Re: cache alias in mmap + write
From: Jamie Lokier
Date: Wed Jan 20 2010 - 20:32:21 EST
KOSAKI Motohiro wrote:
> 2. Add some commnet. almost developer only have x86 machine. so, arm
> specific trick need additional explicit explanation. otherwise anybody
> might break this code in the future.
What's being discussed here is not ARM-specific, although it appears
maintainers of different architecture (ARM and MIPS for a start) may
have different ideas about what they are guaranteeing to userspace.
It sounds like MIPS expects userspace to use msync() sometimes (even
though Linux msync(MS_INVALIDATE) is quite broken), and ARM expects to
to keep mappings coherent automatically (which is sometimes slower
than necessary, but usually very helpful).
> 3. Resend the patch. original mail isn't good patch format. please
> consider to reduce akpm suffer.
This type of change in generic code would need review from a number of
architecture maintainers, I'd expect.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/