setitimer() bug? (Linux, glibc 2.2)

[Michael Kerrisk]

Gidday,

I'm not sure whether this is the right place to file this or not, but here 
goes:

I've a question for you of a slightly different type.  Austin d5 (and the 
older SUSv2) both say

17221 The setitimer( ) function shall fail if:
17222 [EINVAL] The value argument is not in canonical form. (In canonical 
form, the number of
17223 microseconds is a non-negative integer less than 1,000,000 and the 
number of
17224 seconds is a non-negative integer.)

Now currently, setitimer does not fail in this case.  Specifying usecs > 
1000000 does the sensible, but non-conformant thing - i.e. translates to 
corresponding seconds.  Specifying a negative seconds results in rounding 
down (unsigned treatment) to 2gig jiffies within the kernel.

Now I notice that there does not seem to be a setitimer() for Linux in 
glibc 2.2.  So it is not clear to me whether the bug report should be to 
you folks (i.e. so that an API with appropriate checking should be added 
to glibc) or, should I file it elsewhere (i.e. Linux kernel dev.)

Cheers

Michael

__________________________________________
Michael Kerrisk
mailto: address@hidden

"I can't believe that every time you hear a newborn scream,
you just can't see the shaping of a life."

                                             Roland Orzabel