Hi,
On Monday, 11 September 2006 12:03, Stefan Seyfried wrote:
> Hi,
>
> building with compiler flags "-Wall -O2 -D_FORTIFY_SOURCE=2" throws some
> warnings:
>
> [EMAIL PROTECTED]:~/projects/powermanagement/cvs/suspend> make CFLAGS='-Wall
> -O2 -D_FORTIFY_SOURCE=2'
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -c vt.c -o vt.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -DHAVE_INTTYPES_H -DHAVE_STDINT_H -c md5.c
> -o md5.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -DHAVE_INTTYPES_H -DHAVE_STDINT_H
> -I/usr/local/include -c encrypt.c -o encrypt.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -I/usr/local/include -c config.c -o config.o
> cc -g -Wall -O2 -D_FORTIFY_SOURCE=2 -I/usr/local/include -c bootsplash.c -o
> bootsplash.o
> cc -g -Wall -O2 -D_FORTIFY_SOURCE=2 -I/usr/local/include -c splash.c -o
> splash.o
> cc -g -Wall -O2 -D_FORTIFY_SOURCE=2 -I/usr/local/include vt.o md5.o encrypt.o
> config.o suspend.c -o s2disk splash.o bootsplash.o -L/usr/local/lib
> suspend.c: In function ‘prepare_console’:
> suspend.c:816: warning: ignoring return value of ‘write’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘get_kernel_console_loglevel’:
> suspend.c:886: warning: ignoring return value of ‘fscanf’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘get_swappiness’:
> suspend.c:919: warning: ignoring return value of ‘fscanf’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘main’:
> suspend.c:1192: warning: ignoring return value of ‘chdir’, declared with
> attribute warn_unused_result
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -c vbetool/lrmi.c -o vbetool/lrmi.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -c vbetool/x86-common.c -o
> vbetool/x86-common.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -DS2RAM -c vbetool/vbetool.c -o
> vbetool/vbetool.o
> vbetool/vbetool.c: In function ‘save_state’:
> vbetool/vbetool.c:347: warning: ignoring return value of ‘write’, declared
> with attribute warn_unused_result
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -DS2RAM -c radeontool.c -o radeontool.o
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -DS2RAM -c dmidecode.c -o dmidecode.o
> cc -g -Wall -O2 -D_FORTIFY_SOURCE=2 -DCONFIG_BOTH -I/usr/local/include md5.o
> encrypt.o config.o suspend.c s2ram.c -o s2both vt.o vbetool/lrmi.o
> vbetool/x86-common.o vbetool/vbetool.o radeontool.o dmidecode.o splash.o
> bootsplash.o -L/usr/local/lib -lpci
> suspend.c: In function ‘prepare_console’:
> suspend.c:816: warning: ignoring return value of ‘write’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘get_kernel_console_loglevel’:
> suspend.c:886: warning: ignoring return value of ‘fscanf’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘get_swappiness’:
> suspend.c:919: warning: ignoring return value of ‘fscanf’, declared with
> attribute warn_unused_result
> suspend.c: In function ‘main’:
> suspend.c:1192: warning: ignoring return value of ‘chdir’, declared with
> attribute warn_unused_result
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -I/usr/local/include md5.o encrypt.o
> config.o vt.o resume.c splash.o bootsplash.o -static -o resume
> -L/usr/local/lib
> resume.c: In function ‘main’:
> resume.c:737: warning: ignoring return value of ‘fgets’, declared with
> attribute warn_unused_result
> cc -Wall -O2 -D_FORTIFY_SOURCE=2 -g s2ram.c vt.o vbetool/lrmi.o
> vbetool/x86-common.o vbetool/vbetool.o radeontool.o dmidecode.o -lpci -o s2ram
>
> In our build system, _FORTIFY_SOURCE=2 is mandatory and sooner or later those
> warnings will become errors. I can always work around them by casting to void,
> but as was pointed out to me by our toolchain team, some of them are probably
> real errors.
Which ones?
> If we _know_ that this write or fscanf cannot fail because we also _know_ what
> we are doing on the kernel side, then we should probably add a comment at the
> appropriate places and maybe just add a (void) cast. This is ugly to look at,
> but tells everybody else that we did not just forget to check for the values.
Well, we can. We can also return the result of the write or fscanf from the
function that's calling them and ignore it in the caller.
> Even leaving out the void cast, an comment like "this will throw a warning,
> but we know that this cannot fail" is better so i can defend my local patches
> against theevil overlords of the buildsystem :-)
I personally don't like the _FORTIFY_SOURCE=2 or equivalent ideas, because
usually they force people to work around them which results in code that's
difficult to read.
Greetings,
Rafael
--
You never change things by fighting the existing reality.
R. Buckminster Fuller
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Suspend-devel mailing list
Suspend-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/suspend-devel
