Skip to contentSkip to conversation

FAILED: patch "[PATCH] riscv: stacktrace: fix the riscv stacktrace when" failed to apply to 5.10-stable tree

PATCH [email protected] 1 participant 1 message May 30, 2021

Subsystems (2)
Stable branch RISC-V
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From eac2f3059e02382d91f8c887462083841d6ea2a3 Mon Sep 17 00:00:00 2001
From: Chen Huang <[email protected]>
Date: Thu, 29 Apr 2021 07:03:48 +0000
Subject: [PATCH] riscv: stacktrace: fix the riscv stacktrace when
CONFIG_FRAME_POINTER enabled

As [1] and [2] said, the arch_stack_walk should not to trace itself, or it will leave the trace unexpectedly when called. The example is when we do "cat /sys/kernel/debug/page_owner", all pages' stack is the same.

arch_stack_walk+0x18/0x20
stack_trace_save+0x40/0x60
register_dummy_stack+0x24/0x5e
init_page_owner+0x2e

So we use __builtin_frame_address(1) as the first frame to be walked. And mark the arch_stack_walk() noinline.

We found that pr_cont will affact pages' stack whose task state is RUNNING when testing "echo t > /proc/sysrq-trigger". So move the place of pr_cont and mark the function dump_backtrace() noinline.

Also we move the case when task == NULL into else branch, and test for it in "echo c > /proc/sysrq-trigger".

[1] https://lore.kernel.org/lkml/[email protected]/
[2] https://lore.kernel.org/lkml/[email protected]/

Signed-off-by Chen Huang <[email protected]>
Fixes 5d8544e2d007 ("RISC-V: Generic library routines and assembly")
Cc [email protected]
Signed-off-by Palmer Dabbelt <[email protected]>
arch/riscv/kernel/stacktrace.c +7 -7
index 2b3e0cb90d78..bde85fc53357 100644
@@ -27,10 +27,10 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 		fp = frame_pointer(regs);
 		sp = user_stack_pointer(regs);
 		pc = instruction_pointer(regs);
-	} else if (task == NULL || task == current) {
-		fp = (unsigned long)__builtin_frame_address(0);
-		sp = sp_in_global;
-		pc = (unsigned long)walk_stackframe;
+	} else if (task == current) {
+		fp = (unsigned long)__builtin_frame_address(1);
+		sp = (unsigned long)__builtin_frame_address(0);
+		pc = (unsigned long)__builtin_return_address(0);
 	} else {
 		/* task blocked in __switch_to */
 		fp = task->thread.s[0];

@@ -106,15 +106,15 @@ static bool print_trace_address(void *arg, unsigned long pc)
 	return true;
 }
 
-void dump_backtrace(struct pt_regs *regs, struct task_struct *task,
+noinline void dump_backtrace(struct pt_regs *regs, struct task_struct *task,
 		    const char *loglvl)
 {
-	pr_cont("%sCall Trace:\n", loglvl);
 	walk_stackframe(task, regs, print_trace_address, (void *)loglvl);
 }
 
 void show_stack(struct task_struct *task, unsigned long *sp, const char *loglvl)
 {
+	pr_cont("%sCall Trace:\n", loglvl);
 	dump_backtrace(NULL, task, loglvl);
 }
 

@@ -139,7 +139,7 @@ unsigned long get_wchan(struct task_struct *task)
 
 #ifdef CONFIG_STACKTRACE
 
-void arch_stack_walk(stack_trace_consume_fn consume_entry, void *cookie,
+noinline void arch_stack_walk(stack_trace_consume_fn consume_entry, void *cookie,
 		     struct task_struct *task, struct pt_regs *regs)
 {
 	walk_stackframe(task, regs, consume_entry, cookie);