connect(server_addr) InterruptedError: [Errno 4] Interrupted system call |
|
|
msg239906 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-02 11:42 |
New changeset 75fcc7a3738a by Victor Stinner in branch 'default': Issue #23618: socket.socket.connect() now waits until the connection completes https://hg.python.org/cpython/rev/75fcc7a3738a |
|
|
msg239911 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2015-04-02 12:25 |
http://buildbot.python.org/all/builders/AMD64%20Snow%20Leop%203.x/builds/2904/steps/test/logs/stdio ====================================================================== FAIL: test_connect_ex_error (test.test_ssl.NetworkedTests) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/buildbot/buildarea/3.x.murray-snowleopard/build/Lib/test/test_ssl.py", line 1441, in test_connect_ex_error self.assertIn(rc, (errno.ECONNREFUSED, errno.EWOULDBLOCK)) AssertionError: 36 not found in (61, 35) where 36 is errno.EINPROGRESS |
|
|
msg239912 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-02 12:38 |
New changeset 44adbb5eeb4b by Victor Stinner in branch 'default': Issue #23618: Fix sock_connect_impl(), set the socket error code https://hg.python.org/cpython/rev/44adbb5eeb4b |
|
|
msg239913 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-02 13:20 |
New changeset 09a4d5cc6afd by Victor Stinner in branch 'default': Issue #23618: Ooops, remove abort() added for debug purpose https://hg.python.org/cpython/rev/09a4d5cc6afd |
|
|
msg240312 - (view) |
Author: Roundup Robot (python-dev) |
Date: 2015-04-09 08:32 |
New changeset bff88c866886 by Victor Stinner in branch 'default': Issue #23618: Fix internal_select() for negative timeout (blocking socket) when https://hg.python.org/cpython/rev/bff88c866886 |
|
|