From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andreas Larsson Date: Wed, 03 Sep 2014 17:14:37 +0200 Subject: [Buildroot] Sparc Leon support in Buildroot In-Reply-To: <20140903164536.1a28b0d1@free-electrons.com> References: <20140701202206.153b2cf6@free-electrons.com> <53F4800D.2060307@gaisler.com> <20140820140514.67f9fb23@free-electrons.com> <54072204.3020402@gaisler.com> <20140903164536.1a28b0d1@free-electrons.com> Message-ID: <5407305D.3070508@gaisler.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net On 2014-09-03 16:45, Thomas Petazzoni wrote: > Dear Andreas Larsson, > > On Wed, 03 Sep 2014 16:13:24 +0200, Andreas Larsson wrote: > >>> Great, thanks! Also, currently, we don't have any Sparc toolchain in >>> our autobuilder testing. Normally, we try to have all architectures we >>> support tested in our autobuilders. If we were to add the Sparc >>> architecture to our testing infrastructure, would you be willing to >>> look at the potential build failures that may result from this testing? >>> (Look at http://autobuild.buildroot.org to see what the results look >>> like for other architectures). >> >> If that would be set up, it is something I could keep an eye on. >> However, I do not have the possibility to take responsibility for broken >> packages. > > Well, it's not really a "responsibility": most of the Buildroot > contributors are volunteers so there are no obligations. But if you can > give some help from time to time for some SPARC specific issues, that > would be nice. > > Have you had some time to look at the toolchain components? As I said > earlier, we are currently keeping gcc 4.4.x only for SPARC Leon, so if > you could test if gcc 4.8.x and/or 4.9.x work fine for Leon, it would > allow us to get rid of the support for gcc 4.4. gcc 4.8.x and 4.9.x produces a bug that seems to be due to memset optimizations and a surprising memset behavior in sparc Linux. I have submitted a kernel patch to fix this (https://lkml.org/lkml/2014/8/29/333). This (or some other fix to the problem) should go into mainline and stable branches before claiming gcc 4.8/4.9 working for leon in my opinion. I'll have to test more if there are other problems with the newer gcc:s as well. Cheers, Andreas