2022-07-12 02:21:28

by Li Huafei

[permalink] [raw]
Subject: [PATCH 2/5] ARM: stacktrace: Avoid duplicate saving of exception PC value

Because an exception stack frame is not created in the exception entry,
save_trace() does special handling for the exception PC, but this is
only needed when CONFIG_FRAME_POINTER_UNWIND=y. When
CONFIG_ARM_UNWIND=y, unwind annotations have been added to the exception
entry and save_trace() will repeatedly save the exception PC:

[0x7f000090] hrtimer_hander+0x8/0x10 [hrtimer]
[0x8019ec50] __hrtimer_run_queues+0x18c/0x394
[0x8019f760] hrtimer_run_queues+0xbc/0xd0
[0x8019def0] update_process_times+0x34/0x80
[0x801ad2a4] tick_periodic+0x48/0xd0
[0x801ad3dc] tick_handle_periodic+0x1c/0x7c
[0x8010f2e0] twd_handler+0x30/0x40
[0x80177620] handle_percpu_devid_irq+0xa0/0x23c
[0x801718d0] generic_handle_domain_irq+0x24/0x34
[0x80502d28] gic_handle_irq+0x74/0x88
[0x8085817c] generic_handle_arch_irq+0x58/0x78
[0x80100ba8] __irq_svc+0x88/0xc8
[0x80108114] arch_cpu_idle+0x38/0x3c
[0x80108114] arch_cpu_idle+0x38/0x3c <==== duplicate saved exception PC
[0x80861bf8] default_idle_call+0x38/0x130
[0x8015d5cc] do_idle+0x150/0x214
[0x8015d978] cpu_startup_entry+0x18/0x1c
[0x808589c0] rest_init+0xd8/0xdc
[0x80c00a44] arch_post_acpi_subsys_init+0x0/0x8

We can move the special handling of the exception PC in save_trace() to
the unwind_frame() of the frame pointer unwinder.

Signed-off-by: Li Huafei <[email protected]>
---
arch/arm/include/asm/stacktrace.h | 6 +++++
arch/arm/kernel/return_address.c | 1 +
arch/arm/kernel/stacktrace.c | 38 +++++++++++++++++++------------
3 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/arch/arm/include/asm/stacktrace.h b/arch/arm/include/asm/stacktrace.h
index 3e78f921b8b2..25282ff645fb 100644
--- a/arch/arm/include/asm/stacktrace.h
+++ b/arch/arm/include/asm/stacktrace.h
@@ -21,6 +21,9 @@ struct stackframe {
struct llist_node *kr_cur;
struct task_struct *tsk;
#endif
+#ifdef CONFIG_UNWINDER_FRAME_POINTER
+ bool ex_frame;
+#endif
};

static __always_inline
@@ -34,6 +37,9 @@ void arm_get_current_stackframe(struct pt_regs *regs, struct stackframe *frame)
frame->kr_cur = NULL;
frame->tsk = current;
#endif
+#ifdef CONFIG_UNWINDER_FRAME_POINTER
+ frame->ex_frame = in_entry_text(frame->pc) ? true : false;
+#endif
}

extern int unwind_frame(struct stackframe *frame);
diff --git a/arch/arm/kernel/return_address.c b/arch/arm/kernel/return_address.c
index 8aac1e10b117..38f1ea9c724d 100644
--- a/arch/arm/kernel/return_address.c
+++ b/arch/arm/kernel/return_address.c
@@ -47,6 +47,7 @@ void *return_address(unsigned int level)
frame.kr_cur = NULL;
frame.tsk = current;
#endif
+ frame.ex_frame = false;

walk_stackframe(&frame, save_return_addr, &data);

diff --git a/arch/arm/kernel/stacktrace.c b/arch/arm/kernel/stacktrace.c
index af87040b0353..3acf51ee46bb 100644
--- a/arch/arm/kernel/stacktrace.c
+++ b/arch/arm/kernel/stacktrace.c
@@ -82,6 +82,21 @@ int notrace unwind_frame(struct stackframe *frame)
if (frame_pointer_check(frame))
return -EINVAL;

+ /*
+ * When we unwind through an exception stack, include the saved PC
+ * value into the stack trace.
+ */
+ if (frame->ex_frame) {
+ struct pt_regs *regs = (struct pt_regs *)frame->sp;
+
+ if ((unsigned long)&regs[1] > ALIGN(frame->sp, THREAD_SIZE))
+ return -EINVAL;
+
+ frame->pc = regs->ARM_pc;
+ frame->ex_frame = false;
+ return 0;
+ }
+
/* restore the registers from the stack frame */
#ifdef CONFIG_CC_IS_CLANG
frame->sp = frame->fp;
@@ -98,6 +113,9 @@ int notrace unwind_frame(struct stackframe *frame)
(void *)frame->fp, &frame->kr_cur);
#endif

+ if (in_entry_text(frame->pc))
+ frame->ex_frame = true;
+
return 0;
}
#endif
@@ -128,7 +146,6 @@ static int save_trace(struct stackframe *frame, void *d)
{
struct stack_trace_data *data = d;
struct stack_trace *trace = data->trace;
- struct pt_regs *regs;
unsigned long addr = frame->pc;

if (data->no_sched_functions && in_sched_functions(addr))
@@ -139,19 +156,6 @@ static int save_trace(struct stackframe *frame, void *d)
}

trace->entries[trace->nr_entries++] = addr;
-
- if (trace->nr_entries >= trace->max_entries)
- return 1;
-
- if (!in_entry_text(frame->pc))
- return 0;
-
- regs = (struct pt_regs *)frame->sp;
- if ((unsigned long)&regs[1] > ALIGN(frame->sp, THREAD_SIZE))
- return 0;
-
- trace->entries[trace->nr_entries++] = regs->ARM_pc;
-
return trace->nr_entries >= trace->max_entries;
}

@@ -193,6 +197,9 @@ static noinline void __save_stack_trace(struct task_struct *tsk,
frame.kr_cur = NULL;
frame.tsk = tsk;
#endif
+#ifdef CONFIG_UNWINDER_FRAME_POINTER
+ frame.ex_frame = false;
+#endif

walk_stackframe(&frame, save_trace, &data);
}
@@ -214,6 +221,9 @@ void save_stack_trace_regs(struct pt_regs *regs, struct stack_trace *trace)
frame.kr_cur = NULL;
frame.tsk = current;
#endif
+#ifdef CONFIG_UNWINDER_FRAME_POINTER
+ frame.ex_frame = in_entry_text(frame.pc) ? true : false;
+#endif

walk_stackframe(&frame, save_trace, &data);
}
--
2.17.1


2022-07-18 09:37:17

by Linus Walleij

[permalink] [raw]
Subject: Re: [PATCH 2/5] ARM: stacktrace: Avoid duplicate saving of exception PC value

On Tue, Jul 12, 2022 at 4:18 AM Li Huafei <[email protected]> wrote:

> Because an exception stack frame is not created in the exception entry,
> save_trace() does special handling for the exception PC, but this is
> only needed when CONFIG_FRAME_POINTER_UNWIND=y. When
> CONFIG_ARM_UNWIND=y, unwind annotations have been added to the exception
> entry and save_trace() will repeatedly save the exception PC:
>
> [0x7f000090] hrtimer_hander+0x8/0x10 [hrtimer]
> [0x8019ec50] __hrtimer_run_queues+0x18c/0x394
> [0x8019f760] hrtimer_run_queues+0xbc/0xd0
> [0x8019def0] update_process_times+0x34/0x80
> [0x801ad2a4] tick_periodic+0x48/0xd0
> [0x801ad3dc] tick_handle_periodic+0x1c/0x7c
> [0x8010f2e0] twd_handler+0x30/0x40
> [0x80177620] handle_percpu_devid_irq+0xa0/0x23c
> [0x801718d0] generic_handle_domain_irq+0x24/0x34
> [0x80502d28] gic_handle_irq+0x74/0x88
> [0x8085817c] generic_handle_arch_irq+0x58/0x78
> [0x80100ba8] __irq_svc+0x88/0xc8
> [0x80108114] arch_cpu_idle+0x38/0x3c
> [0x80108114] arch_cpu_idle+0x38/0x3c <==== duplicate saved exception PC
> [0x80861bf8] default_idle_call+0x38/0x130
> [0x8015d5cc] do_idle+0x150/0x214
> [0x8015d978] cpu_startup_entry+0x18/0x1c
> [0x808589c0] rest_init+0xd8/0xdc
> [0x80c00a44] arch_post_acpi_subsys_init+0x0/0x8
>
> We can move the special handling of the exception PC in save_trace() to
> the unwind_frame() of the frame pointer unwinder.
>
> Signed-off-by: Li Huafei <[email protected]>

This is another very nice patch!
Reviewed-by: Linus Waleij <[email protected]>

Nitpick:

> + if ((unsigned long)&regs[1] > ALIGN(frame->sp, THREAD_SIZE))
> + return -EINVAL;

It'd be nice to add a comment saying what is in regs[1] at this point
so it is easier to understand the code. Not your fault as it is just
moved code, but if you have time please add a small comment.

Yours,
Linus Walleij

2022-07-26 09:41:23

by Li Huafei

[permalink] [raw]
Subject: Re: [PATCH 2/5] ARM: stacktrace: Avoid duplicate saving of exception PC value



On 2022/7/18 17:01, Linus Walleij wrote:
> On Tue, Jul 12, 2022 at 4:18 AM Li Huafei <[email protected]> wrote:
>
>> Because an exception stack frame is not created in the exception entry,
>> save_trace() does special handling for the exception PC, but this is
>> only needed when CONFIG_FRAME_POINTER_UNWIND=y. When
>> CONFIG_ARM_UNWIND=y, unwind annotations have been added to the exception
>> entry and save_trace() will repeatedly save the exception PC:
>>
>> [0x7f000090] hrtimer_hander+0x8/0x10 [hrtimer]
>> [0x8019ec50] __hrtimer_run_queues+0x18c/0x394
>> [0x8019f760] hrtimer_run_queues+0xbc/0xd0
>> [0x8019def0] update_process_times+0x34/0x80
>> [0x801ad2a4] tick_periodic+0x48/0xd0
>> [0x801ad3dc] tick_handle_periodic+0x1c/0x7c
>> [0x8010f2e0] twd_handler+0x30/0x40
>> [0x80177620] handle_percpu_devid_irq+0xa0/0x23c
>> [0x801718d0] generic_handle_domain_irq+0x24/0x34
>> [0x80502d28] gic_handle_irq+0x74/0x88
>> [0x8085817c] generic_handle_arch_irq+0x58/0x78
>> [0x80100ba8] __irq_svc+0x88/0xc8
>> [0x80108114] arch_cpu_idle+0x38/0x3c
>> [0x80108114] arch_cpu_idle+0x38/0x3c <==== duplicate saved exception PC
>> [0x80861bf8] default_idle_call+0x38/0x130
>> [0x8015d5cc] do_idle+0x150/0x214
>> [0x8015d978] cpu_startup_entry+0x18/0x1c
>> [0x808589c0] rest_init+0xd8/0xdc
>> [0x80c00a44] arch_post_acpi_subsys_init+0x0/0x8
>>
>> We can move the special handling of the exception PC in save_trace() to
>> the unwind_frame() of the frame pointer unwinder.
>>
>> Signed-off-by: Li Huafei <[email protected]>
>
> This is another very nice patch!
> Reviewed-by: Linus Waleij <[email protected]>
>
> Nitpick:
>
>> + if ((unsigned long)&regs[1] > ALIGN(frame->sp, THREAD_SIZE))
>> + return -EINVAL;
>
> It'd be nice to add a comment saying what is in regs[1] at this point
> so it is easier to understand the code. Not your fault as it is just
> moved code, but if you have time please add a small comment.
>

It is necessary to add the comment. This check is to ensure that 'regs +
sizeof(struct pt_regs)' (that is, &regs[1]) does not go beyond the
bottom of the stack, to avoid accessing data outside the task's stack,
see commit 40ff1ddb5570 ("ARM: 8948/1: Prevent OOB access in
stacktrace") for details. I will add a comment in the next version.

Thanks,
Huafei

> Yours,
> Linus Walleij
> .
>

2022-07-26 10:54:56

by Russell King (Oracle)

[permalink] [raw]
Subject: Re: [PATCH 2/5] ARM: stacktrace: Avoid duplicate saving of exception PC value

On Tue, Jul 12, 2022 at 10:15:24AM +0800, Li Huafei wrote:
> @@ -34,6 +37,9 @@ void arm_get_current_stackframe(struct pt_regs *regs, struct stackframe *frame)
> frame->kr_cur = NULL;
> frame->tsk = current;
> #endif
> +#ifdef CONFIG_UNWINDER_FRAME_POINTER
> + frame->ex_frame = in_entry_text(frame->pc) ? true : false;

in_entry_text() returns a bool, so there's no need for the ternary
operator. The same comment applies throughout this patch.

--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!

2022-07-26 13:05:36

by Li Huafei

[permalink] [raw]
Subject: Re: [PATCH 2/5] ARM: stacktrace: Avoid duplicate saving of exception PC value



On 2022/7/26 18:22, Russell King (Oracle) wrote:
> On Tue, Jul 12, 2022 at 10:15:24AM +0800, Li Huafei wrote:
>> @@ -34,6 +37,9 @@ void arm_get_current_stackframe(struct pt_regs *regs, struct stackframe *frame)
>> frame->kr_cur = NULL;
>> frame->tsk = current;
>> #endif
>> +#ifdef CONFIG_UNWINDER_FRAME_POINTER
>> + frame->ex_frame = in_entry_text(frame->pc) ? true : false;
>
> in_entry_text() returns a bool, so there's no need for the ternary
> operator. The same comment applies throughout this patch.
>
OK, I will fix it in v3.

Thanks,
Huafei