[Xcb] Bug#486507: libx11-6: VMWare Server sees X responses out of order

Julien Cristau jcristau at debian.org
Wed Dec 3 12:47:34 PST 2008

tag 486507 moreinfo

On Sat, Oct  4, 2008 at 13:25:17 -0700, Jamey Sharp wrote:

> The locking assertion failures and their associated backtraces should be
> looked into, but may be unrelated to the assertion failure that's
> actually causing VMWare Server to abort():
> vmware: ../../src/xcb_lock.c:77: _XGetXCBBuffer: assertion  `((int) ((xcb_req) - (dpy->request)) >= 0)' failed 
> First: Could somebody who has encountered this bug please get a stack
> trace from the actual assert?
> Does anybody know if VMWare Server is calling Xlib from multiple threads
> concurrently? A `thread apply all bt full` in GDB when the assertion
> triggers might be informative, especially with libx11-6-dbg installed.
> I suspect this problem would be fixed by the socket handoff work that
> Josh and I have been trying to finish for months. If you're willing to
> recompile XCB and Xlib yourself, you might try the v2 handoff patch set:
> 	http://lists.freedesktop.org/archives/xcb/2008-March/003392.html
> There are some known bugs in that version, but you're unlikely to
> encounter them until your app has been running for hours. Unfortunately
> the XCB patches don't apply to current upstream git, so you might need
> to apply them against commit 7a74ba3d0212f9bfe021d6da9070f71cbc53f85b.
It would be nice to get some feedback on this from the people who can
reproduce the issue.  This means either getting a backtrace from the
failing assert(), with debug packages for libxcb and libx11 installed,
or upgrading libx11-6 to the version in experimental, which uses the
work mentioned above.


More information about the Xcb mailing list