Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!umd5!vrdxhq!daitc!csed-1!roskos
From: roskos@csed-1.IDA.ORG (Eric Roskos)
Newsgroups: comp.os.minix
Subject: Re: V1.3 posting # 15 - test1.c and test3.c
Message-ID: <385@csed-47.csed-1.IDA.ORG>
Date: 23 Jun 88 17:51:43 GMT
References: <800@ast.cs.vu.nl> <2879@cs.utexas.edu>
Organization: IDA, Alexandria, VA
Lines: 19

> Test 5 would hang and never come back.

I reported a bug with this about 6 months ago, when I first tried running the
tests on version 1.2.  The problem was caused by the wrong field name being
used in one of the messages being passed between mm and the clock task.  If
I remember correctly, it is because line 6706 uses m_sig.PROC_NR (which is
field m2_i2) and line 3156 uses m_ptr->CLOCK_PROC_NR, which is field
m6_i1.  This caused all alarm signals to get sent to the wrong process.
I don't have the diff listing any more, but I think I just changed line 
6706 to m_sig.CLOCK_PROC_NR and it fixed it.  It took a couple of days to
track it down.
-- 
Eric Roskos, IDA (...daitc!csed-1!roskos, or csed-1!roskos@DAITC.ARPA, or
		     Roskos@DOCKMASTER.ARPA)

"... should have risen as noiselessly as the sun out of the sea, and
     we sailors have found ourselves steering between the tropics as if
     the broad day had lasted forever.  You know how the sun comes up 
     from the sea when you stand on the cliff, and doesn't startle you,
     but every thing, and you too are helping it."