From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Mundt Date: Tue, 15 Sep 2009 13:07:01 +0000 Subject: Re: User mode address error problems on 7763 Message-Id: <20090915130701.GA23923@linux-sh.org> List-Id: References: <6e6aa1a80909111106l6dbcfccag4e9e7a3abb922688@mail.gmail.com> In-Reply-To: <6e6aa1a80909111106l6dbcfccag4e9e7a3abb922688@mail.gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-sh@vger.kernel.org On Tue, Sep 15, 2009 at 07:55:14AM -0500, Dale Larson wrote: > > I assumed SH7763 was an SH-X2 based core, but it turns out it is a > > first-generation SH-4A, in which case it does not implement the extended > > mode TLB. So the next available page size above 4k is 64k. I wasn't aware > > there was anyone still running first generation SH-4A, I'll have to dust > > some of mine off to see if I can reproduce your issues. > > We're finally replacing our 7760 (yep, you heard right) based product > with the 7763. > While you're dusting your's off we'll just be putting polish on it. ;-) > That's actually good to know, it means we'll have to do more aggressive regression testing on first-generation SH-X parts (we tend to scale back regression testing efforts on older parts when there are no visible users of them on the current kernel). In any event, I'll give both the SH7780 and SH7770 highlanders a go tomorrow and see how things go.. and if that doesn't work, I'm sure there's a 7763 lying around the office somwehere. ;-) > > page size. That will bring up a menu where you can cycle through the > > available page sizes. I don't recommend 1MB for a system page size, as it > > tends to interact badly with some of the slab allocators. 64k will give > > you an idea of whether there are indeed any aliasing bugs lurking about. > > I'll try that. > > Thanks! One other thing I forgot to mention is that you might also want to try a 4k page size with write-through caching enabled. This you can configure via System type -> Cache configuration.