Re: [PATCH v2] tracing: Function stack size and its name mismatch in arm64

2019-08-02 Thread Steven Rostedt
On Fri, 2 Aug 2019 10:43:03 +0800 Jiping Ma wrote: > > *Why* does the frame appear to be off-by-one? > Because the PC is LR in ARM64 stack actually.  Following is ARM64 stack > layout. Please refer to the figure 3 in > http://infocenter.arm.com/help/topic/com.arm.doc.ihi0055b/IHI0055B_aapcs64

Re: [PATCH v2] tracing: Function stack size and its name mismatch in arm64

2019-08-01 Thread Jiping Ma
On 2019年08月01日 17:41, Will Deacon wrote: On Thu, Aug 01, 2019 at 04:33:40PM +0800, Jiping Ma wrote: In arm64, the PC of the frame is matched to the last frame function, rather than the function of his frame. For the following example, the stack size of occupy_stack_init function should be 337

Re: [PATCH v2] tracing: Function stack size and its name mismatch in arm64

2019-08-01 Thread Steven Rostedt
On Thu, 1 Aug 2019 16:33:40 +0800 Jiping Ma wrote: > diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c > index 5d16f73898db..ed80b95abf06 100644 > --- a/kernel/trace/trace_stack.c > +++ b/kernel/trace/trace_stack.c > @@ -40,16 +40,28 @@ static void print_max_stack(void) > >

Re: [PATCH v2] tracing: Function stack size and its name mismatch in arm64

2019-08-01 Thread Will Deacon
On Thu, Aug 01, 2019 at 04:33:40PM +0800, Jiping Ma wrote: > In arm64, the PC of the frame is matched to the last frame function, > rather than the function of his frame. For the following example, the > stack size of occupy_stack_init function should be 3376, rather than 176. > > Wrong info: >

[PATCH v2] tracing: Function stack size and its name mismatch in arm64

2019-08-01 Thread Jiping Ma
In arm64, the PC of the frame is matched to the last frame function, rather than the function of his frame. For the following example, the stack size of occupy_stack_init function should be 3376, rather than 176. Wrong info: Depth Size Location (16 entries) - 0) 5400 16 __updat