On 26 Oct 2022 19:57, Tsukasa OI wrote:
> On 2022/10/26 17:59, Mike Frysinger wrote:
> > On 25 Oct 2022 06:27, Tsukasa OI wrote:
> >> Because of Binutils/GCC hack, getopt on GNU libc (2.25 or earlier) is
> >> currently unusable on sim, causing a regression on CentOS 7.
> >>
> >> This is caused as follows:
> >>
> >> 1.  If HAVE_DECL_GETOPT is defined (getopt with known prototype is
> >>     declared), a declaration of getopt in "include/getopt.h" is suppressed.
> >>     The author started to define HAVE_DECL_GETOPT in sim with the commit
> >>     340aa4f6872c ("sim: Check known getopt definition existence").
> >> 2.  GNU libc (2.25 or earlier)'s <unistd.h> includes <getopt.h> to declare
> >>     getopt function (only, not getopt_long or getopt_long_only) but it
> >>     causes <unistd.h> to include Binutils/GCC's "include/getopt.h".
> >> 3.  If both 1. and 2. are satisfied, despite that <unistd.h> tries to
> >>     declare getopt by including <getopt.h>, "include/getopt.h" does not
> >>     define one, causing getopt function unusable.
> >>
> >> Getting rid of "include/getopt.h" (e.g. renaming this header file) is the
> >> best solution to avoid hacking but as a short-term solution, this commit
> >> replaces getopt with getopt_long under sim/.
> >> ---
> >>  sim/igen/igen.c | 6 ++++--
> >>  sim/m32c/main.c | 5 ++++-
> >>  sim/ppc/dgen.c  | 6 ++++--
> >>  sim/ppc/igen.c  | 9 ++++++---
> >>  sim/rl78/main.c | 4 +++-
> >>  5 files changed, 21 insertions(+), 9 deletions(-)
> >>
> >> diff --git a/sim/igen/igen.c b/sim/igen/igen.c
> >> index ba856401fa9..22cfd30ec43 100644
> >> --- a/sim/igen/igen.c
> >> +++ b/sim/igen/igen.c
> >> @@ -989,6 +989,7 @@ main (int argc, char **argv, char **envp)
> >>    char *real_file_name = NULL;
> >>    int is_header = 0;
> >>    int ch;
> >> +  struct option dummy_longopts = { 0 };
> > 
> > just call it "longopts" so we don't have to rename it in the future if we
> > decide to actually add long options.  comes up in the other files too.
> > 
> > otherwise lgtm.
> > -mike
> 
> To prepare actual long options, not just renaming, making them array of
> struct option seems better.  Moving longopts out from the caller is
> avoided for now (since it might not get big so that it requires option
> definition outside a function).  That means...
> 
> Before: struct option dummy_longopts = { 0 };
> After:  struct option longopts[] = { { 0 } };
> 
> I fixed like this and I'll submit PATCH v3 (with fix above and minor
> commit message clarification) soon.  Finally, I can clean up the mess I
> created.

i would make it static const too in that case so it isn't constructed on
the stack ... it's just part of the rodata and loaded right away.

> Ah, a minor question to you, Mike.  Can I consider your "lgtm" as an
> approval for specific area you are responsible?

"lgtm" is equiv to "approved, please push when you're ready"
-mike