From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:53683) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROEdX-0004bt-72 for qemu-devel@nongnu.org; Wed, 09 Nov 2011 15:20:44 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ROEdW-0007xe-BH for qemu-devel@nongnu.org; Wed, 09 Nov 2011 15:20:43 -0500 Received: from v220110690675601.yourvserver.net ([78.47.199.172]:56209) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROEdW-0007xW-47 for qemu-devel@nongnu.org; Wed, 09 Nov 2011 15:20:42 -0500 Message-ID: <4EBAE06F.7000803@weilnetz.de> Date: Wed, 09 Nov 2011 21:19:59 +0100 From: Stefan Weil MIME-Version: 1.0 References: <1320861815-14317-1-git-send-email-rth@twiddle.net> In-Reply-To: <1320861815-14317-1-git-send-email-rth@twiddle.net> Content-Type: text/plain; charset=ISO-8859-15; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 0/3] Standardize on TCGReg for entry points List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Richard Henderson Cc: Blue Swirl , Anthony Liguori , qemu-devel@nongnu.org, Aurelien Jarno , Alexander Graf Am 09.11.2011 19:03, schrieb Richard Henderson: > This is the direction in which I believe we should fix the s390 > compilation failure. I've cross-compiled --target-list=i386-softmmu > for every tcg host to make sure that they all build. > > The second patch cannot be split into per-host patches, lest it > introduce regressions on the way to fixing them. > > > r~ > > > Richard Henderson (3): > tcg: Standardize on TCGReg as the enum for hard registers > tcg: Use TCGReg for standard tcg-target entry points. > tcg-sparc: Fix set-but-not used warnings. > > tcg/arm/tcg-target.c | 13 +++++++------ > tcg/arm/tcg-target.h | 4 ++-- > tcg/hppa/tcg-target.c | 12 ++++++------ > tcg/hppa/tcg-target.h | 4 ++-- > tcg/i386/tcg-target.c | 13 +++++++------ > tcg/i386/tcg-target.h | 4 ++-- > tcg/ia64/tcg-target.c | 12 ++++++------ > tcg/ia64/tcg-target.h | 4 ++-- > tcg/mips/tcg-target.c | 13 +++++++------ > tcg/mips/tcg-target.h | 4 ++-- > tcg/ppc/tcg-target.c | 8 ++++---- > tcg/ppc/tcg-target.h | 4 ++-- > tcg/ppc64/tcg-target.c | 8 ++++---- > tcg/ppc64/tcg-target.h | 4 ++-- > tcg/sparc/tcg-target.c | 17 +++++++---------- > tcg/sparc/tcg-target.h | 4 ++-- > tcg/tcg.c | 8 ++++---- > tcg/tci/tcg-target.c | 8 ++++---- > tcg/tci/tcg-target.h | 2 +- > tci.c | 38 +++++++++++++++++++------------------- > 20 files changed, 92 insertions(+), 92 deletions(-) Hi Richard, cross compiling for all tcg targets must have been a really hard job, so thank you very much for what you have done! Did you get all needed cross packages from your host's distribution? Which one? There remain small risks caused by the modified argument types: * TCGReg is an enum, so it might be unsigned while int was signed. As Blue already wrote, signedness should not matter here, so the risk seems to be very low. * Using TCGReg instead of TCGArg (ia64) should also be a very small risk. * Using tcg_target_long instead of int32 (mips) potentially allows larger arguments. Only the lower 16 bits are used, so this should pose no problem. Therefore for the first two patches, I can add a Reviewed-by: Stefan Weil PS. I still think TCGRegister would be nicer, but having a patch with TCGReg is better than nothing. Anthony, patches 1 and 2 are needed for QEMU 1.0 because they fix compilation for ia64 and s390.