Re: Capabilities: ALPHA time

Jeremy Fitzhardinge (jeremy@goop.org)
Tue, 04 May 1999 10:16:57 -0700 (PDT)


On 04-May-99 Pavel Machek wrote:
> Well, I'm still not sure your patch is better. I've seen it and it is
> much more complicated. _If_ we want to support multiple different CAPS
> notes with different versions simultaneously, _then_ your patch is the
> way to go.I'm just not sure having multiple CAPS notes is good idea:
> it makes kernel code more complicated, it is hard to imagine interface
> to manipulate such multiple capabilities, and I do not think it is
> neccessary at all. [You want single capabilities per executable, not
> arbitrary number of capabilities. The later is too hard to use for
> use and unneccessary, IMO.]

You miss my point.

Notes are a generic mechanism, and there could be notes in an executable for
reasons other than capabilties. You can't rely on the CAPS note being first
because something might put another note before it. Those other notes may be
something associated with CAPS, like a digital signature. Your technique would
immediately break as soon as someone tries to do something else with ELF notes.

The reason I made "addnote" very generic (even to the generic name) is that
there's a real possibility that notes in an executable will be more generally
useful, and I wanted to make sure the tools were available for doing it
properly. For the same reason, the kernel code which parses the ELF notes
should conform to the spec, rather than some half-assed hack. The code to do
it properly is a little longer, but its not conceptually more complex and has
negligable performance impact.

I would consider changing so that it would only load and parse, say, 4k of the
notes segment at most.

J

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/