linux-kernel - [PATCH for-next v2 4/4] riscv: stacktrace: Implement stacktrace for irq

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20220921125128.33913-5-chenzhongjin@huawei.com>
Date:   Wed, 21 Sep 2022 20:51:27 +0800
From:   Chen Zhongjin <chenzhongjin@...wei.com>
To:     <linux-kernel@...r.kernel.org>, <linux-riscv@...ts.infradead.org>,
        <linux-perf-users@...r.kernel.org>
CC:     <paul.walmsley@...ive.com>, <palmer@...belt.com>,
        <aou@...s.berkeley.edu>, <peterz@...radead.org>,
        <mingo@...hat.com>, <acme@...nel.org>, <mark.rutland@....com>,
        <alexander.shishkin@...ux.intel.com>, <namhyung@...nel.org>,
        <jolsa@...nel.org>, <guoren@...nel.org>, <frederic@...nel.org>,
        <vincent.chen@...ive.com>, <ardb@...nel.org>,
        <mhiramat@...nel.org>, <rostedt@...dmis.org>,
        <keescook@...omium.org>, <catalin.marinas@....com>,
        <chenzhongjin@...wei.com>
Subject: [PATCH for-next v2 4/4] riscv: stacktrace: Implement stacktrace for irq

After adding encoded fp onto stack to record pt_regs, now the
unwinder have ability to unwind frame through irq.

There is two steps to unwind irq frame and the interrupted frame:

1. When there is an encoded fp on stack, we can get the pt_regs
and unwind frame by (regs->epc) and (regs->s0).

2. To unwind the interrupted frame, there is two possibilities,
we can determine the situation by checking whether the value in
frame->ra position is a fp value.

If there is a fp in ra position:
We are inside a leaf frame and there is only fp on ra position.
Get fp from ra position and get next pc from pt_regs.
Else:
Just get fp and next pc from stack frame.

Stacktrace before this patch:

 Call Trace:
  ...
  [<ffffffff800aa692>] __flush_smp_call_function_queue+0xde/0x1fa
  [<ffffffff800ab404>] generic_smp_call_function_single_interrupt+0x22/0x2a
  [<ffffffff800077b2>] handle_IPI+0xaa/0x108
  [<ffffffff803f827e>] riscv_intc_irq+0x56/0x6e
  [<ffffffff808d94b6>] generic_handle_arch_irq+0x4c/0x76
  [<ffffffff80003ad0>] ret_from_exception+0x0/0xc

Stacktrace after this patch:

 Call Trace:
  ...
  [<ffffffff800aa6da>] __flush_smp_call_function_queue+0xde/0x1fa
  [<ffffffff800ab44c>] generic_smp_call_function_single_interrupt+0x22/0x2a
  [<ffffffff800077fa>] handle_IPI+0xaa/0x108
  [<ffffffff803f82c6>] riscv_intc_irq+0x56/0x6e
  [<ffffffff808d94fe>] generic_handle_arch_irq+0x4c/0x76
  [<ffffffff80003ad0>] ret_from_exception+0x0/0xc
+ [<ffffffff80003d52>] arch_cpu_idle+0x22/0x28
+ [<ffffffff808e23a8>] default_idle_call+0x44/0xee
+ [<ffffffff80056ece>] do_idle+0x116/0x126
+ [<ffffffff8005706e>] cpu_startup_entry+0x36/0x38
+ [<ffffffff808d99ae>] kernel_init+0x0/0x15a
+ [<ffffffff80a007a0>] arch_post_acpi_subsys_init+0x0/0x38
+ [<ffffffff80a0100c>] start_kernel+0x7c4/0x7f2

Signed-off-by: Chen Zhongjin <chenzhongjin@...wei.com>
---
 arch/riscv/kernel/stacktrace.c | 45 ++++++++++++++++++++++++++++------
 1 file changed, 38 insertions(+), 7 deletions(-)

diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c
index e84e21868a3e..976dc298ab3b 100644
--- a/arch/riscv/kernel/stacktrace.c
+++ b/arch/riscv/kernel/stacktrace.c
@@ -16,29 +16,60 @@
 
 #ifdef CONFIG_FRAME_POINTER
 
+static struct pt_regs *decode_frame_pointer(unsigned long fp)
+{
+	if (!(fp & 0x1))
+		return NULL;
+
+	return (struct pt_regs *)(fp & ~0x1);
+}
+
 static int notrace unwind_next(struct unwind_state *state)
 {
 	unsigned long low, high, fp;
 	struct stackframe *frame;
+	struct pt_regs *regs;
 
-	fp = state->fp;
+	regs = decode_frame_pointer(state->fp);
 
 	/* Validate frame pointer */
-	low = state->sp + sizeof(struct stackframe);
+	if (regs) {
+		if user_mode(regs)
+			return -1;
+
+		fp = (unsigned long)regs;
+		low = state->sp;
+	} else {
+		fp = state->fp;
+		low = state->sp + sizeof(struct stackframe);
+	}
 	high = ALIGN(low, THREAD_SIZE);
 
 	if (fp < low || fp > high || fp & 0x7)
 		return -EINVAL;
 
-	/* Unwind stack frame */
 	frame = (struct stackframe *)fp - 1;
 	state->sp = fp;
 
-	if (state->regs && state->regs->epc == state->pc &&
-		fp & 0x7) {
-		state->fp = frame->ra;
-		state->pc = state->regs->ra;
+	if (regs) {
+	/* Unwind from irq to interrupted function */
+		state->fp = regs->s0;
+		state->pc = regs->epc;
+		state->regs = regs;
+	} else if (state->regs && state->regs->epc == state->pc) {
+	/* Unwind from interrupted function to caller*/
+		if (frame->ra < low || frame->ra > high) {
+		/* normal function */
+			state->fp = frame->fp;
+			state->pc = frame->ra;
+		} else {
+		/* leaf function */
+			state->fp = frame->ra;
+			state->pc = state->regs->ra;
+		}
+		state->regs = NULL;
 	} else {
+	/* Unwind from normal stack frame */
 		state->fp = frame->fp;
 		state->pc = ftrace_graph_ret_addr(current, NULL, frame->ra,
 							(unsigned long *)fp - 1);
-- 
2.17.1