Re: [Lse-tech] Re: [PATCH]updated ipc lock patch

[Manfred Spraul <manfred@colorfullife.com>]

[-- Attachment #1: Type: text/plain, Size: 918 bytes --]

mingming cao wrote:

>Here is what I saw:
>
>[root@elm3b83 shmctl]# ./shmctl01
>shmctl01    1  PASS  :  pid, size, # of attaches and mode are correct -
>pass #1
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>
These failures are caused by a bug in the ltp test. See the attached patch.

>shmctl01    2  PASS  :  pid, size, # of attaches and mode are correct -
>pass #2
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    0  INFO  :  shmdt() failed - 22
>shmctl01    3  FAIL  :  # of attaches is incorrect - 0
>
This one is odd. The testcase contains races, but they can only increase 
# of attaches.
Could you strace shmctl01?
The testcase with shmat(), then fork() fails.

--
    Manfred

[-- Attachment #2: patch-ltp --]
[-- Type: text/plain, Size: 714 bytes --]

diff -u ltp-orig/testcases/kernel/syscalls/ipc/shmctl/shmctl01.c ltp-20021008/testcases/kernel/syscalls/ipc/shmctl/shmctl01.c
--- ltp-orig/testcases/kernel/syscalls/ipc/shmctl/shmctl01.c	Tue May 21 15:55:56 2002
+++ ltp-20021008/testcases/kernel/syscalls/ipc/shmctl/shmctl01.c	Fri Oct 25 22:14:23 2002
@@ -252,10 +252,12 @@
 
 			if (stat_time == FIRST) {
 				test = set_shmat();
+			} else {
+				test = set_shared;
 			}
 
 			/* do an assignement for fun */
-			(int *)test = i;
+			*(int *)test = i;
 
 			/* pause until we get a signal from stat_cleanup() */
 			rval = pause();
@@ -273,7 +275,7 @@
 		}
 	}
 	/* sleep briefly to ensure correct execution order */
-	usleep(25000);
+	usleep(250000);
 }
 
 /*