From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rich Felker Date: Fri, 14 Aug 2015 01:10:27 +0000 Subject: Re: SH2 kernel fails to build with recent GCC Message-Id: <20150814011027.GG31018@brightrain.aerifal.cx> List-Id: References: <20150812015616.GA19843@brightrain.aerifal.cx> In-Reply-To: <20150812015616.GA19843@brightrain.aerifal.cx> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-sh@vger.kernel.org On Wed, Aug 12, 2015 at 10:01:49AM -0400, Rich Felker wrote: > On Wed, Aug 12, 2015 at 09:34:32AM +0200, Geert Uytterhoeven wrote: > > Hi Rich, > > > > On Wed, Aug 12, 2015 at 3:56 AM, Rich Felker wrote: > > > The commit that resolved GCC issue #54089 broke building the kernel > > > for SH2, by causing GCC to generate calls to new libgcc functions > > > which the kernel's fake libgcc does not provide. See the GCC issue > > > tracker for details: > > > > > > https://gcc.gnu.org/bugzilla/show_bug.cgi?idT089 > > > > > > The relevant commit is linked from comment 16; comments 33-43 (and > > > perhaps beyond) discuss the issue in detail and possible fixes. > > > > > > Before I work on a fix, would dropping the kernel's copy of libgcc > > > functions for SH and using the real libgcc (like some archs do; see > > > comment 43) be acceptable, or do we need to get updated > > > libgcc[-compatible] code into the kernel? > > > > Which libgcc functions are missing? I couldn't find that list in the bug > > report... > > At least __ashlsi3_r0 and __lshrsi3_r0. It's been a while since I > tried building with an affected GCC (see the dates on the comments I > linked) so I don't remember if there were any others. I can try to > build again and let you know. I just retried the build and confirmed that these are the only two unresolved symbols reported. So just adding them sounds practical. Rich