152728 - XpuWaitForPrintFileChild() needs to call XFlush() before waiting for the consumer child

Status: RESOLVED FIXED

(Core Graveyard :: Printing: Xprint, defect)

Description

[Roland Mainz]

XpuWaitForPrintFileChild() needs to call XFlush() before waiting for the
consumer child, otherwise it will may cause a deadlock between consumer child
and main process/thread.

The problem is burried deeply in the Xlib internals, the deadlock occurs because
XpuWaitForPrintFileChild() waits for a child process (or thread) while the child
is still expecting data which will never appear because the registered hooks
have to be called (via any Xlib event) function first. But the parent is waiting
for the child and does not do any event processing at that moment --> deadlock.

Mozilla's Xprint module seems to be _not_ affected by this problem since we wait
for the XpExtension events (which triggers a |XFlush()| automagically when there
are no events queued yet (maybe we hit this issue on large multi-CPU-machines -
but it is still very unlikely)).

Comment 1

[Roland Mainz]

Suffering is all mine...

Comment 2

[Roland Mainz]

Attachment: Patch for 2002-06-13-08-trunk

Comment 3

[Roland Mainz]

Requesting r=/sr= ...

Comment 4

[Calvin Liu]

Comment on attachment 88230 [details] [diff] [review]
Patch for 2002-06-13-08-trunk

r=[e-mail]

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 88230 [details] [diff] [review]
Patch for 2002-06-13-08-trunk

sr=bzbarsky

Comment 6

[Roland Mainz]

Patch checked-in
(http://bonsai.mozilla.org/cvsquery.cgi?module=MozillaTinderboxAll&branch=HEAD&cvsroot=/cvsroot&date=explicit&mindate=1030709640&maxdate=1030710900&who=peterv%25netscape.com),
marking bug as FIXED.

Comment 7

[Roland Mainz]

Filed bug http://xprint.mozdev.org/bugs/show_bug.cgi?id=2146
("XpuWaitForPrintFileChild() needs to call XFlush() before waiting for the
consumer child") to check this patch "in" into the XprintUtil copy in
xprint.mozdev.org's source tree...