From: Namhyung Kim <namhyung@kernel.org>
To: Yanbo Zhao <yzhao62@ncsu.edu>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>,
Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>,
Mark Rutland <mark.rutland@arm.com>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Jiri Olsa <jolsa@kernel.org>, Ian Rogers <irogers@google.com>,
Adrian Hunter <adrian.hunter@intel.com>,
James Clark <james.clark@linaro.org>, Zecheng Li <zli94@ncsu.edu>,
Xu Liu <xliuprof@google.com>,
linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/3] perf dwarf-aux: Add C++ class, inheritance and vtable helpers
Date: Mon, 24 Aug 2026 14:04:06 -0700 [thread overview]
Message-ID: <aoyxxgALuI_ljpQI@google.com> (raw)
In-Reply-To: <20260821050207.4517-3-yzhao62@ncsu.edu>
On Fri, Aug 21, 2026 at 01:02:06AM -0400, Yanbo Zhao wrote:
> Add DWARF helper functions for C++ virtual function support:
> - cu_get_language() and cu_is_cplusplus() to detect source language.
> - die_get_base_class() and die_get_vtable_index() for virtual table
> inspection.
> - die_get_parent() to retrieve enclosing scopes.
> - die_find_member_by_offset() and die_find_virtual_func() to locate
> class members and virtual function DIEs.
> - die_is_vptr_member() to check if a member is a virtual table pointer.
> - Define DW_AT_vtable_elem_index and DW_LANG_C_plus_plus_17/20
> fallbacks for older elfutils that lack these DWARF 5 constants.
>
> Signed-off-by: Yanbo Zhao <yzhao62@ncsu.edu>
> ---
> tools/perf/util/dwarf-aux.c | 192 ++++++++++++++++++++++++++++++++++++
> tools/perf/util/dwarf-aux.h | 24 +++++
> 2 files changed, 216 insertions(+)
>
> diff --git a/tools/perf/util/dwarf-aux.c b/tools/perf/util/dwarf-aux.c
> index 88ba0ec23a68..e991eed963a8 100644
> --- a/tools/perf/util/dwarf-aux.c
> +++ b/tools/perf/util/dwarf-aux.c
> @@ -12,6 +12,19 @@
> #include "dwarf-regs.h"
> #include "strbuf.h"
> #include "string2.h"
> +#include "symbol.h"
> +
> +#ifndef DW_AT_vtable_elem_index
> +#define DW_AT_vtable_elem_index 0x8c
> +#endif
I'm not sure about this. My system has dwarf.h with the following.
DW_AT_vtable_elem_location = 0x4d,
DW_AT_loclists_base = 0x8c,
> +
> +/* Added in DWARF 5, may be missing in older elfutils */
> +#ifndef DW_LANG_C_plus_plus_17
> +#define DW_LANG_C_plus_plus_17 0x2a
> +#endif
> +#ifndef DW_LANG_C_plus_plus_20
> +#define DW_LANG_C_plus_plus_20 0x2b
> +#endif
Also, these are enum constants, not macros. So it won't be defined even
on recent versions. And there are DW_LANG_C_plus_plus_23 (0x3a) too.
>
> /**
> * cu_find_realpath - Find the realpath of the target file
> @@ -60,6 +73,37 @@ const char *cu_get_comp_dir(Dwarf_Die *cu_die)
> return dwarf_formstring(&attr);
> }
>
> +Dwarf_Word cu_get_language(Dwarf_Die *cu_die)
> +{
> + Dwarf_Attribute attr;
> + Dwarf_Word lang;
> +
> + if (dwarf_attr(cu_die, DW_AT_language, &attr) == NULL)
> + return 0;
> +
> + if (dwarf_formudata(&attr, &lang) != 0)
> + return 0;
> +
> + return lang;
> +}
If it's not used elsewhere, probably better to make it static.
