[Bug translator/12138] New: sdt_misc.exp fails most ptr types on i686

[Bug translator/12138] New: sdt_misc.exp fails most ptr types on i686

[jistone at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=12138

           Summary: sdt_misc.exp fails most ptr types on i686
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: translator
        AssignedTo: systemtap@sources.redhat.com
        ReportedBy: jistone@redhat.com


The sdt_types part of sdt_misc.exp fails on i686 for SDT V3

Host: Linux vm-f13-32 2.6.34.6-54.fc13.i686 #1 SMP Sun Sep 5 17:52:31 UTC 2010
i686 i686 i386 GNU/Linux
Snapshot: version 1.4/0.148 commit release-1.3-209-ga44a7cb
GCC: 4.4.4 [gcc (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)]
Distro: Fedora release 13 (Goddard)
[...]
spawn /usr/local/bin/stap -g -c ./sdt_types.x
/home/jistone/systemtap/testsuite/systemtap.base/sdt_types.stp ./sdt_types.x
PASS: char_var
PASS: const_char_var
PASS: volatile_char_var
FAIL: ptr_char_var
FAIL: ptr_const_char_var
FAIL: char_ptr_const_var
FAIL: ptr_volatile_char_var
PASS: char_ptr_volatile_var
PASS: short_int_var
PASS: const_short_int_var
PASS: volatile_short_int_var
FAIL: ptr_short_int_var
FAIL: ptr_const_short_int_var
FAIL: short_int_ptr_const_var
FAIL: ptr_volatile_short_int_var
PASS: short_int_ptr_volatile_var
PASS: int_var
PASS: const_int_var
PASS: volatile_int_var
FAIL: ptr_const_int_var
FAIL: ptr_const_int_var
FAIL: int_ptr_const_var
FAIL: ptr_volatile_int_var
PASS: int_ptr_volatile_var
PASS: long_int_var
PASS: const_long_int_var
PASS: volatile_long_int_var
FAIL: ptr_long_int_var
FAIL: ptr_const_long_int_var
FAIL: long_int_ptr_const_var
FAIL: ptr_volatile_long_int_var
PASS: long_int_ptr_volatile_var
PASS: long_long_int_var
PASS: const_long_long_int_var
PASS: volatile_long_long_int_var
FAIL: ptr_long_long_int_var
FAIL: ptr_const_long_long_int_var
FAIL: long_long_int_ptr_const_var
FAIL: ptr_volatile_long_long_int_var
PASS: long_long_int_ptr_volatile_var
PASS: arr_char_var
PASS: arr_struct_var
PASS: bitfields_bit_var
PASS: primary_colors_var
FAIL: sdt_misc types types  ptr_char_var  ptr_const_char_var 
char_ptr_const_var  ptr_volatile_char_var  ptr_short_int_var 
ptr_const_short_int_var  short_int_ptr_const_var  ptr_volatile_short_int_var 
ptr_const_int_var  ptr_const_int_var  int_ptr_const_var  ptr_volatile_int_var 
ptr_long_int_var  ptr_const_long_int_var  long_int_ptr_const_var 
ptr_volatile_long_int_var  ptr_long_long_int_var  ptr_const_long_long_int_var 
long_long_int_ptr_const_var  ptr_volatile_long_long_int_var V3_uprobe

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Bug translator/12138] sdt_misc.exp fails most ptr types on i686

[jistone at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=12138

--- Comment #1 from Josh Stone <jistone at redhat dot com> 2010-10-20 00:58:19 UTC ---
The test isn't very helpful about diagnostics, but I'll pick the first FAIL:

From sdt_types.c:
  char *ptr_char_var = &char_var;
  STAP_PROBE2(provider,ptr_char_var,ptr_char_var,&char_var);

From sdt_types.stp:
probe process(@1).mark("ptr_char_var") {
  if ($arg1 != $arg2)
    printf("FAIL: ptr_char_var\n")
  else
    printf("PASS: ptr_char_var\n")
}


$ stap -c ./sdt_types.x -e 'probe process("./sdt_types.x").mark("ptr_char_var")
{ printf("%p %p\n", $arg1, $arg2) }'
0xffffffffbf81541a 0xbf81541a

Along the way, the sign got extended...

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Bug translator/12138] sdt_misc.exp fails most ptr types on i686

[jistone at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=12138

--- Comment #2 from Josh Stone <jistone at redhat dot com> 2010-10-20 01:02:59 UTC ---
(In reply to comment #1)
> Along the way, the sign got extended...

Which reminds me, this same test has short() and int() functions which appear
to be kludging the sign-extension into shape for other parts of the test.  This
seems to be masking real problems...

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Bug translator/12138] sdt_misc.exp fails most ptr types on i686

[jistone at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=12138

--- Comment #3 from Josh Stone <jistone at redhat dot com> 2010-10-20 01:32:09 UTC ---
(In reply to comment #1)
> The test isn't very helpful about diagnostics, but I'll pick the first FAIL:
> 
> From sdt_types.c:
>   char *ptr_char_var = &char_var;
>   STAP_PROBE2(provider,ptr_char_var,ptr_char_var,&char_var);

Roland helped me figure out that .note.stapsdt gets:
.asciz "ptr_char_var"
.asciz "4@124(%esp) 4@%eax"

> printf("%p %p\n", $arg1, $arg2)

The translator does:
$arg1: user_int((%{ /* unprivileged */ /* pure */u_fetch_register(4) %}) +
(124))
$arg2: %{ /* unprivileged */ /* pure */u_fetch_register(0) %}

So the trip through user_int() extends the sign.  We could add user_uint(), if
there's enough info for the SDT-translator to know it should be unsigned.  I
think that's indicated by "4@..." instead of "-4@..."?

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Bug translator/12138] sdt_misc.exp fails most ptr types on i686

[scox at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=12138

Stan Cox <scox at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |scox at redhat dot com
         AssignedTo|systemtap at sources dot    |scox at redhat dot com
                   |redhat.com                  |

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.