Re: [PATCH] proc_file_read() (Was: Re: proc_file_read() question)

[Martin Wilck]

> PAGE_OFFSET definitely works for me, but a quick scan of the headers
> suggests that non-sun3 m68k builds define PAGE_OFFSET as 0, as does
> s390.

Hum - is there no simple way to determine whether a pointer is
a valid pointer to something returned by __get_free_pages ()? You are
right, S390 in particular seems to allow arbitrary addresses starting from
0.

> Sure, the overloading is self-admittedly hacky, but (again I assume)
> the motivation was to avoid breaking the clients, many of which are
> not in the kernel.org tree. Your proposed change overloads a third
> interpretation on start, namely an arbitrary pointer, outside the
> page allocation.

For some reason I was convinced that this was the originally intended
use of start. The only quotes I find right now are Ori Pomerantz'
Module Programming Guide (http://www.linuxdoc.org/LDP/lkmpg/node16.html)
and Rubini's "Writing Device Drivers", chapter 4. Also, the comment in the
code
		if (!start) {
			/*
			 * For proc files that are less than 4k
			 */
		...
		}
supports this notion somehow (start only set if data size > page size).

After all, unless you want to mangle the file position as intended by
the hack, there is no point in touching start at all in proc_read (),
ppos will be updated automatically.

Perhaps I have misunderstood something here.
Who wrote the original code, after all?

Martin

-- 
Martin Wilck     <[e-mail]>
FSC EP PS DS1, Paderborn      Tel. [phone]