From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH RFC] sparc64: eBPF JIT Date: Fri, 21 Apr 2017 21:19:37 -0400 (EDT) Message-ID: <20170421.211937.393214469332032143.davem@davemloft.net> References: <20170418054444.GA36377@ast-mbp.thefacebook.com> <20170418.143732.1494140903085244508.davem@davemloft.net> <20170418225707.GA65449@ast-mbp.thefacebook.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: sparclinux@vger.kernel.org, netdev@vger.kernel.org, ast@kernel.org, daniel@iogearbox.net To: alexei.starovoitov@gmail.com Return-path: Received: from shards.monkeyblade.net ([184.105.139.130]:41138 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1425547AbdDVBTj (ORCPT ); Fri, 21 Apr 2017 21:19:39 -0400 In-Reply-To: <20170418225707.GA65449@ast-mbp.thefacebook.com> Sender: netdev-owner@vger.kernel.org List-ID: From: Alexei Starovoitov Date: Tue, 18 Apr 2017 15:57:09 -0700 > Alternative idea: can the above > 'add FP, STACK_BIAS, one_of_local_regs' be done once in prologue > and that register used as substitue for R10 ? After much consideration I think this is what I'm going to end up doing, I should have listened to you from the beginning :-) It will simplify the JIT significantly. Now I will just have to completely trust that the verifier has audited all the FP accesses properly. :-) From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Date: Sat, 22 Apr 2017 01:19:37 +0000 Subject: Re: [PATCH RFC] sparc64: eBPF JIT Message-Id: <20170421.211937.393214469332032143.davem@davemloft.net> List-Id: References: <20170418054444.GA36377@ast-mbp.thefacebook.com> <20170418.143732.1494140903085244508.davem@davemloft.net> <20170418225707.GA65449@ast-mbp.thefacebook.com> In-Reply-To: <20170418225707.GA65449@ast-mbp.thefacebook.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: alexei.starovoitov@gmail.com Cc: sparclinux@vger.kernel.org, netdev@vger.kernel.org, ast@kernel.org, daniel@iogearbox.net From: Alexei Starovoitov Date: Tue, 18 Apr 2017 15:57:09 -0700 > Alternative idea: can the above > 'add FP, STACK_BIAS, one_of_local_regs' be done once in prologue > and that register used as substitue for R10 ? After much consideration I think this is what I'm going to end up doing, I should have listened to you from the beginning :-) It will simplify the JIT significantly. Now I will just have to completely trust that the verifier has audited all the FP accesses properly. :-)