Message-ID: <20190716184549.GA26084@redhat.com>
Date:   Tue, 16 Jul 2019 14:45:49 -0400
From:   Joe Lawrence <joe.lawrence@...hat.com>
To:     Miroslav Benes <mbenes@...e.cz>
Cc:     heiko.carstens@...ibm.com, gor@...ux.ibm.com,
        borntraeger@...ibm.com, linux-s390@...r.kernel.org,
        linux-kernel@...r.kernel.org, jpoimboe@...hat.com,
        jikos@...nel.org, pmladek@...e.com, nstange@...e.de,
        live-patching@...r.kernel.org
Subject: Re: [PATCH] s390/livepatch: Implement reliable stack tracing for the
 consistency model

On Wed, Jul 10, 2019 at 12:59:18PM +0200, Miroslav Benes wrote:
> The livepatch consistency model requires reliable stack tracing
> architecture support in order to work properly. In order to achieve
> this, two main issues have to be solved. First, reliable and consistent
> call chain backtracing has to be ensured. Second, the unwinder needs to
> be able to detect stack corruptions and return errors.
> 
> The "zSeries ELF Application Binary Interface Supplement" says:
> 
>   "The stack pointer points to the first word of the lowest allocated
>   stack frame. If the "back chain" is implemented this word will point to
>   the previously allocated stack frame (towards higher addresses), except
>   for the first stack frame, which shall have a back chain of zero (NULL).
>   The stack shall grow downwards, in other words towards lower addresses."
> 
> "back chain" is optional. GCC option -mbackchain enables it. Quoting
> Martin Schwidefsky [1]:
> 
>   "The compiler is called with the -mbackchain option, all normal C
>   function will store the backchain in the function prologue. All
>   functions written in assembler code should do the same, if you find one
>   that does not we should fix that. The end result is that a task that
>   *voluntarily* called schedule() should have a proper backchain at all
>   times.
> 
>   Dependent on the use case this may or may not be enough. Asynchronous
>   interrupts may stop the CPU at the beginning of a function, if kernel
>   preemption is enabled we can end up with a broken backchain.  The
>   production kernels for IBM Z are all compiled *without* kernel
>   preemption. So yes, we might get away without the objtool support.
> 
>   On a side-note, we do have a line item to implement the ORC unwinder for
>   the kernel, that includes the objtool support. Once we have that we can
>   drop the -mbackchain option for the kernel build. That gives us a nice
>   little performance benefit. I hope that the change from backchain to the
>   ORC unwinder will not be too hard to implement in the livepatch tools."
> 
> Thus, the call chain backtracing should be currently ensured and objtool
> should not be necessary for livepatch purposes.

Hi Miroslav,

Should there be a CONFIG? dependency on -mbackchain and/or kernel
preemption, or does the following ensure that we don't need a explicit
build time checks?

