1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
|
<html>
<head>
<title>Diffs on diffs :-) by Larry V. Streepy, Jr.</title>
<base target=_top>
<meta name="GENERATOR" content="FindMail Communications">
<meta name="Date" content="Monday, October 07, 1996 02:03 PM PST">
<meta name="Author" content="Larry V. Streepy, Jr.">
</head>
<body background="/gifs/betafm.gif" bgcolor="#ffffff" text="#000000" link="#0000ee" vlink="#ff0000" alink="#000099">
<h3>Diffs on diffs :-)</h3>
Larry V. Streepy, Jr. (<a href="mailto.html?mid=2079859748&num=398" target="_top">@healthcare.com</a>)<br>Monday, October 07, 1996 02:03 PM PST<br>
<p>
This is a multi-part message in MIME format.<p>
--------------65BE18E23639BCDD7BE55F7F <br>
Content-Type: text/plain; charset=us-ascii <br>
Content-Transfer-Encoding: 7bit<p>
Unfortunately, there are a couple of bugs in my pthread_cancel support (no, say it isn't so :-)<p>
Oh well, I cam across a couple of cases that I missed in my testing last week. Here are the bugs:<p>
1. If a thread calls pthread_testcancel during it's cleanup processing after being cancelled, the pthread kernel would hang.<p>
2. I didn't realize that threads in PS_SLEEP_WAIT state are *NOT* on any queue, they are handled using a linked list. So, when cancelling a thread that was sleeping, a PANIC() I put in possiblymakeRunnable would go off.<p>
Both of these are fixed. The diffs are attached.<br>
-- <br>
Larry V. Streepy, Jr. <br>
Chief Technical Officer, Healthcare Communications, Inc. mailto:<a href="mailto.html?mid=2079859748&num=398" target="_top">@healthcare.com</a> <br>
(214) 851-7033 (Dallas Main #) <br>
(970) 626-5028 (My office #) (970) 626-4425 (Fax)<p>
--------------65BE18E23639BCDD7BE55F7F <br>
Content-Type: text/plain; charset=us-ascii; name="cancel.diffs" Content-Transfer-Encoding: 7bit <br>
Content-Disposition: inline; filename="cancel.diffs"<p>
Index: pthread_cancel.c<br>
=================================================================== RCS file: /usr/cvssrc/pthreads-1_60_beta5/pthreads/pthread_cancel.c,v retrieving revision 1.1 <br>
diff -c -r1.1 pthread_cancel.c<br>
*** pthread_cancel.c 1996/10/06 00:31:27 1.1<br>
--- pthread_cancel.c 1996/10/07 18:33:27<br>
***************<br>
*** 187,192 ****<br>
--- 187,197 ----<br>
return; /* Can't be cancelled */<br>
}<br>
<br>
+ /* Ensure that we aren't in the process of exiting already */<br>
+ if( TEST_PF_RUNNING_TO_CANCEL(pthread_run) ) {<br>
+ return;<br>
+ }<br>
+ <br>
/* See if we have been cancelled */<br>
if( TEST_PF_CANCELLED(pthread_run) ) {<br>
/* Set this flag to avoid recursively calling pthread_exit */<br>
***************<br>
*** 266,277 ****<br>
if( pthread->state == PS_RUNNING )<br>
return; /* will happen at context switch */<br>
<br>
! /* Otherwise, we need to take it off the queue and make it runnable */<br>
! if( pthread->queue == NULL ) {<br>
! PANIC(); /* Must be on a queue */<br>
! }<br>
<br>
- pthread_queue_remove(pthread->queue, pthread);<br>
pthread_prio_queue_enq(pthread_current_prio_queue, pthread);<br>
pthread->old_state = pthread->state;<br>
pthread->state = PS_RUNNING;<br>
--- 271,291 ----<br>
if( pthread->state == PS_RUNNING )<br>
return; /* will happen at context switch */<br>
<br>
! /* If the thread is sleeping, the it isn't on a queue. */<br>
! if( pthread->state == PS_SLEEP_WAIT ) {<br>
! sleep_cancel( pthread ); /* Remove from sleep list */<br>
! } else {<br>
! /* Otherwise, we need to take it off the queue and make it runnable */<br>
! <br>
! if( pthread->queue == NULL ) {<br>
! PANIC(); /* Must be on a queue */<br>
! }<br>
! <br>
! pthread_queue_remove(pthread->queue, pthread);<br>
! }<br>
! <br>
! /* And make it runnable */<br>
<br>
pthread_prio_queue_enq(pthread_current_prio_queue, pthread);<br>
pthread->old_state = pthread->state;<br>
pthread->state = PS_RUNNING;<p>
<pre><font size=-1>
--------------65BE18E23639BCDD7BE55F7F--
</pre><p></pre>
<hr>
<a href="http://www.findmail.com/" target="_top"><font size=-1>Powered by FindMail Communications</font><br></a>
<br>Please email comments and suggestions to:<a href="/cgi-bin/comments.py" target="_top">comments@findmail.com</a>
<br><font size=-3 color="#ffffff">xmlarchive</font>
</body></html>
|
