[Bug 191260] New: [patch] dtrace fbt entry function gets the wrong values from arg5 to arg9 on amd64 platform

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=191260

            Bug ID: 191260
           Summary: [patch] dtrace fbt entry function gets the wrong
                    values from arg5 to arg9 on amd64 platform
           Product: Base System
           Version: 11.0-CURRENT
          Hardware: Any
                OS: Any
            Status: Needs Triage
          Severity: Affects Some People
          Priority: ---
         Component: kern
          Assignee: freebsd-bugs at FreeBSD.org
          Reporter: luke.tw at gmail.com

Created attachment 144012
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=144012&action=edit
patch for dtrace_getarg()

There are two problems in the dtrace_getarg() implementation on amd64 platform.
In sys/cddl/dev/dtrace/amd64/dtrace_isa.c, 

1. dtrace_getarg() cannot find the dtrace_invop stack frame
    The return address of function dtrace_invop() may be different than
dtrace_invop_callsite, because the later is aligned on 16-byte boundary on
amd64 platform. As shown in the following disassembly code, there is 14 bytes
nop between them. 

00000000000249f0 <dtrace_invop_start>:
   249f0:   48 8b bc 24 98 00 00    mov    0x98(%rsp),%rdi
   249f7:   00
   249f8:   48 ff cf                dec    %rdi
   249fb:   48 8b b4 24 b0 00 00    mov    0xb0(%rsp),%rsi
   24a02:   00
   24a03:   48 8b 54 24 30          mov    0x30(%rsp),%rdx
   24a08:   ff 36                   pushq  (%rsi)
   24a0a:   48 89 e6                mov    %rsp,%rsi
   24a0d:   e8 00 00 00 00          callq  24a12 <dtrace_invop_start+0x22>
   24a12:   66 66 66 66 66 2e 0f    nopw   %cs:0x0(%rax,%rax,1)
   24a19:   1f 84 00 00 00 00 00

0000000000024a20 <dtrace_invop_callsite>:
   24a20:   48 83 c4 08             add    $0x8,%rsp
   24a24:   83 f8 01                cmp    $0x1,%eax

2. struct trapframe should be used to match the struct regs used in illumos.

* experiment:
   I write a simple kernel module with a function traceme to print its ten
arguments:

void traceme(long arg0, long arg1, long arg2, long arg3, long arg4, 
                    long arg5, long arg6, long arg7, long arg8, long arg9) {
        printf("test:%ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n",
                        arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8,
arg9);
}

   And the calling the function like this:

traceme(0,1,2,3,4,5,6,7,8,9);


* before patch
 # dtrace -n 'fbt:example:traceme:entry {printf("%d %d %d %d %d %d %d %d %d
%d\n", arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9);}'
dtrace: description 'fbt:example:traceme:entry ' matched 1 probe
CPU     ID                    FUNCTION:NAME
  0  48648                    traceme:entry 0 1 2 3 4 -2118041099 0 1 2 3

* after patch
 # dtrace -n 'fbt:example:traceme:entry {printf("%d %d %d %d %d %d %d %d %d
%d\n", arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9);}'
dtrace: description 'fbt:example:traceme:entry ' matched 1 probe
CPU     ID                    FUNCTION:NAME
  1  48648                    traceme:entry 0 1 2 3 4 5 6 7 8 9

-- 
You are receiving this mail because:
You are the assignee for the bug.