Re: perf probe: can't find unnamed union members

[Masami Hiramatsu]

Hi,

(2015/03/06 0:32), Arnaldo Carvalho de Melo wrote:
> 
> Hi Masami,
> 
> [root@zoo ~]# perf probe -x ~/bin/perf -L lock__delete <lock__delete@/home/git/linux/tools/perf/util/annotate.c:0>
>       0  static void lock__delete(struct ins_operands *ops)
>       1  {
>       2         struct ins *ins = ops->locked.ins;
>          
>       4         if (ins && ins->ops->free)
>       5                 ins->ops->free(ops->locked.ops);
>                 else
>       7                 ins__delete(ops->locked.ops);
>          
>       9         zfree(&ops->locked.ops);
>      10         zfree(&ops->target.raw);
>      11         zfree(&ops->target.name);
>      12  }
> 
> And:
> 
> [root@zoo ~]# pahole ~/bin/perf -C ins_operands
> struct ins_operands {
> 	char *                     raw;                  /*     0     8 */
> 	struct {
> 		char *             raw;                  /*     8     8 */
> 		char *             name;                 /*    16     8 */
> 		u64                addr;                 /*    24     8 */
> 		u64                offset;               /*    32     8 */
> 	} target;                                        /*     8    32 */
> 	union {
> 		struct {
> 			char *     raw;                  /*    40     8 */
> 			char *     name;                 /*    48     8 */
> 			u64        addr;                 /*    56     8 */
> 		} source;                                /*          24 */
> 		struct {
> 			struct ins * ins;                /*    40     8 */
> 			struct ins_operands * ops;       /*    48     8 */
> 		} locked;                                /*          16 */
> 	};                                               /*    40    24 */
> 	/* --- cacheline 1 boundary (64 bytes) --- */
> 
> 	/* size: 64, cachelines: 1, members: 3 */
> };
> [root@zoo ~]#
> 
> But:
> 
> [root@zoo ~]# perf probe -x ~/bin/perf lock__delete ops 'locked_ops=ops->locked.ops'
> ops(type:ins_operands) has no member locked.
>   Error: Failed to add events.
> [root@zoo ~]# 
> 
> Yes, it has :-)

Oh, I see. That is not yet supported.
Here, I got the structure in debuginfo.

 [ fac4e]    structure_type
             name                 (strp) "ins_operands"
             byte_size            (data1) 64
             decl_file            (data1) 5
             decl_line            (data1) 16
             sibling              (ref4) [ fac79]
 [ fac5a]      member
               name                 (string) "raw"
               decl_file            (data1) 5
               decl_line            (data1) 17
               type                 (ref4) [ f8bdb]
               data_member_location (data1) 0
 [ fac66]      member
               name                 (strp) "target"
               decl_file            (data1) 5
               decl_line            (data1) 23
               type                 (ref4) [ fab9c]
               data_member_location (data1) 8
 [ fac72]      member					<= here is the unnamed one
               type                 (ref4) [ fac7f]
               data_member_location (data1) 28
 [ fac79]    pointer_type
             byte_size            (data1) 8
             type                 (ref4) [ fac4e]
 [ fac7f]    union_type
             byte_size            (data1) 24
             decl_file            (data1) 5
             decl_line            (data1) 24
             sibling              (ref4) [ fac9e]
 [ fac87]      member
               name                 (strp) "source"
               decl_file            (data1) 5
               decl_line            (data1) 29
               type                 (ref4) [ fabd5]
 [ fac92]      member					<= here is actual member
               name                 (strp) "locked"
               decl_file            (data1) 5
               decl_line            (data1) 33
               type                 (ref4) [ fac02]

Let me see how I can solve this...

Thank you for reporting!

> 
> - Arnaldo
> 


-- 
Masami HIRAMATSU
Software Platform Research Dept. Linux Technology Research Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@xxxxxxxxxxx


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/