> Regarding the second issue, stack corruptions and non-reliable states
> have to be recognized by the unwinder. Mainly it means to detect
> preemption or page faults, the end of the task stack must be reached,
> return addresses must be valid text addresses and hacks like function
> graph tracing and kretprobes must be properly detected.
> 
> Unwinding a running task's stack is not a problem, because there is a
> livepatch requirement that every checked task is blocked, except for the
> current task. Due to that, the implementation can be much simpler
> compared to the existing non-reliable infrastructure. We can consider a
> task's kernel/thread stack only and skip the other stacks.
> 
> Idle tasks are a bit special. Their final back chains point to no_dat
> stacks. See for reference CALL_ON_STACK() in smp_start_secondary()
> callback used in __cpu_up(). The unwinding is stopped there and it is
> not considered to be a stack corruption.
> 
> Signed-off-by: Miroslav Benes <mbenes@...e.cz>
> ---
> - based on Linus' master
> - passes livepatch kselftests
> - passes tests from https://github.com/lpechacek/qa_test_klp, which
>   stress the consistency model and the unwinder a bit more
> 
>  arch/s390/Kconfig                  |  1 +
>  arch/s390/include/asm/stacktrace.h |  5 ++
>  arch/s390/include/asm/unwind.h     | 19 ++++++
>  arch/s390/kernel/dumpstack.c       | 28 +++++++++
>  arch/s390/kernel/stacktrace.c      | 78 +++++++++++++++++++++++++
>  arch/s390/kernel/unwind_bc.c       | 93 ++++++++++++++++++++++++++++++
>  6 files changed, 224 insertions(+)
> 
> diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
> index fdb4246265a5..ea73e555063d 100644
> --- a/arch/s390/Kconfig
> +++ b/arch/s390/Kconfig
> @@ -170,6 +170,7 @@ config S390
>  	select HAVE_PERF_EVENTS
>  	select HAVE_RCU_TABLE_FREE
>  	select HAVE_REGS_AND_STACK_ACCESS_API
> +	select HAVE_RELIABLE_STACKTRACE
>  	select HAVE_RSEQ
>  	select HAVE_SYSCALL_TRACEPOINTS
>  	select HAVE_VIRT_CPU_ACCOUNTING
> diff --git a/arch/s390/include/asm/stacktrace.h b/arch/s390/include/asm/stacktrace.h
> index 0ae4bbf7779c..2b5c913c408f 100644
> --- a/arch/s390/include/asm/stacktrace.h
> +++ b/arch/s390/include/asm/stacktrace.h
> @@ -23,6 +23,11 @@ const char *stack_type_name(enum stack_type type);
>  int get_stack_info(unsigned long sp, struct task_struct *task,
>  		   struct stack_info *info, unsigned long *visit_mask);
>  
> +#ifdef CONFIG_HAVE_RELIABLE_STACKTRACE
> +int get_stack_info_reliable(unsigned long sp, struct task_struct *task,
> +			    struct stack_info *info);
> +#endif
> +
>  static inline bool on_stack(struct stack_info *info,
>  			    unsigned long addr, size_t len)
>  {
> diff --git a/arch/s390/include/asm/unwind.h b/arch/s390/include/asm/unwind.h
> index d827b5b9a32c..1cc96c54169c 100644
> --- a/arch/s390/include/asm/unwind.h
> +++ b/arch/s390/include/asm/unwind.h
> @@ -45,6 +45,25 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task,
>  bool unwind_next_frame(struct unwind_state *state);
>  unsigned long unwind_get_return_address(struct unwind_state *state);
>  
> +#ifdef CONFIG_HAVE_RELIABLE_STACKTRACE
> +void __unwind_start_reliable(struct unwind_state *state,
> +			     struct task_struct *task, unsigned long sp);
> +bool unwind_next_frame_reliable(struct unwind_state *state);
> +
> +static inline void unwind_start_reliable(struct unwind_state *state,
> +					 struct task_struct *task)
> +{
> +	unsigned long sp;
> +
> +	if (task == current)
> +		sp = current_stack_pointer();
> +	else
> +		sp = task->thread.ksp;
> +
> +	__unwind_start_reliable(state, task, sp);
> +}
> +#endif
> +
>  static inline bool unwind_done(struct unwind_state *state)
>  {
>  	return state->stack_info.type == STACK_TYPE_UNKNOWN;
> diff --git a/arch/s390/kernel/dumpstack.c b/arch/s390/kernel/dumpstack.c
> index ac06c3949ab3..b21ef2a766ff 100644
> --- a/arch/s390/kernel/dumpstack.c
> +++ b/arch/s390/kernel/dumpstack.c
> @@ -127,6 +127,34 @@ int get_stack_info(unsigned long sp, struct task_struct *task,
>  	return -EINVAL;
>  }
>  
> +#ifdef CONFIG_HAVE_RELIABLE_STACKTRACE
> +int get_stack_info_reliable(unsigned long sp, struct task_struct *task,
> +			    struct stack_info *info)
> +{
> +	if (!sp)
> +		goto error;
> +
> +	/* Sanity check: ABI requires SP to be aligned 8 bytes. */
> +	if (sp & 0x7)
> +		goto error;
> +

Does SP alignment only need to be checked for the initial frame, or
should it be verified everytime it's moved in
unwind_next_frame_reliable()?

