From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754229AbbJPO2g (ORCPT ); Fri, 16 Oct 2015 10:28:36 -0400 Received: from mail-pa0-f50.google.com ([209.85.220.50]:35384 "EHLO mail-pa0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752490AbbJPO2e (ORCPT ); Fri, 16 Oct 2015 10:28:34 -0400 From: Jungseok Lee To: catalin.marinas@arm.com, will.deacon@arm.com, linux-arm-kernel@lists.infradead.org Cc: linux-kernel@vger.kernel.org, AKASHI Takahiro , James Morse , Mark Rutland Subject: [PATCH v2] arm64: Synchronise dump_backtrace() with perf callchain Date: Fri, 16 Oct 2015 14:28:16 +0000 Message-Id: <1445005696-7887-1-git-send-email-jungseoklee85@gmail.com> X-Mailer: git-send-email 2.5.0 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Unlike perf callchain relying on walk_stackframe(), dump_backtrace() has its own backtrace logic. A major difference between them is the moment a symbol is recorded. Perf writes down a symobl *before* calling unwind_frame(), but dump_backtrace() prints it out *after* unwind_frame(). As a result, the last valid symbol cannot be hooked in case of dump_backtrace(). This patch addresses the issue as synchronising dump_backtrace() with perf callchain. A simple test and its results are as follows: - crash trigger $ sudo echo c > /proc/sysrq-trigger - current status Call trace: [] sysrq_handle_crash+0x24/0x30 [] __handle_sysrq+0x128/0x19c [] write_sysrq_trigger+0x60/0x74 [] proc_reg_write+0x84/0xc0 [] __vfs_write+0x44/0x104 [] vfs_write+0x98/0x1a8 [] SyS_write+0x50/0xb0 - with this change Call trace: [] sysrq_handle_crash+0x24/0x30 [] __handle_sysrq+0x128/0x19c [] write_sysrq_trigger+0x60/0x74 [] proc_reg_write+0x84/0xc0 [] __vfs_write+0x44/0x104 [] vfs_write+0x98/0x1a8 [] SyS_write+0x50/0xb0 [] el0_svc_naked+0x20/0x28 Note that this patch does not cover a case where MMU is disabled. The last stack frame of swapper, for example, has PC in a form of physical address. Unfortunately, a simple conversion using phys_to_virt() cannot cover all scenarios since PC is retrieved from LR - 4, not LR. It is a big tradeoff to change both head.S and unwind_frame() for only a few of symbols in *.S. Thus, this hunk does not take care of the case. Cc: AKASHI Takahiro Cc: James Morse Cc: Mark Rutland Signed-off-by: Jungseok Lee --- Changes since v1: - Added an example to the commit msg, per Will - Modified a comment. arch/arm64/kernel/traps.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c index f93aae5..e9b9b53 100644 --- a/arch/arm64/kernel/traps.c +++ b/arch/arm64/kernel/traps.c @@ -103,12 +103,12 @@ static void dump_mem(const char *lvl, const char *str, unsigned long bottom, set_fs(fs); } -static void dump_backtrace_entry(unsigned long where, unsigned long stack) +static void dump_backtrace_entry(unsigned long where) { + /* + * Note that 'where' can have a physical address, but it's not handled. + */ print_ip_sym(where); - if (in_exception_text(where)) - dump_mem("", "Exception stack", stack, - stack + sizeof(struct pt_regs), false); } static void dump_instr(const char *lvl, struct pt_regs *regs) @@ -172,12 +172,17 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk) pr_emerg("Call trace:\n"); while (1) { unsigned long where = frame.pc; + unsigned long stack; int ret; + dump_backtrace_entry(where); ret = unwind_frame(&frame); if (ret < 0) break; - dump_backtrace_entry(where, frame.sp); + stack = frame.sp; + if (in_exception_text(where)) + dump_mem("", "Exception stack", stack, + stack + sizeof(struct pt_regs), false); } } -- 2.5.0