Re: [PATCH 1/2] WorkStruct: Add assign_bits() to give an atomic-bitops safe assignment

[David Howells <dhowells@redhat.com>]

Russell King <rmk+lkml@arm.linux.org.uk> wrote:

> This seems to be a very silly question (and I'm bound to be utterly
> wrong as proven in my last round) but why are we implementing a new
> set of atomic primitives which effectively do the same thing as our
> existing set?
> 
> Why can't we just use atomic_t for this?

atomic_t is the wrong thing as it's basically an int, not an unsigned long.

atomic64_t/atomic_long_t is also probably the wrong thing to use as it's a
signed long (and the long is also volatile on some platforms - x86_64 for
example).  Bitops operate on unsigned long.

But the most important point is that assign_bits() has to take the same pointer
type as test_bit(), set_bit(), test_and_set_bit(), etc., and none of those
operate on an atomic*_t.

We could change that, of course, but I don't fancy tackling the task just at
the moment.  It oughtn't to be a difficult change, but there are a lot of flags
words in the kernel.

David