Messages in this thread | | | Date | Wed, 25 Mar 2009 17:26:22 -0400 (EDT) | From | Steven Rostedt <> | Subject | Re: [PATCH][GIT PULL] x86, function-graph: only save return values on x86_64 |
| |
On Thu, 26 Mar 2009, Jaswinder Singh Rajput wrote: > > > > diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S > > index a331ec3..1ac9986 100644 > > --- a/arch/x86/kernel/entry_64.S > > +++ b/arch/x86/kernel/entry_64.S > > @@ -147,27 +147,14 @@ END(ftrace_graph_caller) > > GLOBAL(return_to_handler) > > subq $80, %rsp > > > > + /* Save the return values */ > > movq %rax, (%rsp) > > - movq %rcx, 8(%rsp) > > - movq %rdx, 16(%rsp) > > - movq %rsi, 24(%rsp) > > - movq %rdi, 32(%rsp) > > - movq %r8, 40(%rsp) > > - movq %r9, 48(%rsp) > > - movq %r10, 56(%rsp) > > - movq %r11, 64(%rsp) > > + movq %rdx, 8(%rsp) > > > > call ftrace_return_to_handler > > > > movq %rax, 72(%rsp) > > - movq 64(%rsp), %r11 > > - movq 56(%rsp), %r10 > > - movq 48(%rsp), %r9 > > - movq 40(%rsp), %r8 > > - movq 32(%rsp), %rdi > > - movq 24(%rsp), %rsi > > - movq 16(%rsp), %rdx > > - movq 8(%rsp), %rcx > > + movq 8(%rsp), %rdx > > movq (%rsp), %rax > > addq $72, %rsp > > retq > > hmm, is this gonna work: > movq %rax, 16(%rsp) > movq 8(%rsp), %rdx > movq (%rsp), %rax > addq $16, %rsp > retq
Interesting. I was just talking over IRC with Frederic about this. One theory about his earlier failures was not having a proper stack frame when calling the function. This is probably a bogus theory and there was probably something else that broke the code. But there is no harm in allocating 80 bytes (it worked before). Thus, I rather error on the side of least change than risk it.
-- Steve
| |