> On Thu, Nov 14, 2002 at 10:36:03AM -0800, Linus Torvalds wrote:
>
> >Actually, I think we should do the reverse (for testing), and make the
> >name be something small like 8 bytes, and make sure that everybody who
> >writes the name uses strncpy() and snprintf() instead of just blindly
> >writing whatever is in the database.
> >
> >Otherwise we'll always end up having fragile magic constants.
> >
> >Anybody willing to do that cleanup?
>
>
> Sure, I can do that. That leads me to think that maybe we should
> delete name from struct device and just use the one in struct kobject
> (which is already a mere 16 bytes). But if we're going to go as far
> down as the kobject... that has a dentry. And dentrys have names.
> So how about eliminating that too and just creating a dentry with the
> almost infinitely long name?
Remember that the names we are talking about here is the English
descriptive name of the PCI device... there is only _one_ need for
these names: /proc/pci
If there is going to be this much thought put into it (what a waste of
brain cycles <g>), then let's just remove CONFIG_PCI_NAMES and the name
field completely. lspci prints them out from its own database anyway,
and the only _real_ use is to provide descriptive names for devices in
/proc/pci. Why bother? For existing code, just use pdev->slot_name
instead, which is what happens anyway for hotplugged devices that appear
after we drop the PCI name database during the __init phase.
Comments?
Jeff
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/