From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id E2FA2C6FA91 for ; Wed, 21 Sep 2022 12:55:18 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229632AbiIUMzS (ORCPT ); Wed, 21 Sep 2022 08:55:18 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:38212 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229880AbiIUMzN (ORCPT ); Wed, 21 Sep 2022 08:55:13 -0400 Received: from szxga08-in.huawei.com (szxga08-in.huawei.com [45.249.212.255]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id EB24B2E9C9; Wed, 21 Sep 2022 05:55:11 -0700 (PDT) Received: from dggpemm500023.china.huawei.com (unknown [172.30.72.53]) by szxga08-in.huawei.com (SkyGuard) with ESMTP id 4MXdZH14Zxz14RgJ; Wed, 21 Sep 2022 20:51:03 +0800 (CST) Received: from dggpemm500013.china.huawei.com (7.185.36.172) by dggpemm500023.china.huawei.com (7.185.36.83) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.31; Wed, 21 Sep 2022 20:55:10 +0800 Received: from ubuntu1804.huawei.com (10.67.175.36) by dggpemm500013.china.huawei.com (7.185.36.172) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.31; Wed, 21 Sep 2022 20:55:09 +0800 From: Chen Zhongjin To: , , CC: , , , , , , , , , , , , , , , , , , Subject: [PATCH for-next v2 3/4] riscv: stacktrace: Save pt_regs in ENCODE_FRAME_POINTER Date: Wed, 21 Sep 2022 20:51:26 +0800 Message-ID: <20220921125128.33913-4-chenzhongjin@huawei.com> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20220921125128.33913-1-chenzhongjin@huawei.com> References: <20220921125128.33913-1-chenzhongjin@huawei.com> MIME-Version: 1.0 Content-Type: text/plain X-Originating-IP: [10.67.175.36] X-ClientProxiedBy: dggems704-chm.china.huawei.com (10.3.19.181) To dggpemm500013.china.huawei.com (7.185.36.172) X-CFilter-Loop: Reflected Precedence: bulk List-ID: X-Mailing-List: linux-perf-users@vger.kernel.org To support stack unwinding when there is a pt_regs on stack, the position of pt_regs is nessesary. Because for some functions, compiler only push s0/fp on stack without ra. As the situation described in commit f766f77a74f5("riscv/stacktrace: Fix stack output without ra on the stack top") When irq happens there, the function frame looks like: prev function | ... | | | normal function +-----------------+ | ra to prev | | s0 of prev | | ... |<-+ leaf function +-----------------+ | | s0 of normal | | | empty slot | | irq pt_regs +-----------------+ | | epc (ra to leaf)| | | ra (ra to norm)| | | ... | | | s0 of leaf |--+ | ... | +-----------------+ When unwinding from unwinding from leaf to normal, beacause (ra to norm) is saved in pt_regs, but not stackframe of leaf, we have to get pt_regs for that. To get pt_regs position on stack, we can save the encoded *pt_regs in s0, as x86 architecture did. Then we can get s0, epc and ra easily. Add ENCODE_FRAME_POINTER in irq, ftrace and kretprobe trampoline entries. For ftrace and kretprobes we should also set pt_regs as kernel mode so that they are not mistaken as user_mode pt_regs. Signed-off-by: Chen Zhongjin Reviewed-by: Guo Ren --- arch/riscv/include/asm/frame.h | 45 +++++++++++++++++++ arch/riscv/kernel/entry.S | 3 ++ arch/riscv/kernel/mcount-dyn.S | 7 +++ arch/riscv/kernel/probes/kprobes_trampoline.S | 7 +++ 4 files changed, 62 insertions(+) create mode 100644 arch/riscv/include/asm/frame.h diff --git a/arch/riscv/include/asm/frame.h b/arch/riscv/include/asm/frame.h new file mode 100644 index 000000000000..2a1f45cf3a4e --- /dev/null +++ b/arch/riscv/include/asm/frame.h @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _ASM_RISCV_FRAME_H +#define _ASM_RISCV_FRAME_H + +#include + +#ifdef CONFIG_FRAME_POINTER + +#ifdef __ASSEMBLY__ + +/* + * This is a sneaky trick to help the unwinder find pt_regs on the stack. The + * frame pointer is replaced with an encoded pointer to pt_regs. The encoding + * is just setting the LSB, which makes it an invalid stack address and is also + * a signal to the unwinder that it's a pt_regs pointer in disguise. + * + * This macro must be used when sp point to pt_regs + */ +.macro ENCODE_FRAME_POINTER + add s0, sp, 0x1 +.endm + +#else /* !__ASSEMBLY__ */ + +#define ENCODE_FRAME_POINTER \ + "add s0, sp, 0x1\n\t" + +#endif /* __ASSEMBLY__ */ + +#else /* !CONFIG_FRAME_POINTER */ + +#ifdef __ASSEMBLY__ + +.macro ENCODE_FRAME_POINTER ptregs_offset=0 +.endm + +#else /* !__ASSEMBLY */ + +#define ENCODE_FRAME_POINTER + +#endif /* !__ASSEMBLY */ + +#endif /* CONFIG_FRAME_POINTER */ + +#endif /* _ASM_RISCV_FRAME_H */ diff --git a/arch/riscv/kernel/entry.S b/arch/riscv/kernel/entry.S index b9eda3fcbd6d..ecb15c7430b4 100644 --- a/arch/riscv/kernel/entry.S +++ b/arch/riscv/kernel/entry.S @@ -13,6 +13,7 @@ #include #include #include +#include #if !IS_ENABLED(CONFIG_PREEMPTION) .set resume_kernel, restore_all @@ -95,6 +96,8 @@ _save_context: REG_S s4, PT_CAUSE(sp) REG_S s5, PT_TP(sp) + ENCODE_FRAME_POINTER + /* * Set the scratch register to 0, so that if a recursive exception * occurs, the exception vector knows it came from the kernel diff --git a/arch/riscv/kernel/mcount-dyn.S b/arch/riscv/kernel/mcount-dyn.S index d171eca623b6..a362521e030e 100644 --- a/arch/riscv/kernel/mcount-dyn.S +++ b/arch/riscv/kernel/mcount-dyn.S @@ -10,6 +10,8 @@ #include #include #include +#include +#include .text @@ -172,6 +174,11 @@ ENDPROC(ftrace_caller) #ifdef CONFIG_DYNAMIC_FTRACE_WITH_REGS ENTRY(ftrace_regs_caller) SAVE_ALL +#ifdef CONFIG_FRAME_POINTER + li s0, SR_PP + REG_S s0, PT_STATUS(sp) + ENCODE_FRAME_POINTER +#endif addi a0, ra, -FENTRY_RA_OFFSET la a1, function_trace_op diff --git a/arch/riscv/kernel/probes/kprobes_trampoline.S b/arch/riscv/kernel/probes/kprobes_trampoline.S index 7bdb09ded39b..70760a16784f 100644 --- a/arch/riscv/kernel/probes/kprobes_trampoline.S +++ b/arch/riscv/kernel/probes/kprobes_trampoline.S @@ -6,6 +6,8 @@ #include #include +#include +#include .text .altmacro @@ -78,6 +80,11 @@ ENTRY(__kretprobe_trampoline) addi sp, sp, -(PT_SIZE_ON_STACK) save_all_base_regs +#ifdef CONFIG_FRAME_POINTER + li s0, SR_PP + REG_S s0, PT_STATUS(sp) + ENCODE_FRAME_POINTER +#endif move a0, sp /* pt_regs */ -- 2.17.1