On 2012年09月04日 14:35, Jan Stancek wrote:
>
> ----- Original Message -----
>> From: "Kang Kai"<[email protected]>
>> To: "Kang Kai"<[email protected]>
>> Cc: [email protected], "Zhenfeng
>> Zhao"<[email protected]>
>> Sent: Monday, 3 September, 2012 10:45:58 AM
>> Subject: Re: [LTP] [PATCH 2/3] mq_open/16-1: use tmp file to share info
>>
>> On 2012年08月09日 18:30, Kang Kai wrote:
>>
>> Hi Wanlong,
>>
>>> In this test case, it uses a variable to share data between child
>>> and
>>> parent processes. But after fork there is a copy of the variable in
>>> child process and modify it will not affect the variable in the
>>> parent
>>> process. Then when the child process call mq_open() before parent
>>> process,
>>> the case will fail.
>>>
>>> Use tmp file to replace the variable. Any modification in child
>>> process
>>> can be seen in parent process.
>> Any comment for this patch?
> I had one question, see here:
> http://article.gmane.org/gmane.linux.ltp/16611
>
> Regards,
> Jan
>
>> Thanks,
>> Kai
>>
>>> Signed-off-by: Kang Kai<[email protected]>
>>> ---
>>> .../conformance/interfaces/mq_open/16-1.c | 38
>>> +++++++++++++++++--
>>> 1 files changed, 34 insertions(+), 4 deletions(-)
>>>
>>> diff --git
>>> a/testcases/open_posix_testsuite/conformance/interfaces/mq_open/16-1.c
>>> b/testcases/open_posix_testsuite/conformance/interfaces/mq_open/16-1.c
>>> index b9a3215..ebd697b 100644
>>> ---
>>> a/testcases/open_posix_testsuite/conformance/interfaces/mq_open/16-1.c
>>> +++
>>> b/testcases/open_posix_testsuite/conformance/interfaces/mq_open/16-1.c
>>> @@ -20,9 +20,11 @@
>>> * this is fine (will have some false positives, but no false
>>> negatives).
>>> */
>>>
>>> +#include<sys/mman.h>
>>> #include<sys/stat.h>
>>> #include<sys/types.h>
>>> #include<sys/wait.h>
>>> +#include<errno.h>
>>> #include<fcntl.h>
>>> #include<mqueue.h>
>>> #include<signal.h>
>>> @@ -32,11 +34,15 @@
>>> #include "posixtest.h"
>>>
>>> #define NAMESIZE 50
>>> +#define TNAME "mq_open/16-1.c"
>>>
>>> int main()
>>> {
>>> char qname[NAMESIZE];
>>> + char fname[NAMESIZE];
>>> int pid, succeeded=0;
>>> + int fd;
>>> + void *pa = NULL;
>>> mqd_t childqueue, queue;
>>>
>>> /*
>>> @@ -47,6 +53,26 @@ int main()
>>>
>>> sprintf(qname, "/mq_open_16-1_%d", getpid());
>>>
>>> + sprintf(fname, "/tmp/pts_mq_open_16_1_%d", getpid());
>>> + unlink(fname);
>>> + fd = open(fname, O_CREAT | O_RDWR | O_EXCL,
>>> + S_IRUSR | S_IWUSR);
>>> + if (fd == -1) {
>>> + printf(TNAME " Error at open(): %s\n", strerror(errno));
>>> + exit(PTS_UNRESOLVED);
>>> + }
Hi Jan,
Copy your comment here
>>> + /* file is empty now, will cause "Bus error" */
>>> + write(fd, "\0", 1);
| Isn't there a risk of SIGBUS, when the file length is just 1 byte
| and it's mapped to sizeof(int)?
|
| SIGBUS Attempted access to a portion of the buffer that does not correspond
to the file (for example, beyond
| the end of the file, including the case where another process has
truncated the file).
I think why it doen't fail:
mmap implemented on linux willl map one page size if length pass to mmap less
than one page size. So we can use the memory in this one page size.
But empty file is special case cause SIGBUS.
>>> + unlink(fname);
>>> +
>>> + pa = mmap(NULL, sizeof(int), PROT_READ | PROT_WRITE, MAP_SHARED,
>>> fd, 0);
>>> + if (pa == MAP_FAILED) {
>>> + printf(TNAME " Error at mmap: %s\n", strerror(errno));
>>> + close(fd);
>>> + exit(PTS_FAIL);
>>> + }
I add test code here:
>>> + *(int *)pa = 0;
>>> +
int *pi = (int *)pa;
++pi;
*pi = 4;
It still doesn't fail.
So that may implement dependent. I will update the patch to write a int
to the file.
Regards,
Kai
>>> if ((pid = fork()) == 0) {
>>> sigset_t mask;
>>> int sig;
>>> @@ -62,7 +88,7 @@ int main()
>>> childqueue = mq_open(qname, O_CREAT|O_EXCL|O_RDWR,
>>> S_IRUSR | S_IWUSR, NULL);
>>> if (childqueue != (mqd_t)-1) {
>>> - succeeded++;
>>> + ++*(int *)pa;
>>> #ifdef DEBUG
>>> printf("mq_open() in child succeeded\n");
>>> } else {
>>> @@ -79,7 +105,7 @@ int main()
>>> queue = mq_open(qname, O_CREAT | O_EXCL |O_RDWR,
>>> S_IRUSR | S_IWUSR, NULL);
>>> if (queue != (mqd_t)-1) {
>>> - succeeded++;
>>> + ++*(int *)pa;
>>> #ifdef DEBUG
>>> printf("mq_open() in parent succeeded\n");
>>> } else {
>>> @@ -93,13 +119,15 @@ int main()
>>> mq_close(queue);
>>> mq_close(childqueue);
>>> mq_unlink(qname);
>>> + close(fd);
>>> + munmap(pa, sizeof(int));
>>> return PTS_UNRESOLVED;
>>> }
>>>
>>> mq_close(queue);
>>> mq_close(childqueue);
>>> mq_unlink(qname);
>>> -
>>> + succeeded = *(int *)pa;
>>> if (succeeded==0) {
>>> printf("Test FAILED - mq_open() never succeeded\n");
>>> return PTS_FAIL;
>>> @@ -111,8 +139,10 @@ int main()
>>> }
>>>
>>> printf("Test PASSED\n");
>>> + close(fd);
>>> + munmap(pa, sizeof(int));
>>> return PTS_PASS;
>>> }
>>>
>>> return PTS_UNRESOLVED;
>>> -}
>>> \ No newline at end of file
>>> +}
>>
>> ------------------------------------------------------------------------------
>> Live Security Virtual Conference
>> Exclusive live event will cover all the ways today's security and
>> threat landscape has changed and how IT managers can respond.
>> Discussions
>> will include endpoint security, mobile security and the latest in
>> malware
>> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>> _______________________________________________
>> Ltp-list mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/ltp-list
>>
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
