Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751633AbdIJSkP (ORCPT ); Sun, 10 Sep 2017 14:40:15 -0400 Received: from merlin.infradead.org ([205.233.59.134]:37184 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751527AbdIJSkN (ORCPT ); Sun, 10 Sep 2017 14:40:13 -0400 Date: Sun, 10 Sep 2017 20:40:02 +0200 From: Peter Zijlstra To: Guenter Roeck Cc: Byungchul Park , linux-kernel@vger.kernel.org, Chris Zankel , Max Filippov , linux-xtensa@linux-xtensa.org Subject: Re: xtensa compiler error triggered by 'locking/lockdep: Implement the 'crossrelease' feature' Message-ID: <20170910184002.sdzru2hx5pzrseru@hirez.programming.kicks-ass.net> References: <20170910183150.GA2172@roeck-us.net> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <20170910183150.GA2172@roeck-us.net> User-Agent: NeoMutt/20170609 (1.8.3) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1088 Lines: 27 On Sun, Sep 10, 2017 at 11:31:50AM -0700, Guenter Roeck wrote: > Hi, > > xtensa:allmodconfig fails to build in mainline with compiler errors > as follows. > > drivers/staging/rtl8723bs/core/rtw_ap.c: In function ‘expire_timeout_chk’: > drivers/staging/rtl8723bs/core/rtw_ap.c:442:1: internal compiler error: > in change_address_1, at emit-rtl.c:2126 > > drivers/staging/rtl8188eu/core/rtw_ap.c: In function ‘expire_timeout_chk’: > drivers/staging/rtl8188eu/core/rtw_ap.c:445:1: internal compiler error: > in change_address_1, at emit-rtl.c:2150 > > I tried gcc 6.3, 6.4, and 7.2; they all have the same problem. > > Bisect points to commit b09be676e0f ("locking/lockdep: Implement the > 'crossrelease' feature"). > > The compile error is still seen if I disable CONFIG_LOCKDEP_CROSSRELEASE. > The image builds fine if I disable both CONFIG_RTL8723BS and CONFIG_R8188EU. I'm taking it this is specific to the GCC xtensa backend, I've not seen or heard of any other ICEs wrt this code. I'd take this up with your compiler team, this is not something we can do much about.