Re: Possible shared mapping bug in 2.4.23 (at least MIPS/Sparc)

[Peter Horton <pdh@colonel-panic.org>]

On Sat, Dec 13, 2003 at 05:41:16PM -0800, Linus Torvalds wrote:
> 
> On Sat, 13 Dec 2003, Jamie Lokier wrote:
> >
> > Peter Horton wrote:
> > > A quick look at sparc and sparc64 seem to show the same problem.
> >
> > D-cache incoherence with unsuitably aligned multiple MAP_FIXED
> > mappings is also observed on SH4, SH5, PA-RISC 1.1d.  The kernel may
> > have the same behaviour on those platforms: allowing a mapping that
> > should not be allowed.
> 
> Why?
> 
> If the user asks for it, it's the users own damn fault. Nobody guarantees
> cache coherency to users who require fixed addresses.
> 
> Just document it as a bug in the user program if this causes problems.
> Don't blame the kernel - the kernel is only doing what the user asked it
> to do.
> 

I've seen code written for X86 use MAP_FIXED to create self wrapping
ring buffers. Surely it's better to fail the mmap() on other archs
rather than for the code to fail in unexpected ways?

It's a bug either way ... either the test should be fixed up or it
should be removed from arch_get_unmapped_area() to save confusion.

P.