syskill.o unconditionally provides 'kill', but signal.h conditionally declares it

[James Greenhalgh]

Hi,

I'm trying to fix the configurey that libgfortran uses to detect what
features newlib supplies [1], and I'm running in to an issue very similar
to that Ulrichi/Corinna fixed for SPU here:

  https://sourceware.org/ml/newlib/2014/msg00630.html

I see "kill" autodetected by the libgfortran configure, but when
libgfortran tries to use it after including signal.h it does not find
a declaration of the function and errors. readelf tells me
that this is being provided by syskill.c:

     0: 0000000000000000     0 NOTYPE  LOCAL  DEFAULT  UND 
     1: 0000000000000000     0 FILE    LOCAL  DEFAULT  ABS syskill.c
     2: 0000000000000000     0 SECTION LOCAL  DEFAULT    1 
     3: 0000000000000000     0 SECTION LOCAL  DEFAULT    3 
     4: 0000000000000000     0 SECTION LOCAL  DEFAULT    4 
     5: 0000000000000000     0 NOTYPE  LOCAL  DEFAULT    1 $x
     6: 0000000000000000     0 SECTION LOCAL  DEFAULT    5 
     7: 0000000000000000     0 SECTION LOCAL  DEFAULT    7 
     8: 0000000000000000     0 SECTION LOCAL  DEFAULT    8 
     9: 0000000000000000     0 SECTION LOCAL  DEFAULT    9 
    10: 0000000000000000     0 SECTION LOCAL  DEFAULT   11 
    11: 0000000000000000     0 SECTION LOCAL  DEFAULT   13 
    12: 0000000000000010     0 NOTYPE  LOCAL  DEFAULT   15 $d
    13: 0000000000000000     0 SECTION LOCAL  DEFAULT   15 
    14: 0000000000000000     0 SECTION LOCAL  DEFAULT   14 
    15: 0000000000000000    20 FUNC    GLOBAL DEFAULT    1 kill
    16: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND _impure_ptr
    17: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND _kill_r

I guess this was never a problem in the past as libgfortran didn't build
with -Werror=implicit-function-declaration until late 2014, by which
point libgfortran's configure was hardcoding rather than detecting
presence of functions.

I'd like to move libgfortran back to auto-detecting what newlib is
providing for it, so it would seem to make sense to me to make that
declaration of 'kill' in libc/include/sys/signal.h unconditional.

On the other hand, I know very little about building newlib, and the
reasons for this currently being conditional. I've looked back at the
history, and it seems like it has been this way for a sufficiently long
time that it looks intentional.

What am I missing? If nothing, would a patch making the declaration of
kill unconditional be acceptable?

Thanks,
James

---
[1]: Along these lines: https://gcc.gnu.org/ml/gcc-patches/2015-08/msg01181.html