> +
> +bool cu_is_cplusplus(Dwarf_Die *cu_die)
> +{
> + Dwarf_Word lang = cu_get_language(cu_die);
> +
> + switch (lang) {
> + case DW_LANG_C_plus_plus:
> + case DW_LANG_C_plus_plus_03:
> + case DW_LANG_C_plus_plus_11:
> + case DW_LANG_C_plus_plus_14:
> + case DW_LANG_C_plus_plus_17:
> + case DW_LANG_C_plus_plus_20:
> + return true;
> + default:
> + return false;
> + }
> +}
> +
> bool die_is_compound_type(Dwarf_Die *type_die)
> {
> int tag = dwarf_tag(type_die);
> @@ -68,6 +112,90 @@ bool die_is_compound_type(Dwarf_Die *type_die)
> tag == DW_TAG_class_type;
> }
>
> +static int __die_find_inheritance_cb(Dwarf_Die *die_mem, void *arg)
> +{
> + int tag = dwarf_tag(die_mem);
> +
> + if (tag == DW_TAG_inheritance) {
> + *(Dwarf_Die *)arg = *die_mem;
> + return DIE_FIND_CB_END;
> + }
> +
> + return DIE_FIND_CB_SIBLING;
> +}
> +
> +Dwarf_Die *die_get_base_class(Dwarf_Die *class_die, Dwarf_Die *base_die, int *offset)
> +{
> + Dwarf_Die inherit_die;
> + Dwarf_Attribute attr;
> + Dwarf_Word loc;
> +
> + if (die_find_child(class_die, __die_find_inheritance_cb,
> + &inherit_die, &inherit_die) == NULL)
> + return NULL;
> +
> + if (__die_get_real_type(&inherit_die, base_die) == NULL)
> + return NULL;
> +
> + if (dwarf_attr_integrate(&inherit_die, DW_AT_data_member_location, &attr) &&
> + dwarf_formudata(&attr, &loc) == 0) {
As Sashiko said, you need to handle other formats.
> + *offset = loc;
> + } else {
> + *offset = 0;
> + }
> +
> + return base_die;
> +}
> +
> +int die_get_vtable_index(Dwarf_Die *func_die, int *index)
It'd be great to clarify what it returns. It's not clear to me what the
index means here. Is it for an index of the @func_die in the vtable of
the current class?
> +{
> + Dwarf_Attribute attr;
> + Dwarf_Word idx;
> +
> + /* Try DW_AT_vtable_elem_index first (DWARF 5+) */
> + if (dwarf_attr(func_die, DW_AT_vtable_elem_index, &attr) &&
> + dwarf_formudata(&attr, &idx) == 0) {
> + *index = idx;
> + return 0;
> + }
> +
> + /* Fallback to DW_AT_vtable_elem_location (older DWARF) */
> + if (dwarf_attr(func_die, DW_AT_vtable_elem_location, &attr)) {
According to the Sashiko review, it would return a byte offset instead
of an index. If so, it should be converted for consistency.
> + Dwarf_Op *expr;
> + size_t expr_len;
> +
> + /* Compile often emits it as a simple constant expression or block */
> + if (dwarf_getlocation(&attr, &expr, &expr_len) == 0 && expr_len > 0) {
> + if (expr[0].atom == DW_OP_constu) {
> + *index = expr[0].number;
> + return 0;
> + }
> + if (expr[0].atom >= DW_OP_lit0 && expr[0].atom <= DW_OP_lit31) {
> + *index = expr[0].atom - DW_OP_lit0;
> + return 0;
> + }
> + }
> + }
> +
> + return -1;
> +}
> +
> +Dwarf_Die *die_get_parent(Dwarf_Die *die, Dwarf_Die *parent_die)
> +{
> + Dwarf_Die *scopes = NULL;
> + int n = dwarf_getscopes_die(die, &scopes);
> +
> + if (n <= 1) {
> + free(scopes);
> + return NULL;
> + }
> +
> + /* scopes[0] is the DIE itself, scopes[1] is the parent */
> + *parent_die = scopes[1];
> + free(scopes);
> + return parent_die;
> +}
Where is this function used?
> +
> /* Unlike dwarf_getsrc_die(), cu_getsrc_die() only returns statement line */
> static Dwarf_Line *cu_getsrc_die(Dwarf_Die *cu_die, Dwarf_Addr addr)
> {
> @@ -2164,6 +2292,70 @@ Dwarf_Die *die_get_member_type(Dwarf_Die *type_die, int offset,
> return die_mem;
> }
>
> +Dwarf_Die *die_find_member_by_offset(Dwarf_Die *type_die, int offset, Dwarf_Die *member_die)
> +{
> + if (!die_is_compound_type(type_die))
> + return NULL;
> +
> + return die_find_child(type_die, __die_find_member_offset_cb,
> + (void *)(long)offset, member_die);
> +}
Isn't it basically the same as die_get_member_type()?
