Re: [Xen-devel] [PATCH 11 of 18] tools/libxl: fix build errors caused by Werror in disk_eject_xswatch_callback

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2012-04-02 at 20:58 +0100, Olaf Hering wrote:
> On Mon, Apr 02, Ian Jackson wrote:
> 
> > Olaf Hering writes ("[Xen-devel] [PATCH 11 of 18] tools/libxl: fix build 
> > errors caused by Werror in disk_eject_xswatch_callback"):
> > > tools/libxl: fix build errors caused by Werror in 
> > > disk_eject_xswatch_callback
> > > 
> > > -O2 -Wall -Werror triggers these warnings:
> > > 
> > > libxl.c: In function 'disk_eject_xswatch_callback':
> > > libxl.c:928: warning: zero-length printf format string
> > 
> > There is nothing wrong with zero-length format strings.  This warning
> > should be disabled.  Please do send a patch to do so.
> 
> -Wno-format-zero-length -Wall has no effect, and -Wall comes from
> RPM_OPT_FLAGS which will come last with my current EXTRA_CFLAGS patch.

IMHO it's a little bit unreasonable for RPM to think it knows better
than upstream what warning flags are appropriate. Especially given not
all the warnings are useful and/or make sense (which is certainly the
case for Wformat-zero-length[0]).

I'd suggest that RPM_OPT_FLAGS either ought to omit
Wno-format-zero-length or, if it is a member of some umbrella -W option,
grow a -Wno-format-zero-length. Or if you cannot control RPM_OPT_FLAGS
in that way you should pass it alongside RPM_OPT_FLAGS in via
EXTRA_CFLAGS.

[0] Even gcc(1) says:
       -Wno-format-zero-length (C and Objective-C only)
           If -Wformat is specified, do not warn about zero-length formats.
           The C standard specifies that zero-length formats are allowed.
So quite why it appears that Wall is enabling Wformat-zero-length I have
no idea, this is a clear gcc bug in my opinion.

Ian.



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel