linux-kernel - Re: [PATCH 6/6] arm64: Use ftrace_graph_get_ret_stack() instead of curr_ret

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: <f0613902-80ac-5865-26ee-f0aac4d3b906@arm.com>
Date:   Thu, 13 Dec 2018 17:09:35 +0000
From:   James Morse <james.morse@....com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     linux-kernel@...r.kernel.org, linux-arch@...r.kernel.org,
        Ingo Molnar <mingo@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        linux-arm-kernel@...ts.infradead.org,
        Will Deacon <will.deacon@....com>,
        Mark Rutland <mark.rutland@....com>,
        Catalin Marinas <catalin.marinas@....com>
Subject: Re: [PATCH 6/6] arm64: Use ftrace_graph_get_ret_stack() instead of
 curr_ret_stack

Hi Steven,

On 10/12/2018 19:30, Steven Rostedt wrote:
> From: "Steven Rostedt (VMware)" <rostedt@...dmis.org>
> 
> [
>   Folks, I'm working on rewriting the function graph tracer. In order to
>   do so, some changes need to be done that affect architecture specific
>   code. I'm only able to compile test these changes. I would like to
>   have folks check out my repo and give them a test.
> 
>     git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace.git
>   ftrace/core
> 
>   Head SHA1: 51584396cff54aaf57ed0bd353767d71429f77b4
> ]
> 
> The structure of the ret_stack array on the task struct is going to
> change, and accessing it directly via the curr_ret_stack index will no
> longer give the ret_stack entry that holds the return address. To access
> that, architectures must now use ftrace_graph_get_ret_stack() to get the
> associated ret_stack that matches the saved return address.

I gave this branch a spin, but I hit the WARN_ON() fairly easily:

root@...ian-guest:~# echo function_graph > /sys/kernel/debug/tracing/current_tracer
root@...ian-guest:~# ps
[   27.744824] WARNING: CPU: 0 PID: 2457 at arch/arm64/kernel/stacktrace.c:70
unwind_frame+0x78/0x128
[   27.745174] Modules linked in:
[   27.745375] CPU: 0 PID: 2457 Comm: ps Not tainted
4.20.0-rc3-00038-g51584396cff5 #10814
[   27.745697] Hardware name: linux,dummy-virt (DT)
[   27.745918] pstate: 80400005 (Nzcv daif +PAN -UAO)
[   27.746151] pc : unwind_frame+0x78/0x128
[   27.746337] lr : unwind_frame+0x114/0x128
[   27.746511] sp : ffff00000c893a80
[   27.746685] x29: ffff00000c893a80 x28: 0000000000000000
[   27.746914] x27: ffffffffffffffff x26: 0000000000000001
[   27.747150] x25: 0000000000000049 x24: ffff000009662940
[   27.747375] x23: 0000000000000001 x22: ffff0000096496c8
[   27.747622] x21: ffff0000096496c8 x20: 000000000000000f
[   27.747851] x19: ffff00000c893ad0 x18: 0000000000000000
[   27.748078] x17: 0000000000000000 x16: 0000000000000000
[   27.748337] x15: 0000000000000000 x14: 0000000000000000
[   27.748563] x13: ffff80000a6b2500 x12: 0000000000300000
[   27.748790] x11: 0000000000281bad x10: 0000000000000528
[   27.749322] x9 : ffff80000c808400 x8 : 0000000000000030
[   27.749557] x7 : ffff80000a503520 x6 : 0000000000000528
[   27.749790] x5 : 0000000000019a3b x4 : ffff80000a6b2580
[   27.750010] x3 : ffff80000c1a0d80 x2 : 0000000000000001
[   27.750236] x1 : 000000000000000a x0 : 0000000000000000
[   27.750460] Call trace:
[   27.750587]  unwind_frame+0x78/0x128
[   27.750761]  get_wchan+0xa4/0x110
[   27.750926]  do_task_stat+0x8d4/0x9e8
[   27.751097]  proc_tgid_stat+0x40/0x50
[   27.751271]  proc_single_show+0x5c/0xb8
[   27.751448]  seq_read+0x138/0x450
[   27.751614]  __vfs_read+0x60/0x170
[   27.751782]  vfs_read+0x94/0x150
[   27.751948]  ksys_read+0x6c/0xd0
[   27.752129]  __arm64_sys_read+0x24/0x30
[   27.752309]  el0_svc_common+0x94/0xe8
[   27.752479]  el0_svc_handler+0x38/0x78
[   27.752642]  el0_svc+0x8/0xc
[   27.752804] ---[ end trace fdae6e80cafbfff9 ]---
  PID TTY          TIME CMD
 2401 hvc0     00:00:00 login
 2451 hvc0     00:00:00 bash
 2457 hvc0     00:00:00 ps

This is single cpu kvm guest, with arm64's defconfig and:
| CONFIG_GENERIC_TRACER=y
| CONFIG_FUNCTION_TRACER=y
| CONFIG_FUNCTION_GRAPH_TRACER=y

This doesn't happen with the same configuration on v4.20-rc4.


> diff --git a/arch/arm64/kernel/stacktrace.c b/arch/arm64/kernel/stacktrace.c
> index 7723dadf25be..1a29f2695ff2 100644
> --- a/arch/arm64/kernel/stacktrace.c
> +++ b/arch/arm64/kernel/stacktrace.c
> @@ -59,15 +59,17 @@ int notrace unwind_frame(struct task_struct *tsk, struct stackframe *frame)
>  #ifdef CONFIG_FUNCTION_GRAPH_TRACER
>  	if (tsk->ret_stack &&
>  			(frame->pc == (unsigned long)return_to_handler)) {
> -		if (WARN_ON_ONCE(frame->graph == -1))
> -			return -EINVAL;
> +		struct ftrace_ret_stack *ret_stack;
>  		/*
>  		 * This is a case where function graph tracer has
>  		 * modified a return address (LR) in a stack frame
>  		 * to hook a function return.
>  		 * So replace it to an original value.
>  		 */
> -		frame->pc = tsk->ret_stack[frame->graph--].ret;
> +		ret_stack = ftrace_graph_get_ret_stack(tsk, frame->graph++);
> +		if (WARN_ON_ONCE(!ret_stack))
> +			return -EINVAL;
> +		frame->pc = ret_stack->ret;
>  	}
>  #endif /* CONFIG_FUNCTION_GRAPH_TRACER */


Thanks,

James