[Bug 51289] New: ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

Bug ID: 51289
           Summary: ws2_32:sock has siocatmark failures on wvistau64
           Product: Wine
           Version: 6.10
          Hardware: x86-64
                OS: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: winsock
          Assignee: [e-mail]
          Reporter: [e-mail]

ws2_32:sock fails quite consistently on wvistau64:

https://test.winehq.org/data/patterns.html#ws2_32:sock

sock.c:3838: Test failed: FIONBIO returned 0
sock.c:3838: Test failed: SIOCATMARK returned 1
sock.c:3848: Test failed: FIONBIO returned 0

These failures happen consistently for both 32- and 64-bit tests on wvistau64
but only the 64-bit WineTest is run daily. 
They don't happen at all on wvista though so there must be some difference
between these two VMs.

wvista has the older ws2_32 dll (6.0.6001.18000 vs. 6.0.6002.18005 for
wvistau64), which means the failures happen on the more recent ws2_32. This
makes no sense so the root cause is probably elsewhere.

The failures look like they might be timing related but while adding a
Sleep(50) before the failing lines avoid the first 3838 failure, it's not
reliable and does nothing for the 3848 failure.

The failing tests were introduced in this commit:

commit 2795fa9c00338bfff6304c1b756d1a58f9f6ca60
Author: Zebediah Figura 
Date:   Thu May 27 18:57:31 2021 -0500

ws2_32/tests: Separate and expand tests for FIONREAD and SIOCATMARK.

Signed-off-by: Zebediah Figura 
    Signed-off-by: Alexandre Julliard

But the TestBot only reported the failures for the next patch in the series:
https://www.winehq.org/pipermail/wine-devel/2021-May/187782.html

[Bug 51289] ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

François Gouget  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
    Regression SHA1|                            |2795fa9c00338bfff6304c1b756
                   |                            |d1a58f9f6ca60
           Keywords|                            |source, testcase

[Bug 51289] ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

Zebediah Figura  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

[Bug 51289] ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

--- Comment #1 from Zebediah Figura  ---
Vista has an older networking stack, and the loopback path isn't reliably
synchronous, which means that we generally have to wait for data to be
available before doing nonblocking operations. I've had to deal with it many
times by now.

I don't know why wvista isn't affected, though.

[Bug 51289] ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

François Gouget  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |WONTFIX

--- Comment #2 from François Gouget  ---
Vista is no longer part of the Windows versions that the Wine tests are
expected to succeed on so I'm marking this as resolved.

[Bug 51289] ws2_32:sock has siocatmark failures on wvistau64

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=51289

Gijs Vermeulen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #3 from Gijs Vermeulen  ---
Closing WONTFIX.