utils.c: include <signal.h> for siginfo_t

[aitap]

POSIX says:

The <signal.h> header shall define the siginfo_t type as a structure

So <sys/wait.h> is not enough to see the definition (not just a forward declaration) of siginfo_t.

Fixes: #7516

Tested manually on FreeBSD:

Details

$ R CMD check data.table_1.18.99.tar.gz --no-manual

* using R version 4.5.1 (2025-06-13)                                                   
* using platform: amd64-portbld-freebsd15.0                                            
* R was compiled by                                                                                                                                                            
    FreeBSD clang version 19.1.7 (https://github.com/llvm/llvm-project.git llvmorg-19.1.7-0-gcd708029e0b2)
    GNU Fortran (FreeBSD Ports Collection) 13.3.0                                      
* running under: FreeBSD freebsd 15.0-RELEASE FreeBSD 15.0-RELEASE releng/15.0-n280995-7aedc8de6446 GENERIC amd64
* checking whether package ‘data.table’ can be installed ... OK                        
* used C compiler: ‘FreeBSD clang version 19.1.7 (https://github.com/llvm/llvm-project.git llvmorg-19.1.7-0-gcd708029e0b2)’                                                    
* checking compiled code ... NOTE                                                      
File ‘data.table/libs/data_table.so’:   
  Found non-API calls to R: ‘IS_GROWABLE’, ‘LEVELS’, ‘SETLENGTH’,
    ‘SET_GROWABLE_BIT’, ‘SET_TRUELENGTH’, ‘TRUELENGTH’, ‘XTRUELENGTH’
Status: 1 NOTE                                                                         

...and OpenBSD:

Details

$ LANG=C.UTF-8 R CMD check data.table_1.18.99.tar.gz --no-manual

* R was compiled by                                                                    
    egcc (GCC) 8.4.0                                                                   
    GNU Fortran (GCC) 8.4.0                                                            
* running under: OpenBSD openbsd.my.domain 7.8 GENERIC#54 amd64                        

* checking whether package ‘data.table’ can be installed ... OK
* used C compiler: ‘OpenBSD clang version 19.1.7’

Status: OK

Also tested on GNU/Linux with PKG_CFLAGS=--std=c99.

@jszhao, could you please pull from this branch (posix_siginfo_t) and run R CMD build, then install the resulting tarball?

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.96%. Comparing base (3c06138) to head (a5a2644).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7517   +/-   ##
=======================================
  Coverage   98.96%   98.96%           
=======================================
  Files          87       87           
  Lines       16737    16737           
=======================================
  Hits        16563    16563           
  Misses        174      174           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

• HEAD=posix_siginfo_t stopped early for DT[by,verbose=TRUE] improved in #6296
• HEAD=posix_siginfo_t stopped early for isoweek improved in #7144
  

Generated via commit a5a2644

Download link for the artifact containing the test results: ↓ atime-results.zip

Task	Duration
R setup and installing dependencies	2 minutes and 44 seconds
Installing different package versions	45 seconds
Running and plotting the test cases	4 minutes and 29 seconds

[jszhao]

POSIX says:

The <signal.h> header shall define the siginfo_t type as a structure

So <sys/wait.h> is not enough to see the definition (not just a forward declaration) of siginfo_t.

Fixes: #7516

Tested manually on FreeBSD:
Details

...and OpenBSD:
Details

Also tested on GNU/Linux with PKG_CFLAGS=--std=c99.

@jszhao, could you please pull from this branch (posix_siginfo_t) and run R CMD build, then install the resulting tarball?

I have used R CMD INSTALL ... to install this branch (in fact, I am not familiar with github, so I just downloaded the branch with zip file, unzip, and tar it to tarball). It works. Thanks a lot for your effect to fix it for FreeBSD. Thanks again.

[MichaelChirico]

Great!

[MichaelChirico]

cc @aitap for extra eyes on a5a2644 -- should we add the same to the other jobs with -D_FORTIFY_SOURCE=2? just doing so in the strict job for now.

[aitap]

Ironically, setting -D_POSIX_C_SOURCE=200809L makes gettimeofday invisible in FreeBSD 15 (indeed, POSIX.1-2008 marks it as obsolete, recommending clock_gettime() instead):

user@freebsd:~ $ echo '#include <sys/time.h>' | cpp | grep gettimeofday                                   
int gettimeofday(struct timeval *, struct timezone *);
user@freebsd:~ $ echo '#include <sys/time.h>' | cpp -D_POSIX_C_SOURCE=200809L | grep gettimeofday || echo not found
not found

We use gettimeofday in wallclock. (The code that wants to use clock_gettime() deprives itself of the opportunity by testing for #ifdef CLOCK_REALTIME before a header file can #define it. A fix to that will need to be careful not to reintroduce #5374.)

Unlike _POSIX_C_SOURCE, which controls symbol visibility, _FORTIFY_SOURCE rewrites some function calls to perform stricter checks. Is the idea to make strict checks use #define _POSIX_C_SOURCE 200809L because it will hide non-standard or obsolete symbols?

[MichaelChirico]

I'm just looking for a way to fortify our checks such that we would have caught this issue before release

[aitap]

Then the strict checks must use --std=c99 to hide even more symbols, so that data.table could request some of them back using #define _POSIX_C_SOURCE=200809L:
https://gitlab.com/Rdatatable/data.table/-/commit/f02dcb783903727cbc043620f764b9d7f25c7af9/pipelines?ref=stricter_c99

(Right now the only failure is the -vanilla test, due to the C locale.)

Should I push f02dcb7 to this branch?

Drat. R itself uses configure-time tests for C23 features in R_ext/Rboolean.h, so when we later request C99 compatibility, this causes annoying warnings (because the configure-time tests aren't updated).