Re: unremovable files and possible fs corruption (2.1.123)

Matthias Urlichs (smurf@noris.de)
30 Oct 1998 14:45:40 +0100

David Holland <dholland@cs.toronto.edu> writes:
>> How would that work at the system call level? Would each change to a
>> structure that gets passed between user space and kernel space require
>> a new system call?
>
> That's how it's done, yes. Look through syscall.h on your favorite
> system (linux or bsd) and you'll find plenty of places where a call
> was renumbered to make a change like this transparent.
>
Either that, or the kernel interface is dropped and the library handles the
conversion between structure formats.

This also works the other way round -- for instance, glibc has 32-bit
device numbers and UIDs (which the kernel doesn't support yet) so for now
it translates the values down until the new kernel interface is there
(which will happen post-2.2).

> This is also how jump-table type shared libraries have been done for
> years and years.
>
ELF versioned symbols are much better, because you don't need to have one
central authority for defining the One-And-Only Correct Jump Table.

-- 
Matthias Urlichs  |  noris network GmbH   |   smurf@noris.de  |  ICQ: 20193661
The quote was selected randomly. Really.    |      http://www.noris.de/~smurf/

-- 
For peace of mind, resign as general manager of the universe.
			-Larry Eisenberg


-
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/