Re: [PATCH] perf: fix -Wstring-compare

[Arnaldo Carvalho de Melo <arnaldo.melo@gmail.com>]

Em Tue, Feb 25, 2020 at 01:28:50PM -0800, Nick Desaulniers escreveu:
> On Tue, Feb 25, 2020 at 1:35 AM David Laight <David.Laight@aculab.com> wrote:
> >
> > From: Nick Desaulniers
> > > Sent: 24 February 2020 22:06
> > > On Mon, Feb 24, 2020 at 10:20 AM 'Ian Rogers' via Clang Built Linux
> > > <clang-built-linux@googlegroups.com> wrote:
> > > >
> > > > On Mon, Feb 24, 2020 at 8:03 AM David Laight <David.Laight@aculab.com> wrote:
> > > > >
> > > > > From: Ian Rogers
> > > > > > Sent: 24 February 2020 05:56
> > > > > > On Sun, Feb 23, 2020 at 11:35 AM Nick Desaulniers
> > > > > > <nick.desaulniers@gmail.com> wrote:
> > > > > > >
> > > > > > > Clang warns:
> > > > > > >
> > > > > > > util/block-info.c:298:18: error: result of comparison against a string
> > > > > > > literal is unspecified (use an explicit string comparison function
> > > > > > > instead) [-Werror,-Wstring-compare]
> > > > > > >         if ((start_line != SRCLINE_UNKNOWN) && (end_line != SRCLINE_UNKNOWN)) {
> > > > > > >                         ^  ~~~~~~~~~~~~~~~
> > > > > > > util/block-info.c:298:51: error: result of comparison against a string
> > > > > > > literal is unspecified (use an explicit string comparison function
> > > > > > > instead) [-Werror,-Wstring-compare]
> > > > > > >         if ((start_line != SRCLINE_UNKNOWN) && (end_line != SRCLINE_UNKNOWN)) {
> > > > > > >                                                          ^  ~~~~~~~~~~~~~~~
> > > > > > > util/block-info.c:298:18: error: result of comparison against a string
> > > > > > > literal is unspecified (use an explicit string
> > > > > > > comparison function instead) [-Werror,-Wstring-compare]
> > > > > > >         if ((start_line != SRCLINE_UNKNOWN) && (end_line != SRCLINE_UNKNOWN)) {
> > > > > > >                         ^  ~~~~~~~~~~~~~~~
> > > > > > > util/block-info.c:298:51: error: result of comparison against a string
> > > > > > > literal is unspecified (use an explicit string comparison function
> > > > > > > instead) [-Werror,-Wstring-compare]
> > > > > > >         if ((start_line != SRCLINE_UNKNOWN) && (end_line != SRCLINE_UNKNOWN)) {
> > > > > > >                                                          ^  ~~~~~~~~~~~~~~~
> > > > > > > util/map.c:434:15: error: result of comparison against a string literal
> > > > > > > is unspecified (use an explicit string comparison function instead)
> > > > > > > [-Werror,-Wstring-compare]
> > > > > > >                 if (srcline != SRCLINE_UNKNOWN)
> > > > > > >                             ^  ~~~~~~~~~~~~~~~
> > > > > > >
> > > > > > > Link: https://github.com/ClangBuiltLinux/linux/issues/900
> > > > > > > Signed-off-by: Nick Desaulniers <nick.desaulniers@gmail.com>
> > > > > > > ---
> > > > > > > Note: was generated off of mainline; can rebase on -next if it doesn't
> > > > > > > apply cleanly.
> > > >
> > > > Reviewed-by: Ian Rogers <irogers@google.com>
> > > >
> > > > > > Looks good to me. Some more context:
> > > > > > https://clang.llvm.org/docs/DiagnosticsReference.html#wstring-compare
> > > > > > The spec says:
> > > > > > J.1 Unspecified behavior
> > > > > > The following are unspecified:
> > > > > > .. Whether two string literals result in distinct arrays (6.4.5).
> > > > >
> > > > > Just change the (probable):
> > > > > #define SRCLINE_UNKNOWN "unknown"
> > > > > with
> > > > > static const char SRC_LINE_UNKNOWN[] = "unk";
> > > > >
> > > > >         David
> > > >
> > > >
> > > > The SRCLINE_UNKNOWN is used to convey information. Having multiple
> > > > distinct pointers (static) would mean the compiler could likely remove
> > > > all comparisons as the compiler could prove that pointer is never
> > > > returned by a function - ie comparisons are either known to be true
> > > > (!=) or false (==).
> > >
> > > I wouldn't define a static string in a header.  Though I could:
> > > 1. forward declare it in the header with extern linkage.
> > > 2. define it in *one* .c source file.
> > >
> > > Thoughts on that vs the current approach?
> >
> > The string compares are just stupid.
> > If the 'fake' strings are not printed you could use:
> > #define SRCLINE_UNKNOWN ((const char *)1)
> >
> > Otherwise defining the strings in one of the C files is better.
> > Relying on the linker to merge the strings from different compilation
> > units is so broken...
> 
> Note: it looks like free_srcline() already does strcmp, so my patch
> basically does a more consistent job for string comparisons.  Forward
> declaring then defining in tools/perf/util/srcline.c involves changing
> the function signatures and struct members to `const char*` rather
> than `char*`, which is of questionable value.  I wouldn't mind
> changing my patch to just use strcmp instead of strncmp, or convert
> free_srcline() to use strncmp instead, if folks felt strongly about
> being consistent. Otherwise I think my patch with Ian's Reviewed-by is
> the best approach.

Fair enough, applying it with Ian's reviewed-by,

- Arnaldo