Thanks for fixing this... apparently the problem is that what I committed was (for example):
patches/patch-libtiff_tif_dir.h:
$NetBSD$
fix CVE-2014-8128, CVE-2016-5318, CVE-2015-7554 & CVE-2016-10095
per
http://bugzilla.maptools.org/show_bug.cgi?id=2580also CVE-2017-9147
(
http://bugzilla.maptools.org/show_bug.cgi?id=2693)
Index: tif_dir.h
===================================================================
RCS file: /cvs/maptools/cvsroot/libtiff/libtiff/tif_dir.h,v
retrieving revision 1.54
retrieving revision 1.55
diff -w -u -b -r1.54 -r1.55
--- libtiff/tif_dir.h.orig 18 Feb 2011 20:53:05 -0000 1.54
+++ libtiff/tif_dir.h 1 Jun 2017 12:44:04 -0000 1.55
@@ -1,4 +1,4 @@
-/* $Id: tif_dir.h,v 1.54 2011-02-18 20:53:05 fwarmerdam Exp $ */
+/* $Id: tif_dir.h,v 1.55 2017-06-01 12:44:04 erouault Exp $ */
/*
* Copyright (c) 1988-1997 Sam Leffler
@@ -291,6 +291,7 @@
extern int _TIFFMergeFields(TIFF*, const TIFFField[], uint32);
extern const TIFFField* _TIFFFindOrRegisterField(TIFF *, uint32, TIFFDataType);
extern TIFFField* _TIFFCreateAnonField(TIFF *, uint32, TIFFDataType);
+extern int _TIFFCheckFieldIsValidForCodec(TIFF *tif, ttag_t tag);
#if defined(__cplusplus)
}
Which and apparently the multiple cases of $xxx$ caused cvs to do something very very wrong.