[SystemZ][z/OS] Add guard for SI_USER

[abhina-sree]

SI_USER is not defined on z/OS and is causing the following build failure. This patch adds a guard to check whether SI_USER is defined before using it.

error: use of undeclared identifier 'SI_USER'

[llvmorg-github-actions]

@llvm/pr-subscribers-llvm-support

Author: Abhina Sree (abhina-sree)

Changes

SI_USER is not defined on z/OS and is causing the following build failure. This patch adds a guard to check whether SI_USER is defined before using it.

error: use of undeclared identifier 'SI_USER'

---

Full diff: https://github.com/llvm/llvm-project/pull/213101.diff

1 Files Affected:

• (modified) llvm/lib/Support/Unix/Signals.inc (+4-1)

diff --git a/llvm/lib/Support/Unix/Signals.inc b/llvm/lib/Support/Unix/Signals.inc
index 69a80839c99d2..9453222cae546 100644
--- a/llvm/lib/Support/Unix/Signals.inc
+++ b/llvm/lib/Support/Unix/Signals.inc
@@ -477,7 +477,10 @@ static void SignalHandler(int Sig, siginfo_t *Info, void *Context) {
 #endif
 
   // Was the signal generated by kill(), sigqueue(), etc?
-  bool ReraiseSignal = Info->si_code == SI_USER || Info->si_code == SI_QUEUE;
+  bool ReraiseSignal = Info->si_code == SI_QUEUE;
+#if defined(SI_USER)
+  ReraiseSignal |= Info->si_code == SI_USER;
+#endif
 #if defined(SI_LWP)
   // _lwp_kill() on BSDs, Solaris/illumos, possibly others.
   ReraiseSignal |= Info->si_code == SI_LWP;

[alexrp]

Fairly unusual for SI_USER to not be defined. In fact, probably not POSIX-compliant. What does z/OS use when you use kill(2) to send a signal?

[abhina-sree]

Fairly unusual for SI_USER to not be defined. In fact, probably not POSIX-compliant. What does z/OS use when you use kill(2) to send a signal?

On z/OS, based on the documentation here I think si_code will be less than or equal to zero if the signal was generated by another process. Then si_pid and si_uid will have the process id and user id respectively

[alexrp]

If the signal came from another process -- i.e. SI_ASYNCIO or si_code <= 0 -- we definitely want to reraise it. Unfortunately that documentation doesn't really clarify what happens in the case of kill(getpid(), sig)...

[abhina-sree]

If the signal came from another process -- i.e. SI_ASYNCIO or si_code <= 0 -- we definitely want to reraise it. Unfortunately that documentation doesn't really clarify what happens in the case of kill(getpid(), sig)...

I agree, I'm still working on investigating what the proper fix should be. In the interim, is this patch ok to unblock our build? https://lab.llvm.org/staging/#/builders/237/builds/26