Re: GCC 6.1 on Debian build report, a few warnings

On Tue, May 17, 2016 at 10:58:16PM -0400, Dan Collins wrote:
> Hello porters,
> 
> While building a bleadperl under GCC 6.1, I noticed a few compiler
> warnings. Not sure if they're new or not.
> 
> In the miniperl stage, in util.c
> 
> ccache gcc-6.1 -c -DPERL_CORE -fwrapv -DDEBUGGING -fno-strict-aliasing
> -pipe -fstack-protector-strong -I/usr/local/include -D_LARGEFILE_SOURCE
> -D_FILE_OFFSET_BITS=64 -g -Wall -Werror=declaration-after-statement -Wextra
> -Wc++-compat -Wwrite-strings -Wno-format util.c
> util.c: In function ‘Perl_my_vsnprintf’:
> util.c:5299:18: warning: ‘sizeof’ on array function parameter ‘ap’ will
> return size of ‘__va_list_tag *’ [-Wsizeof-array-argument]
>      PERL_UNUSED_ARG(ap);
>                   ^
> util.c:5293:79: note: declared here
>  Perl_my_vsnprintf(char *buffer, const Size_t len, const char *format,
> va_list ap)
> 
>    ^~
> This appears to be an attempt to suppress a -Wunused-parameter that is no
> longer working because GCC got clever.

Now silenced with v5.25.2-10-gbf49eae

> In pp_ctl.c
> 
> ccache gcc-6.1 -c -DPERL_CORE -fwrapv -DDEBUGGING -fno-strict-aliasing
> -pipe -fstack-protector-strong -I/usr/local/include -D_LARGEFILE_SOURCE
> -D_FILE_OFFSET_BITS=64 -g -Wall -Werror=declaration-after-statement -Wextra
> -Wc++-compat -Wwrite-strings -Wno-format pp_ctl.c
> pp_ctl.c: In function ‘Perl_pp_goto’:
> pp_ctl.c:2782:104: warning: self-comparison always evaluates to false
> [-Wtautological-compare]
>        EXTEND(SP, items+1); /* @_ could have been extended. */
> 
>                             ^

I think gcc is wrong here, although I don't have gcc 6 on hand to play
with. I think its falling foul of this from pp.h:

    /* _EXTEND_SAFE_N(n): private helper macro for EXTEND().
     * Tests whether the value of n would be truncated when implicitly cast to
     * SSize_t as an arg to stack_grow(). If so, sets it to -1 instead to
     * trigger a panic. It will be constant folded on platforms where this
     * can't happen.
     */

    #define _EXTEND_SAFE_N(n) \
            (sizeof(n) > sizeof(SSize_t) && ((SSize_t)(n) != (n)) ? -1 : (n))

In particular the (SSize_t)(n) != (n) bit, which should only be executed
if sizeof(n) > sizeof(SSize_t), in which case the cast version if n is
*not* necessarily the same as n, and so the test isn't tautological.

> After miniperl has been built, in inflate.c

inflate.c is part of cpan/Compress-Raw-Zlib, so needs reporting / fixing
upstream (but in fact that file is from the zlib library, so is actually
upstream of Compress-Raw-Zlib).
> 
> ccache gcc-6.1 -c  -I./zlib-src -fwrapv -DDEBUGGING -fno-strict-aliasing
> -pipe -fstack-protector-strong -I/usr/local/include -D_LARGEFILE_SOURCE
> -D_FILE_OFFSET_BITS=64 -Wall -Werror=declaration-after-statement -Wextra
> -Wc++-compat -Wwrite-strings -Wno-format -g   -DVERSION=\"2.069\"
> -DXS_VERSION=\"2.069\" -fPIC "-I../.."  -DNO_VIZ -DZ_SOLO
> -DGZIP_OS_CODE=3  inflate.c
> inflate.c: In function ‘inflateUndermine’:
> inflate.c:1487:9: warning: unused parameter ‘subvert’ [-Wunused-parameter]
>      int subvert)
>          ^~~~~~~

Already reported as https://rt.cpan.org/Public/Bug/Display.html?id=107642


> inflate.c: In function ‘inflateMark’:
> inflate.c:1507:51: warning: left shift of negative value
> [-Wshift-negative-value]
>      if (strm == Z_NULL || strm->state == Z_NULL) return -1L << 16;

I haven't seen that one before.

-- 
"There's something wrong with our bloody ships today, Chatfield."
    -- Admiral Beatty at the Battle of Jutland, 31st May 1916.

• Re: GCC 6.1 on Debian build report, a few warnings by Dave Mitchell
• Re: GCC 6.1 on Debian build report, a few warnings by Dan Collins