On Tue, Jul 7, 2015 at 9:23 AM, Ludovic Courtès <l...@gnu.org> wrote:
> David Thompson <dthomps...@worcester.edu> skribis:
>
>> From: David Thompson <da...@gnu.org>
>>
>> * guix/build/syscalls.scm (clone): New procedure.
>> (CLONE_NEWNS, CLONE_NEWUTS, CLONE_NEWIPC, CLONE_NEWUSER, CLONE_NEWPID,
>> CLONE_NEWNET): New variables.
>> * tests/syscalls.scm: Test it.
>
> [...]
>
>> +;; The libc interface to sys_clone is not useful for Scheme programs, so the
>> +;; low-level system call is wrapped instead.
>> +(define clone
>> + (let* ((ptr (dynamic-func "syscall" (dynamic-link)))
>> + (proc (pointer->procedure int ptr (list int int '*)))
>> + ;; TODO: Handle all supported architectures
>> + (syscall-id (match (utsname:machine (uname))
>> + ("x86_64" 56)
>> + (_ 120))))
>
> Please add the value for at least i386, mips64el, and armv7 (grep -r
> NR_clone arch/ in the kernel tree.)
Done. I should mention that Mark really doesn't like this, and
rightfully so. How can we make this more portable in the future?
>> + (match (clone (logior CLONE_NEWUSER))
>
> No need for logior.
I ended up needing it because I had to add the SIGCHLD flag. :)
>> + (0 (primitive-exit 0))
>
> Maybe exit with code 42 here...
>
>> + (pid
>> + ;; Check if user namespaces are different.
>> + (not (equal? (readlink (user-namespace pid))
>> + (readlink (user-namespace "self")))))))
>
> ... and here to a waitpid and check the status:exit-val.
>
> OK with these changes.
Done and pushed. Thanks!
- Dave