Thanks,
Namhyung
> +
> +struct find_virtual_func_data {
> + int index;
> + Dwarf_Die func_die;
> + bool found;
> +};
> +
> +static int __die_find_virtual_func_cb(Dwarf_Die *die_mem, void *arg)
> +{
> + struct find_virtual_func_data *ad = arg;
> + int tag = dwarf_tag(die_mem);
> +
> + if (tag == DW_TAG_subprogram) {
> + int idx;
> +
> + if (die_get_vtable_index(die_mem, &idx) == 0 && idx == ad->index) {
> + ad->func_die = *die_mem;
> + ad->found = true;
> + return DIE_FIND_CB_END;
> + }
> + }
> +
> + if (tag == DW_TAG_inheritance) {
> + Dwarf_Die base_type;
> +
> + if (__die_get_real_type(die_mem, &base_type)) {
> + if (die_find_child(&base_type,
> + __die_find_virtual_func_cb,
> + ad, &ad->func_die))
> + return DIE_FIND_CB_END;
> + }
> + }
> +
> + return DIE_FIND_CB_SIBLING;
> +}
> +
> +Dwarf_Die *die_find_virtual_func(Dwarf_Die *class_die, int index, Dwarf_Die *die_mem)
> +{
> + struct find_virtual_func_data ad = {
> + .index = index,
> + .found = false,
> + };
> +
> + if (die_find_child(class_die, __die_find_virtual_func_cb, &ad, die_mem))
> + return die_mem;
> +
> + return NULL;
> +}
> +
> +bool die_is_vptr_member(Dwarf_Die *die)
> +{
> + const char *name = dwarf_diename(die);
> +
> + return name && strstarts(name, "_vptr");
> +}
> +
> /**
> * die_deref_ptr_type - Return type info for pointer access
> * @ptr_die: a pointer type DIE
> diff --git a/tools/perf/util/dwarf-aux.h b/tools/perf/util/dwarf-aux.h
> index 855c45fec5bb..0f30f44ad863 100644
> --- a/tools/perf/util/dwarf-aux.h
> +++ b/tools/perf/util/dwarf-aux.h
> @@ -23,9 +23,33 @@ const char *cu_find_realpath(Dwarf_Die *cu_die, const char *fname);
> /* Get DW_AT_comp_dir (should be NULL with older gcc) */
> const char *cu_get_comp_dir(Dwarf_Die *cu_die);
>
> +/* Get source language of CU */
> +Dwarf_Word cu_get_language(Dwarf_Die *cu_die);
> +
> +/* Check if CU is C++ */
> +bool cu_is_cplusplus(Dwarf_Die *cu_die);
> +
> /* Check if DIE is a compound type (structure, union, or class) */
> bool die_is_compound_type(Dwarf_Die *type_die);
>
> +/* Get base class of a class DIE */
> +Dwarf_Die *die_get_base_class(Dwarf_Die *class_die, Dwarf_Die *base_die, int *offset);
> +
> +/* Get vtable index of a virtual function */
> +int die_get_vtable_index(Dwarf_Die *func_die, int *index);
> +
> +/* Get parent DIE */
> +Dwarf_Die *die_get_parent(Dwarf_Die *die, Dwarf_Die *parent_die);
> +
> +/* Find member DIE by offset */
> +Dwarf_Die *die_find_member_by_offset(Dwarf_Die *type_die, int offset, Dwarf_Die *member_die);
> +
> +/* Find virtual function by vtable index */
> +Dwarf_Die *die_find_virtual_func(Dwarf_Die *class_die, int index, Dwarf_Die *die_mem);
> +
> +/* Check if member DIE is a vtable pointer (_vptr) */
> +bool die_is_vptr_member(Dwarf_Die *die);
> +
> /* Get a line number and file name for given address */
> int cu_find_lineinfo(Dwarf_Die *cudie, Dwarf_Addr addr,
> const char **fname, int *lineno);
> --
> 2.25.1
>
next prev parent reply other threads:[~2026-08-24 21:04 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-21 5:02 [PATCH 0/3] perf annotate: Data type profiling support for C++ classes and virtual calls Yanbo Zhao
2026-08-21 5:02 ` [PATCH 1/3] perf dwarf-aux: Add die_is_compound_type() to handle C++ class types Yanbo Zhao
2026-08-21 5:13 ` sashiko-bot
2026-08-24 20:35 ` Namhyung Kim
2026-08-21 5:02 ` [PATCH 2/3] perf dwarf-aux: Add C++ class, inheritance and vtable helpers Yanbo Zhao
2026-08-21 5:12 ` sashiko-bot
2026-08-24 21:04 ` Namhyung Kim [this message]
2026-08-21 5:02 ` [PATCH 3/3] perf disasm & annotate x86: Track C++ vptr and resolve virtual function calls Yanbo Zhao
2026-08-21 5:26 ` sashiko-bot
2026-08-24 21:09 ` Namhyung Kim
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aoyxxgALuI_ljpQI@google.com \
--to=namhyung@kernel.org \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=irogers@google.com \
--cc=james.clark@linaro.org \
--cc=jolsa@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=xliuprof@google.com \
--cc=yzhao62@ncsu.edu \
--cc=zli94@ncsu.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox