Re: [PATCH] cc-can-link.sh: check for linking capability in a more robust way

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Jul 06, 2021 at 01:33:45AM +0900, Masahiro Yamada wrote:
> On Sat, Jul 3, 2021 at 8:21 PM Alexey Dobriyan <adobriyan@xxxxxxxxx> wrote:
> >
> > Compiling "printf("");" doesn't necessarily check for linking capability
> > as printf can be optimised for constants strings even at -O0:
> >
> >         0000000000401106 <main>:
> >           401106:       push   rbp
> >           401107:       mov    rbp,rsp
> >           40110a:       mov    eax,0x0
> >           40110f:       pop    rbp
> >           401110:       ret
> >
> > Pass something from the command line to disable optimisations:
> >
> >         0000000000401126 <main>:
> >           401126:       push   rbp
> >           401127:       mov    rbp,rsp
> >           40112a:       sub    rsp,0x10
> >           40112e:       mov    DWORD PTR [rbp-0x4],edi
> >           401131:       mov    QWORD PTR [rbp-0x10],rsi
> >           401135:       mov    rax,QWORD PTR [rbp-0x10]
> >           401139:       add    rax,0x8
> >           40113d:       mov    rax,QWORD PTR [rax]
> >           401140:       mov    rdi,rax
> >           401143:       mov    eax,0x0
> >           401148:  ***  call   401030 <printf@plt>
> >           40114d:       mov    eax,0x0
> >           401152:       leave
> >           401153:       ret
> >
> > Signed-off-by: Alexey Dobriyan <adobriyan@xxxxxxxxx>
> > ---
> >
> >  scripts/cc-can-link.sh |    4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > --- a/scripts/cc-can-link.sh
> > +++ b/scripts/cc-can-link.sh
> > @@ -3,9 +3,9 @@
> >
> >  cat << "END" | $@ -x c - -o /dev/null >/dev/null 2>&1
> >  #include <stdio.h>
> > -int main(void)
> > +int main(int argc, char *argv[])
> >  {
> > -       printf("");
> > +       printf(argv[1]);
> >         return 0;
> >  }
> >  END
> 
> Ah, right.
> 
> But, we should not merge a bad coding example.
> 
> argv[1] may contain '%' format string, and
> recent GCC versions warn about it.
> 
> 
> 
> $ cat test.c
> #include <stdio.h>
> 
> int main(int argc, char *argv[])
> {
>     printf(argv[1]);
>     return 0;
> }
> $ gcc -c -o test.o  test.c
> test.c: In function ‘main’:
> test.c:5:5: warning: format not a string literal and no format
> arguments [-Wformat-security]
>     5 |     printf(argv[1]);
>       |     ^~~~~~

> I think replacing printf("") with printf("a")
> (or any string you like)
> is enough.

I get putchar() for "a". puts(argv[1]) works too.
I think argv[1] should be used to defeat optimisers, current and future.



[Index of Archives]     [Linux&nblp;USB Development]     [Linux Media]     [Video for Linux]     [Linux Audio Users]     [Yosemite Secrets]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux