ash: waiting machinery: add comments explaining the code

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
This commit is contained in:
Denys Vlasenko committed 2026-10-05 17:40:47 +02:00
1 parent cdd442ebd2
commit e054f9fb4b
1 file changed
+20 -3
+20 -3
View File
@@ -4392,7 +4392,7 @@ static int waitone(int block, struct job *job)
block = (block & ~DOWAIT_JOBSTATUS);
#endif
TRACE(("dowait(0x%x) called\n", block));
TRACE(("waitone(0x%x) called\n", block));
/* It's wrong to call waitpid() outside of INTOFF region:
* signal can arrive just after syscall return and handler can
@@ -4414,7 +4414,7 @@ static int waitone(int block, struct job *job)
*/
INTOFF;
pid = waitproc(block, &status);
TRACE(("wait returns pid %d, status=%d\n", pid, status));
TRACE(("waitproc returns pid %d, status=%d\n", pid, status));
if (pid <= 0)
goto out;
@@ -4504,14 +4504,31 @@ static int dowait(int block, struct job *jp)
rpid = 1;
/* In what cases we loop here:
* = If we waited in blocking mode and got pid > 0,
* but the job we waiting for is still running
* (jp->state == JOBRUNNING), we have to wait in blocking mode again,
* until it's not JOBRUNNING.
* The user is waitforjob(jp) -> dowait(DOWAIT_BLOCK, jp).
* = If we waited in non-blocking mode and got pid > 0,
* we should see whether more processes have terminated,
* we have to wait in non-blocking mode again.
* The users are
* showjobs() -> dowait(DOWAIT_NONBLOCK)
* waitforjob(NULL) -> dowait(DOWAIT_NONBLOCK)
* waitcmd() does its own looping while it sees at least one JOBRUNNING
* job, it does not need us to loop here - only to do
* one nonblocking wait and if it gives nothing, wait for signals.
*/
do {
pid = waitone(block, jp);
rpid &= !!pid;
if (!pid || (jp && jp->state != JOBRUNNING))
block = DOWAIT_NONBLOCK;
} while (pid >= 0);
} while (pid >= 0); //WRONG? should be "pid > 0"
/* Return: 0 if seen a signal, 1 otherwise */
return rpid;
}