[shepherd] branch master updated: service: Fix 'make-kill-destructor' when PGID is zero.

[Ludovic Courtès]

This is an automated email from the git hooks/post-receive script.

civodul pushed a commit to branch master
in repository shepherd.

The following commit(s) were added to refs/heads/master by this push:
     new d190773  service: Fix 'make-kill-destructor' when PGID is zero.
d190773 is described below

commit d190773751ddeddbe0daa8e4a43f76b73c4fd7ac
Author: Mathieu Othacehe <address@hidden>
AuthorDate: Thu May 7 18:39:41 2020 +0200

    service: Fix 'make-kill-destructor' when PGID is zero.
    
    Fixes <https://bugs.gnu.org/40981>.
    
    When a process is forked, and before its GID is changed in "exec-command",
    it will share the parent GID, which is 0 for Shepherd. In that case, use
    the PID instead of the PGID.
    
    Also make sure to reset the SIGTERM handler before forking a process. 
Failing
    to do so, will result in stopping Shepherd if a SIGTERM is received between
    fork and execv calls. Restore the SIGTERM handler once the process has been
    forked.
    
    * modules/shepherd/service.scm (fork+exec-command): Save the current SIGTERM
    handler and reset it before forking. Then, restore it on the parent after
    forking.
    (make-kill-destructor): Handle the case when PGID is zero, between the 
process
    fork and exec.
    
    Signed-off-by: Ludovic Courtès <address@hidden>
---
 modules/shepherd/service.scm | 33 ++++++++++++++++++++++++++-------
 tests/forking-service.sh     | 18 ++++++++++++++++++
 2 files changed, 44 insertions(+), 7 deletions(-)

diff --git a/modules/shepherd/service.scm b/modules/shepherd/service.scm
index 8604d2f..45fcf32 100644
--- a/modules/shepherd/service.scm
+++ b/modules/shepherd/service.scm
@@ -5,6 +5,7 @@
 ;; Copyright (C) 2016 Alex Kost <address@hidden>
 ;; Copyright (C) 2018 Carlo Zancanaro <address@hidden>
 ;; Copyright (C) 2019 Ricardo Wurmus <address@hidden>
+;; Copyright (C) 2020 Mathieu Othacehe <address@hidden>
 ;;
 ;; This file is part of the GNU Shepherd.
 ;;
@@ -884,7 +885,18 @@ its PID."
   (unless %sigchld-handler-installed?
     (sigaction SIGCHLD handle-SIGCHLD SA_NOCLDSTOP)
     (set! %sigchld-handler-installed? #t))
-  (let ((pid (primitive-fork)))
+
+  ;; When forking a process, the signal handlers are inherited, until it
+  ;; forks. If SIGTERM is received by the forked process, before it calls
+  ;; execv, the installed SIGTERM handler, stopping Shepherd will be called.
+  ;; To avoid this, save the SIGTERM handler, disable it, and restore it once,
+  ;; the process has been forked. This way, the forked process will use the
+  ;; default SIGTERM handler stopping the process.
+  (let ((term-handler (match (sigaction SIGTERM)
+                        ((proc . _)
+                         proc)))
+        (pid (and (sigaction SIGTERM SIG_DFL)
+                  (primitive-fork))))
     (if (zero? pid)
         (exec-command command
                       #:user user
@@ -893,7 +905,10 @@ its PID."
                       #:directory directory
                       #:file-creation-mask file-creation-mask
                       #:environment-variables environment-variables)
-        pid)))
+        (begin
+          ;; Restore the initial SIGTERM handler.
+          (sigaction SIGTERM term-handler)
+          pid))))
 
 (define* (make-forkexec-constructor command
                                     #:key
@@ -957,11 +972,15 @@ start."
   "Return a procedure that sends SIGNAL to the process group of the PID given
 as argument, where SIGNAL defaults to `SIGTERM'."
   (lambda (pid . args)
-    ;; Kill the whole process group PID belongs to.  Don't assume that PID
-    ;; is a process group ID: that's not the case when using #:pid-file,
-    ;; where the process group ID is the PID of the process that
-    ;; "daemonized".
-    (kill (- (getpgid pid)) signal)
+    ;; Kill the whole process group PID belongs to.  Don't assume that PID is
+    ;; a process group ID: that's not the case when using #:pid-file, where
+    ;; the process group ID is the PID of the process that "daemonized".  If
+    ;; this procedure is called, between the process fork and exec, the PGID
+    ;; will still be zero (the Shepherd PGID). In that case, use the PID.
+    (let ((pgid (getpgid pid)))
+      (if (= (getpgid 0) pgid)
+          (kill pid signal) ;don't kill ourself
+          (kill (- pgid) signal)))
     #f))
 
 ;; Produce a constructor that executes a command.
diff --git a/tests/forking-service.sh b/tests/forking-service.sh
index 7126c3d..bd9aac9 100644
--- a/tests/forking-service.sh
+++ b/tests/forking-service.sh
@@ -71,6 +71,17 @@ cat > "$conf"<<EOF
                                       #:pid-file "$PWD/$service2_pid")
    #:stop  (make-kill-destructor)
    #:respawn? #t))
+
+(define %command3
+  '("$SHELL" "-c" "sleep 600"))
+
+(register-services
+ (make <service>
+   ;; A service that forks into a different process.
+   #:provides '(test3)
+   #:start (make-forkexec-constructor %command3)
+   #:stop  (make-kill-destructor)
+   #:respawn? #t))
 EOF
 cat $conf
 
@@ -113,3 +124,10 @@ sleep 1;
 $herd status test2 | grep started
 test "`cat $PWD/$service2_started`" = "started
 started"
+
+# Try to trigger eventual race conditions, when killing a process between fork
+# and execv calls.
+for i in `seq 1 50`
+do
+    $herd restart test3
+done