> +	if (!task)
> +		goto error;
> +
> +	/*
> +	 * The unwinding should not start on nodat_stack, async_stack or
> +	 * restart_stack. The task is either current or must be inactive.
> +	 */
> +	if (!in_task_stack(sp, task, info))
> +		goto error;
> +
> +	return 0;
> +error:
> +	info->type = STACK_TYPE_UNKNOWN;
> +	return -EINVAL;
> +}
> +#endif
> +
>  void show_stack(struct task_struct *task, unsigned long *stack)
>  {
>  	struct unwind_state state;
> diff --git a/arch/s390/kernel/stacktrace.c b/arch/s390/kernel/stacktrace.c
> index f6a620f854e1..7d774a325163 100644
> --- a/arch/s390/kernel/stacktrace.c
> +++ b/arch/s390/kernel/stacktrace.c
> @@ -13,6 +13,7 @@
>  #include <linux/export.h>
>  #include <asm/stacktrace.h>
>  #include <asm/unwind.h>
> +#include <asm/kprobes.h>
>  
>  void save_stack_trace(struct stack_trace *trace)
>  {
> @@ -60,3 +61,80 @@ void save_stack_trace_regs(struct pt_regs *regs, struct stack_trace *trace)
>  	}
>  }
>  EXPORT_SYMBOL_GPL(save_stack_trace_regs);
> +
> +#ifdef CONFIG_HAVE_RELIABLE_STACKTRACE
> +/*
> + * This function returns an error if it detects any unreliable features of the
> + * stack.  Otherwise it guarantees that the stack trace is reliable.
> + *
> + * If the task is not 'current', the caller *must* ensure the task is inactive.
> + */
> +static __always_inline int
> +__save_stack_trace_tsk_reliable(struct task_struct *tsk,
> +				struct stack_trace *trace)
> +{
> +	struct unwind_state state;
> +
> +	for (unwind_start_reliable(&state, tsk);
> +	     !unwind_done(&state) && !unwind_error(&state);
> +	     unwind_next_frame_reliable(&state)) {
> +
> +		if (!__kernel_text_address(state.ip))
> +			return -EINVAL;
> +
> +#ifdef CONFIG_KPROBES
> +		/*
> +		 * Mark stacktraces with kretprobed functions on them
> +		 * as unreliable.
> +		 */
> +		if (state.ip == (unsigned long)kretprobe_trampoline)
> +			return -EINVAL;
> +#endif
> +
> +		if (trace->nr_entries >= trace->max_entries)
> +			return -E2BIG;
> +
> +		if (!trace->skip)
> +			trace->entries[trace->nr_entries++] = state.ip;
> +		else
> +			trace->skip--;
> +	}
> +
> +	/* Check for stack corruption */
> +	if (unwind_error(&state))
> +		return -EINVAL;
> +
> +	/* Store kernel_thread_starter, null for swapper/0 */
> +	if (tsk->flags & (PF_KTHREAD | PF_IDLE)) {
> +		if (trace->nr_entries >= trace->max_entries)
> +			return -E2BIG;
> +
> +		if (!trace->skip)
> +			trace->entries[trace->nr_entries++] =
> +				state.regs->psw.addr;
> +		else
> +			trace->skip--;

An idea for a follow up patch: stuff this into a function like
int save_trace_entry(struct stack_trace *trace, unsigned long entry);
which could one day make the trace->entries[] code generic across arches. 

> +	}
> +
> +	return 0;
> +}
> +
> +int save_stack_trace_tsk_reliable(struct task_struct *tsk,
> +				  struct stack_trace *trace)
> +{
> +	int ret;
> +
> +	/*
> +	 * If the task doesn't have a stack (e.g., a zombie), the stack is
> +	 * "reliably" empty.
> +	 */
> +	if (!try_get_task_stack(tsk))
> +		return 0;
> +
> +	ret = __save_stack_trace_tsk_reliable(tsk, trace);
> +
> +	put_task_stack(tsk);
> +
> +	return ret;
> +}
> +#endif
> diff --git a/arch/s390/kernel/unwind_bc.c b/arch/s390/kernel/unwind_bc.c
> index 3ce8a0808059..ada3a8538961 100644
> --- a/arch/s390/kernel/unwind_bc.c
> +++ b/arch/s390/kernel/unwind_bc.c
> @@ -153,3 +153,96 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task,
>  	state->reliable = reliable;
>  }
>  EXPORT_SYMBOL_GPL(__unwind_start);
> +
> +#ifdef CONFIG_HAVE_RELIABLE_STACKTRACE
> +void __unwind_start_reliable(struct unwind_state *state,
> +			     struct task_struct *task, unsigned long sp)
> +{
> +	struct stack_info *info = &state->stack_info;
> +	struct stack_frame *sf;
> +	unsigned long ip;
> +
> +	memset(state, 0, sizeof(*state));
> +	state->task = task;
> +
> +	/* Get current stack pointer and initialize stack info */
> +	if (get_stack_info_reliable(sp, task, info) ||
> +	    !on_stack(info, sp, sizeof(struct stack_frame))) {
> +		/* Something is wrong with the stack pointer */
> +		info->type = STACK_TYPE_UNKNOWN;
> +		state->error = true;
> +		return;
> +	}
> +
> +	/* Get the instruction pointer from the stack frame */
> +	sf = (struct stack_frame *) sp;
> +	ip = READ_ONCE_NOCHECK(sf->gprs[8]);
> +
> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
> +	/* Decode any ftrace redirection */
> +	if (ip == (unsigned long) return_to_handler)
> +		ip = ftrace_graph_ret_addr(state->task, &state->graph_idx,
> +					   ip, NULL);
                                               ^^^^
double checking: we ignore the retp here and not in the next-frame case?

