|To:||Ralf Baechle <firstname.lastname@example.org>|
|Subject:||Re: LTP testing (shmat01)|
|From:||Jun Sun <email@example.com>|
|Date:||Mon, 08 Jul 2002 10:31:23 -0700|
|Cc:||Carsten Langgaard <firstname.lastname@example.org>, email@example.com|
|References:||<3D246924.542682B2@mips.com> <20020704193414.A29570@dea.linux-mips.net> <3D249181.D9147AAE@mips.com> <20020704215614.B29422@dea.linux-mips.net>|
|User-agent:||Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.2.1) Gecko/20010901|
Ralf Baechle wrote:
On Thu, Jul 04, 2002 at 08:18:41PM +0200, Carsten Langgaard wrote:any power of 2 > PAGE_SIZE.Ok, I see, but is there any reason for us to be different than the rest of the world ?Imho the your question already wrong :-) Any assumption about the constant's value in a piece of code is wrong. The reason why the constant's value was choosen are virtually indexed caches. The value allows attaching of shared memory segment without any cache flushes.
I think this is also an effective way to avoid cache aliasing. As long as your cache size is less than 256K, you don't get cache aliasing through shared memory. Perhaps other arches don't have cache aliasing? I know for sure i386 does not have that effect.
|<Prev in Thread]||Current Thread||[Next in Thread>|