Re: [lmi] wxMSW, x86_64-w64-mingw32-gcc-4.9.1: warning in expat

[Vadim Zeitlin]

On Tue, 23 May 2017 12:11:53 +0000 Greg Chicares <address@hidden> wrote:

GC> /opt/lmi/src/lmi[0]$make $coefficiency 
exec_prefix=/opt/lmi/x86_64-w64-mingw32/local host_type=x86_64-w64-mingw32 
--output-sync=recurse -f install_wx.make
GC> ...
GC> checking for expat.h... no
GC> configure: WARNING: system expat library not found, will use built-in 
instead
GC> ...
GC> === configuring in src/expat 
(/opt/lmi/wx-scratch/wxWidgets-3.1.0-p1/gcc-491-89775012799fe5c9bd8ea61e5fa43da7/src/expat)
GC> configure: running /bin/bash ../../../src/expat/configure 
--disable-option-checking '--prefix=/opt/lmi/local'  
'--exec-prefix=/opt/lmi/x86_64-w64-mingw32/local' 
'--build=x86_64-unknown-linux-gnu' '--host=x86_64-w64-mingw32' 
'--disable-apple_ieee' '--disable-aui' '--disable-compat24' 
'--disable-fswatcher' '--disable-gif' '--disable-mediactrl' 
'--disable-propgrid' '--disable-ribbon' '--disable-richtext' '--disable-stc' 
'--disable-webview' '--enable-monolithic' '--enable-shared' 
'--enable-std_iostreams' '--enable-stl' 
'--enable-vendor=gcc-491-89775012799fe5c9bd8ea61e5fa43da7' '--without-libjpeg' 
'--without-libtiff' '--without-opengl' '--without-subdirs' 
'CFLAGS=-fno-omit-frame-pointer' 'CXXFLAGS=-fno-omit-frame-pointer -std=c++11' 
'build_alias=x86_64-unknown-linux-gnu' 'host_alias=x86_64-w64-mingw32' 
--cache-file=/dev/null --srcdir=../../../src/expat
GC> ...
GC> 
GC> <LENGTHY_DIGRESSION>
GC> checking for x86_64-w64-mingw32-gcc... x86_64-w64-mingw32-gcc
GC> ...
GC> checking for x86_64-w64-mingw32-mt... no
GC> checking for mt... mt
GC> configure: WARNING: using cross tools not prefixed with host triplet
GC> checking if mt is a manifest tool... no
... snip ...
GC> So 'configure' is looking for an msw "manifest tool", but finding this:
GC> 
GC> https://linux.die.net/man/1/mt
GC> | mt - control magnetic tape drive operation
GC> 
GC> </LENGTHY_DIGRESSION>

 Yes, this is really weird and I don't like seeing it too. I've never
thought about just following the advice to create a symlink for it, maybe I
should just do this, even if it seems ugly too -- but at least it's ugly to
do once, and then you can forget about it forever.

 BTW, it's not configure per se, it's libtool, I think. As always, all the
strange things in autotools family come from this member...

GC> Anyway, the warning I'm reporting is this:
GC> 
GC> ...
GC> 
/opt/lmi/wx-scratch/wxWidgets-3.1.0-p1/gcc-491-89775012799fe5c9bd8ea61e5fa43da7/bk-deps
 x86_64-w64-mingw32-gcc -c -o wxexpat_xmlparse.o -DNDEBUG -I./src/expat 
-DHAVE_EXPAT_CONFIG_H  -D_FILE_OFFSET_BITS=64 
-I/opt/lmi/wx-scratch/wxWidgets-3.1.0-p1/gcc-491-89775012799fe5c9bd8ea61e5fa43da7/lib/wx/include/x86_64-w64-mingw32-msw-unicode-3.1
 -I../include -Wall -Wundef -O2 -fno-omit-frame-pointer  
../src/expat/lib/xmlparse.c
GC> ../src/expat/lib/xmlparse.c: In function 'generate_hash_secret_salt':
GC> ../src/expat/lib/xmlparse.c:725:42: warning: cast from pointer to integer 
of different size [-Wpointer-to-int-cast]
GC>        gather_time_entropy() ^ getpid() ^ (unsigned long)parser;
GC>                                           ^

 This seems harmless to me, both because we don't care about truncating the
pointer here when we just want to have some random bits and because this
hash secret only matters when parsing untrusted output in order to avoid
denial of service attacks -- so it doesn't affect lmi anyhow. But in any
case it was already fixed in expat which, completely unexpectedly, is being
developed again after a 15+ year interval, in this commit:
https://github.com/libexpat/libexpat/commit/863c4308

 So when we update to the next expat release, this warning wouldn't happen
any more.

 Regards,
VZ