Dmitry Adamushko wrote:
On 11/10/05, Philippe Gerum <[EMAIL PROTECTED]> wrote:


1) don't display such names in /proc;
2) make a common mechanism for both spaces.

rt_mutex_create()   // for other objects as well
- xnobject_copy_name(mutex->name,name);
+xnobject_create_name(mutex->name,name, object);

xnobject_create_name(dst,src, object)
     if (src)
           snprintf(dst, XNOBJECT_NAME_LEN, "%p", object);

// the slightly ugly thing is that we need to be sure that dst is
always object->name so it's really of XNOBJECT_NAME_LEN size.

We should pass the size along with the copy buffer, mainly for readability, and
also as a safety belt.

We don't need that at all, since now we don't want to change a
behaviour of NULL-named objects created from kernel-space.

Frankly, I like this approach better. All objects get some name if
they are NULL-named by a user and the code changes are a bit more
graceful. To show them in /proc or not is still another question.

I'd like we don't generalize an exception case here. The fact that for some
internal reason we need to put a dummy name on anon objects created from
user-space should not make all anon objects from kernel space have a registry
slot too, this would just be overkill (registry slots don't come for free 

That said, the remaining point basically is: how do we avoid anon objects with
auto-generated names from appearing under /proc? I'm not a fan of "magic values"
which would distinguish a real name from a dummy one, so I'd rather use the
existing possibility of passing a NULL pnode pointer to rt_registry_enter(),
which would still index an object without actually exporting it to the /proc
interface. The question is: how could we make the *_create() calls invoking
rt_registry_enter() with a NULL pnode when applicable, without changing the
kernel-based interface of the API.

Well, I guess the only way a given *_create() call may determine a
certain environment (kernel mode vs. user mode) is by analyzing some
of the given arguments (ok, in case of the user mode, a previous call
on the stack has been one of the syscall.c:__rt_object_create() ones,
so we may track the stack... ok-ok, I'm just kidding :)
There are only 2 arguments: rt_object *object or char *name.
We probably can't rely on the first one since it may contain anything
and we don't want to force a user to set it up to something certain:

/* kernel mode */
bzero(&m); // because we use some RT_MUTEX's internal fields when
making a call from syscall.c:__rt_mutex_create(), e.g. m->handle =

rt_mutex_create(&m, NULL);

So the only way is to deal somehow with the "name" argument. There is
a difference already.

rt_mutex_create() being called from:

1) kernel mode:  name == NULL;
2) user mode:    name == "\0" (see, syscall.c:__rt_mutex_create(), so
*name == NULL;

Then a fix would look like:


- if (name && *name)
+ if (name)
+ if (*name)
   err = rt_registry_enter(mutex->name,mutex,&mutex->handle,&__mutex_pnode);
+ else {
+ // this is a NULL-named object being created from the user-mode or
+ // "\0"-named object from the kernel-mode
+xnobject_create_name(mutex->name, OBJNAME_MAX_SIZE, mutex);
+ err = rt_registry_enter(mutex->name,mutex,&mutex->handle,NULL);

        if (err)

Maybe a bit ugly but should work. eerrr... I can't come up with a
better solution at the moment.

At least the magic value here would make some sense, NULL=unregistered, empty=registered,not-exported, otherwise=registered,exported. Looks ok for that purpose, since we do need the additional 2nd state here.



Reply via email to