Re: [PATCH] Add SELinux module to 2.5.74-bk1

[James Morris]

On Tue, 8 Jul 2003, Christoph Hellwig wrote:

> We use this callbacks in a bunch opf places, maybe add hash_value_t
> and keycmp_t typedefs for them to avoid typing the prototypes all
> the time?

I think this is a good idea in general, but am not sure if others agree.

> > +	p = kmalloc(sizeof(*p), GFP_KERNEL);
> 
> Pass the GFP_ mask down to hashtab_create() maybe?

Would someone need to allocate a hashtab in interrupt context?

> > +		if (prev) {
> > +			newnode->next = prev->next;
> > +			prev->next = newnode;
> 
> Use hlists?

They're overkill for this, I think.

> > +config HASHTAB
> > +	tristate "Generic hash table support"
> 
> Should this really be a user option or rather implicitly selected
> by it's users?

As with the crc32 module, we don't know if any out of tree modules will 
need to use it.

(The rest of the suggestions I agree with).

> > + -fno-strict-aliasing -fno-common -g
> 
> accident?

Yes.

Thanks for the feedback.

- James
-- 
James Morris
<[e-mail]>