From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964938AbcFMGud (ORCPT ); Mon, 13 Jun 2016 02:50:33 -0400 Received: from mail.kernel.org ([198.145.29.136]:37966 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S964807AbcFMGuc (ORCPT ); Mon, 13 Jun 2016 02:50:32 -0400 Date: Mon, 13 Jun 2016 15:50:17 +0900 From: Masami Hiramatsu To: David Long Cc: Catalin Marinas , Huang Shijie , James Morse , Marc Zyngier , Pratyush Anand , Sandeepa Prabhu , Will Deacon , William Cohen , linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, Steve Capper , Li Bin , Adam Buchbinder , Alex =?UTF-8?B?QmVubsOpZQ==?= , Andrew Morton , Andrey Ryabinin , Ard Biesheuvel , Christoffer Dall , Daniel Thompson , Dave P Martin , Jens Wiklander , Jisheng Zhang , John Blackwood , Mark Rutland , Petr Mladek , Robin Murphy , Suzuki K Poulose , Vladimir Murzin , Yang Shi , Zi Shen Lim , yalin wang , Mark Brown Subject: Re: [PATCH v13 05/10] arm64: Kprobes with single stepping support Message-Id: <20160613155017.860097875e8bc86563a065ce@kernel.org> In-Reply-To: <575E3235.1020904@linaro.org> References: <1464924384-15269-1-git-send-email-dave.long@linaro.org> <1464924384-15269-6-git-send-email-dave.long@linaro.org> <20160608100714.4abbcebcf356eb72173d0ce3@kernel.org> <575E3235.1020904@linaro.org> X-Mailer: Sylpheed 3.4.3 (GTK+ 2.24.28; x86_64-redhat-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 13 Jun 2016 00:10:29 -0400 David Long wrote: > >> --- > >> arch/arm64/Kconfig | 1 + > >> arch/arm64/include/asm/debug-monitors.h | 5 + > >> arch/arm64/include/asm/insn.h | 4 +- > >> arch/arm64/include/asm/kprobes.h | 60 ++++ > >> arch/arm64/include/asm/probes.h | 44 +++ > >> arch/arm64/kernel/Makefile | 1 + > >> arch/arm64/kernel/debug-monitors.c | 18 +- > >> arch/arm64/kernel/kprobes-arm64.c | 144 +++++++++ > >> arch/arm64/kernel/kprobes-arm64.h | 35 +++ > >> arch/arm64/kernel/kprobes.c | 526 ++++++++++++++++++++++++++++++++ > > > > Not sure why kprobes.c and kprobes-arm64.c are splitted. > > > > > > This comes from the model of the arm32 kprobes code where handling of > the low-level instruction simulation is implemented in separate files > for 32-bit vs. thumb instructions. It should make a little more sense > in the future when additional instruction simulation code will hopefully > be added for those instructions we cannot currently single-step > out-of-line. It also probably *could* be merged into one file. Hmm, at least the name of arch/arm64/kernel/kprobes-arm64.c is meaningless. As we've done in x86, I think we can make it arch/arm64/kernel/kprobes/decode-insn.{c,h} [..] > >> + > >> +/* Return: > >> + * INSN_REJECTED If instruction is one not allowed to kprobe, > >> + * INSN_GOOD If instruction is supported and uses instruction slot, > >> + * INSN_GOOD_NO_SLOT If instruction is supported but doesn't use its slot. > > > > Is there any chance to return INSN_GOOD_NO_SLOT? > > > > Ah, that gets used later when simulation support is added. I've removed > this enum value from this commit and will add it to the later one. > Please no one complain about using an enum instead of a bool, it will > eventually have three possible values. OK :) [..] > >> +enum kprobe_insn __kprobes > >> +arm_kprobe_decode_insn(kprobe_opcode_t *addr, struct arch_specific_insn *asi) > >> +{ > >> + enum kprobe_insn decoded; > >> + kprobe_opcode_t insn = le32_to_cpu(*addr); > >> + kprobe_opcode_t *scan_start = addr - 1; > >> + kprobe_opcode_t *scan_end = addr - MAX_ATOMIC_CONTEXT_SIZE; > >> +#if defined(CONFIG_MODULES) && defined(MODULES_VADDR) > >> + struct module *mod; > >> +#endif > >> + > >> + if (addr >= (kprobe_opcode_t *)_text && > >> + scan_end < (kprobe_opcode_t *)_text) > >> + scan_end = (kprobe_opcode_t *)_text; > >> +#if defined(CONFIG_MODULES) && defined(MODULES_VADDR) > >> + else { > >> + preempt_disable(); > >> + mod = __module_address((unsigned long)addr); > >> + if (mod && within_module_init((unsigned long)addr, mod) && > >> + !within_module_init((unsigned long)scan_end, mod)) > >> + scan_end = (kprobe_opcode_t *)mod->init_layout.base; > >> + else if (mod && within_module_core((unsigned long)addr, mod) && > >> + !within_module_core((unsigned long)scan_end, mod)) > >> + scan_end = (kprobe_opcode_t *)mod->core_layout.base; > > > > What happen if mod == NULL? it should be return error, isn't it? > > > > No, it should be fine. It just means it didn't have to do either of the > extra checks to limit the end of the search through the code to the > boundary of one of the corresponding module text sections. It means the > instruction is in the regular kernel (non-module) text segment. Ah, I see. It is OK then. :) Thank you, -- Masami Hiramatsu