0000959: struct siginfo member si_pid is marked "CX" but is mandatory for the non-CX waitid()

(0002715)
joerg (reporter)
2015-06-18 11:11
edited on: 2015-06-18 11:11

Here is a test program that allows to check whether a implementation
conforms to the standard:

#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
#include <stdio.h>
/*
* Non-standard compliant platforms may need
* #include <signal.h> or something similar
* in addition to the include files above.
*/

int
main()
{
        siginfo_t si;
        pid_t pid;
        int ret;

        if ((pid = fork()) < 0)
                exit(1);
        if (pid == 0) {
                _exit(1234567890);
        }
        ret = waitid(P_PID, pid, &si, WEXITED);
        printf("ret: %d si_pid: %ld si_status: %d si_code: %d\n",
                ret,
                (long) si.si_pid, si.si_status, si.si_code);
        if (pid != si.si_pid)
                printf("si_pid in struct siginfo should be %ld but is %ld\n",
                        (long) pid, (long) si.si_pid);
        if (si.si_status != 1234567890)
                printf("si_status in struct signinfo should be %d (0x%x) but is %d (0x%x)\n",
                        1234567890, 1234567890, si.si_status, si.si_status);
        if (CLD_EXITED != 1)
                printf("CLD_EXITED is %d on this platform\n", CLD_EXITED);
        return (0);
}

(0002716)
geoffclare (manager)
2015-06-18 11:32

This bug report seems to be the result of a misunderstanding of what the CX margin code means. Hint: it does not indicate that something is optional!

(0002717)
joerg (reporter)
2015-06-18 11:46

So do you believe that this is just an extension to ISO C but
mandatory for POSIX?

The interesting thing here is that e.g. Mac OS X returns si_pid == 0
for waitid() and does not seem to implement a useable waitid()
interface.

(0002722)
eblake (manager)
2015-06-18 18:36

See XBD 1.7.1 line 182. Something marked CX is mandatory in POSIX, and merely marked as a convenience to show that it is an extension to C99.

(0002753)
rhansen (manager)
2015-07-10 21:48
edited on: 2015-07-10 21:48

> So do you believe that this is just an extension to ISO C but
> mandatory for POSIX?

Yes. Key quotes from the section Eric mentioned (XBD 1.7.1, definition of CX shading):

"Application developers may make use of an extension as it is supported on all POSIX.1-2008-conforming systems." (page 7 lines 183-184)

"The markings are for information only." (page 7 line 189)

So the CX shading mismatch is not a bug.

Note that 0000690 changed some shading; I don't recall whether there's any overlap with this bug report.

Issue History
Date Modified	Username	Field	Change
2015-06-18 11:10	joerg	New Issue
2015-06-18 11:10	joerg	Status	New => Under Review
2015-06-18 11:10	joerg	Assigned To	=> ajosey
2015-06-18 11:10	joerg	Name	=> Jörg Schilling
2015-06-18 11:10	joerg	Section	=> Headers
2015-06-18 11:10	joerg	Page Number	=> 335-336
2015-06-18 11:10	joerg	Line Number	=> 11179, 11185, 11218-11223
2015-06-18 11:11	joerg	Note Added: 0002715
2015-06-18 11:11	joerg	Note Edited: 0002715
2015-06-18 11:32	geoffclare	Note Added: 0002716
2015-06-18 11:46	joerg	Note Added: 0002717
2015-06-18 18:36	eblake	Note Added: 0002722
2015-07-10 21:39	rhansen	Relationship added	related to 0000690
2015-07-10 21:48	rhansen	Note Added: 0002753
2015-07-10 21:48	rhansen	Note Edited: 0002753
2015-07-30 15:04	Don Cragun	Interp Status	=> ---
2015-07-30 15:04	Don Cragun	Status	Under Review => Closed
2015-07-30 15:04	Don Cragun	Resolution	Open => Withdrawn

Notes
(0002715) joerg (reporter) 2015-06-18 11:11 edited on: 2015-06-18 11:11	Here is a test program that allows to check whether a implementation conforms to the standard: #include <stdlib.h> #include <unistd.h> #include <sys/wait.h> #include <stdio.h> /* * Non-standard compliant platforms may need * #include <signal.h> or something similar * in addition to the include files above. */ int main() { siginfo_t si; pid_t pid; int ret; if ((pid = fork()) < 0) exit(1); if (pid == 0) { _exit(1234567890); } ret = waitid(P_PID, pid, &si, WEXITED); printf("ret: %d si_pid: %ld si_status: %d si_code: %d\n", ret, (long) si.si_pid, si.si_status, si.si_code); if (pid != si.si_pid) printf("si_pid in struct siginfo should be %ld but is %ld\n", (long) pid, (long) si.si_pid); if (si.si_status != 1234567890) printf("si_status in struct signinfo should be %d (0x%x) but is %d (0x%x)\n", 1234567890, 1234567890, si.si_status, si.si_status); if (CLD_EXITED != 1) printf("CLD_EXITED is %d on this platform\n", CLD_EXITED); return (0); }

(0002716) geoffclare (manager) 2015-06-18 11:32	This bug report seems to be the result of a misunderstanding of what the CX margin code means. Hint: it does not indicate that something is optional!

(0002717) joerg (reporter) 2015-06-18 11:46	So do you believe that this is just an extension to ISO C but mandatory for POSIX? The interesting thing here is that e.g. Mac OS X returns si_pid == 0 for waitid() and does not seem to implement a useable waitid() interface.

(0002722) eblake (manager) 2015-06-18 18:36	See XBD 1.7.1 line 182. Something marked CX is mandatory in POSIX, and merely marked as a convenience to show that it is an extension to C99.

(0002753) rhansen (manager) 2015-07-10 21:48 edited on: 2015-07-10 21:48	> So do you believe that this is just an extension to ISO C but > mandatory for POSIX? Yes. Key quotes from the section Eric mentioned (XBD 1.7.1, definition of CX shading): "Application developers may make use of an extension as it is supported on all POSIX.1-2008-conforming systems." (page 7 lines 183-184) "The markings are for information only." (page 7 line 189) So the CX shading mismatch is not a bug. Note that 0000690 changed some shading; I don't recall whether there's any overlap with this bug report.

Aardvark Mark IV