Bogus job number

Bogus job number

[Peter Stephenson]

Reporting for form's sake, though I realise this isn't very useful as it
stands...  After suspending and resuming (in the foreground) a
CPU-intensive make, I got

 jobs.c:950: bogus job number, jn = 140813044, jobtab = 140813008, oldjobtab = 0
[1]  + continued  make...

I haven't seen this before, so I guess we've managed to introduce a new
race to make up for all the old ones that have gone...

pws

Re: Bogus job number

[Bart Schaefer]

On Oct 29,  3:23pm, Peter Stephenson wrote:
} Subject: Bogus job number
}
} Reporting for form's sake, though I realise this isn't very useful as it
} stands...  After suspending and resuming (in the foreground) a
} CPU-intensive make, I got
} 
}  jobs.c:950: bogus job number, jn = 140813044, jobtab = 140813008, oldjobtab = 0
} [1]  + continued  make...
} 
} I haven't seen this before, so I guess we've managed to introduce a new
} race to make up for all the old ones that have gone...

Well, it could just be that the condition tested in DPUTS3() is wrong.
This is coming from:

    944     if (synch > 1 && oldjobtab != NULL)
    945         job = jn - oldjobtab;
    946     else
    947         job = jn - jobtab;
    948     DPUTS3(job < 0 || job > (synch > 1 ? oldmaxjob : maxjob),
    949            "bogus job number, jn = %L, jobtab = %L, oldjobtab = %L",
    950            (long)jn, (long)jobtab, (long)oldjobtab);

And yes, I immediately see a problem with that.


diff --git a/Src/jobs.c b/Src/jobs.c
index 336c5d4..371b8eb 100644
--- a/Src/jobs.c
+++ b/Src/jobs.c
@@ -945,7 +945,7 @@ printjob(Job jn, int lng, int synch)
 	job = jn - oldjobtab;
     else
 	job = jn - jobtab;
-    DPUTS3(job < 0 || job > (synch > 1 ? oldmaxjob : maxjob),
+    DPUTS3(job < 0 || job > (oldjobtab && synch > 1 ? oldmaxjob : maxjob),
 	   "bogus job number, jn = %L, jobtab = %L, oldjobtab = %L",
 	   (long)jn, (long)jobtab, (long)oldjobtab);
 

-- 
Barton E. Schaefer