linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Oleg Nesterov <oleg@redhat.com>
To: Chris Li <sparse@chrisli.org>, Luc Van Oostenryck <lucvoo@kernel.org>
Cc: Alexey Gladkov <legion@kernel.org>, linux-sparse@vger.kernel.org
Subject: [PATCH 4/4] sparse/dissect: sanitize the de-anonymization of SYM_ENUMs
Date: Sun, 4 Jan 2026 16:22:08 +0100	[thread overview]
Message-ID: <aVqFoD7JKxylHanE@redhat.com> (raw)
In-Reply-To: <aVqFaL88QonZYs63@redhat.com>

examine_sym_node() tries to de-anonymize enums just like structs,
and sometimes it makes sense:

	$ ~/tmp/sparse/test-dissect - <<<'enum { X } x;'

	   1:6                    def   E :x                               unsigned int
	   1:8                    def   e :x.X                             unsigned int enum :x
	   1:12                   def   v x                                unsigned int enum :x

But it doesn't make sense to de-anonymize enum types using the name
of its enum_member.

Test-case:

	$ cat -n ENUM_3.c
	     1	enum { A, B };
	     2	typeof(B) x = A;

	$ ./test-dissect ENUM_3.c

	   1:6                    def   E :B                               unsigned int
	   1:8                    def   e :B.A                             unsigned int enum :B
	   1:11                   def   e :B.B                             unsigned int enum :B
	   2:8                    ---   e :B.B                             unsigned int enum :B
	   2:11                   def   v x                                int
	   2:11                   -w-   v x                                int
	   2:15  x                -r-   e :B.A                             unsigned int enum :B

With this patch:

	$ ./test-dissect ENUM_3.c

	   1:8                    def   e A                                unsigned int enum <noident>
	   1:11                   def   e B                                unsigned int enum <noident>
	   2:8                    ---   e B                                unsigned int enum <noident>
	   2:11                   def   v x                                int
	   2:11                   -w-   v x                                int
	   2:15  x                -r-   e A                                unsigned int enum <noident>

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 dissect.c      |  5 +++--
 test-dissect.c | 13 +++++++------
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/dissect.c b/dissect.c
index d36986dd..a0fda09c 100644
--- a/dissect.c
+++ b/dissect.c
@@ -245,10 +245,11 @@ static void examine_sym_node(struct symbol *node, struct symbol *parent)
 			if (toplevel(base->scope))
 				dissect_ctx = NULL;
 
-			if (base->ident || deanon(base, name, parent))
+			if (base->ident || (!node->enum_member &&
+					    deanon(base, name, parent)))
 				reporter->r_symdef(base);
 
-			if (base->ident)
+			if (base->ident || base->type == SYM_ENUM)
 				parent = base;
 			DO_LIST(base->symbol_list, mem,
 				examine_sym_node(mem, parent);
diff --git a/test-dissect.c b/test-dissect.c
index e2b8ec9d..0a7ee98c 100644
--- a/test-dissect.c
+++ b/test-dissect.c
@@ -88,19 +88,20 @@ err:
 static void r_member(unsigned mode, struct position *pos, struct symbol *sym, struct symbol *mem)
 {
 	struct ident *ni, *si, *mi;
-	int mk;
+	int mk, ms = 32;
 
 	ni = built_in_ident("?");
-	si = sym->ident ?: ni;
 	/* mem == NULL means entire struct accessed */
-	mi = mem ? (mem->ident ?: ni) : built_in_ident("*");
 	mk = mem ? mem->kind : 'm';
+	mi = mem ? (mem->ident ?: ni) : built_in_ident("*");
+	si = sym->ident ?: (mk == 'e' ? NULL : ni);
 
 	print_usage(pos, sym, mode, mk);
 
-	printf("%.*s.%-*.*s %s\n",
-		si->len, si->name,
-		32-1 - si->len, mi->len, mi->name,
+	if (si)
+		ms -= printf("%.*s.", si->len, si->name);
+
+	printf("%-*.*s %s\n", ms, mi->len, mi->name,
 		show_typename(mem ? mem->ctype.base_type : sym));
 
 	if (sym->ident && sym->kind != 's' && sym->kind != 'E')
-- 
2.52.0


      parent reply	other threads:[~2026-01-04 15:22 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-04 15:21 [PATCH 0/4] sparse/dissect: handle SYM_ENUM Oleg Nesterov
2026-01-04 15:21 ` [PATCH 1/4] sparse/dissect: examine SYM_ENUM nodes Oleg Nesterov
2026-01-04 15:22 ` [PATCH 2/4] sparse/dissect: report the usage of enumerators Oleg Nesterov
2026-01-04 15:22 ` [PATCH 3/4] sparse/dissect: shift symscope() callsites into print_usage() Oleg Nesterov
2026-01-04 15:22 ` Oleg Nesterov [this message]

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=aVqFoD7JKxylHanE@redhat.com \
    --to=oleg@redhat.com \
    --cc=legion@kernel.org \
    --cc=linux-sparse@vger.kernel.org \
    --cc=lucvoo@kernel.org \
    --cc=sparse@chrisli.org \
    /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;
as well as URLs for NNTP newsgroup(s).