libX11: Changes to 'master'

[Alan Coopersmith]

 specs/libX11/CH01.xml |    2 +-
 specs/libX11/CH14.xml |    2 +-
 src/Iconify.c         |   26 +++++++++++++++-----------
 src/ReconfWM.c        |   37 ++++++++++++++++++++-----------------
 src/Withdraw.c        |   22 ++++++++++++----------
 src/xkb/XKB.c         |   11 ++++-------
 src/xkb/XKBMAlloc.c   |    2 +-
 7 files changed, 54 insertions(+), 48 deletions(-)

New commits:
commit 5525e8433f93bce464412f27cffa203ea628f368
Author: Alan Coopersmith <alan.coopersmith at oracle.com>
Date:   Tue Jul 8 00:05:32 2014 -0700

    specs/libX11: disengender a user reference
    
    Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>

commit d8679eae9317b389ad4acb0430360ee0663e2af8
Author: Alan Coopersmith <alan.coopersmith at oracle.com>
Date:   Fri Jul 11 18:41:42 2014 -0700

    specs/libX11: Correct value of IconicState to match Xutil.h
    
    Xutil.h has always had a value of 3 for IconicState, since 2 was
    previously used for the long-obsolete ZoomState, so make the spec
    match what programs have used for decades.
    
    Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>
    Reviewed-by: Matthieu Herrb <matthieu at herrb.eu>

commit 7ce2b0f12a48fb832f457cbafb0e1144ef557f9a
Author: Alan Coopersmith <alan.coopersmith at oracle.com>
Date:   Fri Jul 11 10:34:08 2014 -0700

    Use C99 named initializers to fill in events passed to XSendEvent
    
    Forces compiler to zero-fill unset fields in the struct (fixing bug 81236)
    and allows optimizer to order field initialization to best fit cache layout
    or other considerations.
    
    Before & after output of gcc -S on AMD64 shows insertion of "rep stosq"
    instructions to rapidly zero-fill structs.
    
    Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>
    Reviewed-by: Matthieu Herrb <matthieu at herrb.eu>

commit 169805e1dc8743b37b00e24cf3a5eb8748f733ad
Author: Alan Coopersmith <alan.coopersmith at oracle.com>
Date:   Sun Jul 6 11:13:49 2014 -0700

    Fix validation of ctrls parameter to XkbGetPerClientControls()
    
    Nothing in the XKB spec states that the memory pointed to by ctrls has to
    be initialized to any given value when passed to the function, only that
    it is set by the function to the values returned by the X server:
    http://www.x.org/releases/X11R7.7/doc/libX11/XKB/xkblib.html#The_Miscellaneous_Per_client_Controls
    
    The check for the incoming value seems to be copied from
    XkbSetPerClientControls without explanation.
    
    Instead change it to checking if ctrls is non-NULL, since there's no
    point asking the X server to return a value the caller won't even see.
    
    Found while investigating report from cppcheck-1.65:
    [src/xkb/XKB.c:699] -> [src/xkb/XKB.c:719]: (warning) Possible null pointer
     dereference: ctrls - otherwise it is redundant to check it against null.
    
    Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>

commit 1e362fac92c6688fb42b195ccad16d7a337a34c1
Author: Alan Coopersmith <alan.coopersmith at oracle.com>
Date:   Sun Jul 6 10:54:57 2014 -0700

    Fix map->num_types check in XkbAddKeyType()
    
    Check is intended to ensure we allocate at least XkbNumRequiredTypes
    in map, but was accidentally marked with a ! causing the wrong check.
    
    Reported-by: Harms <wharms at bfs,de>
    Reviewed-by: Matthieu Herrb <matthieu at herrb.eu>
    Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>