> +#endif
> +
> +	/* Update unwind state */
> +	state->sp = sp;
> +	state->ip = ip;
> +}
> +
> +bool unwind_next_frame_reliable(struct unwind_state *state)
> +{
> +	struct stack_info *info = &state->stack_info;
> +	struct stack_frame *sf;
> +	struct pt_regs *regs;
> +	unsigned long sp, ip;
> +
> +	sf = (struct stack_frame *) state->sp;
> +	sp = READ_ONCE_NOCHECK(sf->back_chain);
> +	/*
> +	 * Idle tasks are special. The final back-chain points to nodat_stack.
> +	 * See CALL_ON_STACK() in smp_start_secondary() callback used in
> +	 * __cpu_up(). We just accept it, go to else branch and look for
> +	 * pt_regs.
> +	 */
> +	if (likely(sp && !(is_idle_task(state->task) &&
> +			   outside_of_stack(state, sp)))) {
> +		/* Non-zero back-chain points to the previous frame */
> +		if (unlikely(outside_of_stack(state, sp)))
> +			goto out_err;
> +
> +		sf = (struct stack_frame *) sp;
> +		ip = READ_ONCE_NOCHECK(sf->gprs[8]);
> +	} else {
> +		/* No back-chain, look for a pt_regs structure */
> +		sp = state->sp + STACK_FRAME_OVERHEAD;
> +		regs = (struct pt_regs *) sp;
> +		if ((unsigned long)regs != info->end - sizeof(struct pt_regs))
> +			goto out_err;
> +		if (!(state->task->flags & (PF_KTHREAD | PF_IDLE)) &&
> +		     !user_mode(regs))
> +			goto out_err;
> +
> +		state->regs = regs;
> +		goto out_stop;
> +	}
> +
> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
> +	/* Decode any ftrace redirection */
> +	if (ip == (unsigned long) return_to_handler)
> +		ip = ftrace_graph_ret_addr(state->task, &state->graph_idx,
> +					   ip, (void *) sp);
> +#endif
> +
> +	/* Update unwind state */
> +	state->sp = sp;
> +	state->ip = ip;

minor nit: maybe the CONFIG_FUNCTION_GRAPH_TRACER and "Update unwind
state" logic could be combined into a function?  (Not a big deal either
way.)

> +	return true;
> +
> +out_err:
> +	state->error = true;
> +out_stop:
> +	state->stack_info.type = STACK_TYPE_UNKNOWN;
> +	return false;
> +}
> +#endif
> -- 
> 2.22.0
> 

I've tested the patch with positive results, however I didn't stress it
very hard (basically only selftests).  The code logic seems
straightforward and correct by inspection.

On a related note, do you think it would be feasible to extend (in
another patchset) the reliable stack unwinding code a bit so that we
could feed it pre-baked stacks ... then we could verify that the code
was finding interesting scenarios.  That was a passing thought I had
back when Nicolai and I were debugging the ppc64le exception frame
marker bug, but didn't think it worth the time/effort at the time.

One more note:  Using READ_ONCE_NOCHECK is probably correct here, but
s390 happens to define a READ_ONCE_TASK_STACK macro which calls
READ_ONCE_NOCHECK when task != current.  According to the code comments,
this "disables KASAN checking when reading a value from another task's
stack".  Is there any scenario here where we would want to use the that
wrapper macro?

-- Joe

Powered by blists - more mailing lists