From: Bian N. <bi...@cn...> - 2011-01-25 03:12:31
|
Garrett Cooper wrote: > On Mon, Jan 24, 2011 at 1:58 AM, Bian Naimeng <bi...@cn...> wrote: >> >> Mitani wrote: >>> Hi, >>> >>> >>> Following POSIX testsets failed: >>> ------------ >>> conformance/interfaces/sched_getparam/6-1: execution: FAILED: Output: >>> Testing with user 'bin' (euid: 1)(uid: 1) >>> The function sched_getparam has successed. >>> ------------ >>> >>> ------------ >>> conformance/interfaces/sched_getscheduler/7-1: execution: FAILED: Output: >>> Testing with user 'bin' (uid: 1) >>> The function sched_getscheduler has successed. >>> ------------ >>> >>> The environments are as follows: >>> - RHEL4.8 --- (x86, x86_64, ia64) >>> - kernel --- kernel-2.6.9-89.EL >>> - glibc --- glibc-2.3.4-2.43 >>> >>> - RHEL5.5 --- (x86, x86_64, ia64) >>> - kernel --- kernel-2.6.18-194.el5 >>> - glibc --- glibc-2.5-49 >>> >>> >>> They are permission tests. >>> >>> The comment of "sched_getparam/6-1" testset source is as follows: >>> ------------ >>> * On Linux, e.g, the kernel makes no check on user permission to call this >>> * API. So basically we don't know on what condition a system should return >>> * EPERM. It is implementation defined. >>> ------------ >>> >> Hi Garrett, >> >> I think it's better to report UNRESOLVED, whatever the os is. > > It's not implementation defined according to 2008.1, and it's flat > out a requirement failure (not UNRESOLVED): > > The sched_getparam() function shall fail if: > > [EPERM] > The requesting process does not have permission to obtain the > scheduling parameters of the specified process. > > `shall' -> `must' in POSIX terms. Thanks very much for telling me this information. But the spec does not point out the what condition a process does not have permission to do it. > > The test is incorrect because it fails to setgid the group to the > non-root user's group. Ditto. The POSIX spec does not point out non-root user(non-root group too) does not have permission to obtain the scheduling parameters of the init process. Maybe i miss something. If so, please tell me, thanks very much. Thanks Bian > Thanks, > -Garrett > > -- Regards Bian Naimeng |