linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Christopher Li" <sparse@chrisli.org>
To: Alexey Dobriyan <adobriyan@sw.ru>, Josh Triplett <josh@freedesktop.org>
Cc: linux-sparse@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH] Re: sparse breakage triggered by rcu_read_lock() lockdep annotations
Date: Fri, 19 Oct 2007 14:08:16 -0700	[thread overview]
Message-ID: <70318cbf0710191408mfb372bah292028f7bafb6eec@mail.gmail.com> (raw)
In-Reply-To: <70318cbf0710191244v59bed9fekbef0ae1fcda22a6a@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1870 bytes --]

OK, I get a trivial fix after all. The test case is fixed now.
I haven't done much test otherwise.

See the patch attached.

Chris


On 10/19/07, Chris Li <christ.li@gmail.com> wrote:
> Err,
>
> Sparse does not support the local label syntax yet. It just treats the
> second label "x:" as the same as the first one. Then the linearize
> code gets serious confused when it saw one label get define in two
> places.
>
> The fix seems not trivial from the first look.
>
> Chris
>
> On 10/16/07, Alexey Dobriyan <adobriyan@sw.ru> wrote:
> > FWIW, commit 851a67b825540a8e00c0be3ee25e4627ba8b133b
> > aka "lockdep: annotate rcu_read_{,un}lock{,_bh}"
> > causes sparse to trigger internal assertion in quite a few places over
> > allyesconfig run.
> >
> >         sparse: flow.c:805: rewrite_parent_branch: Assertion `changed' failed.
> >
> > Trimmed down testcase:
> >
> >         void f(unsigned long ip);
> >         static void g(void)
> >         {
> >                 if (1) {
> >                       f(({ __label__ x; x: (unsigned long)&&x; }));
> >                 }
> >                 f(({ __label__ x; x: (unsigned long)&&x; }));
> >         }
> >
> > #0  0x4001c410 in __kernel_vsyscall ()
> > (gdb) bt
> > #0  0x4001c410 in __kernel_vsyscall ()
> > #1  0x40050701 in raise () from /lib/libc.so.6
> > #2  0x40051e38 in abort () from /lib/libc.so.6
> > #3  0x40049fcc in __assert_fail () from /lib/libc.so.6
> > #4  0x08064947 in pack_basic_blocks (ep=0x411a1c6c) at flow.c:812
> > #5  0x0805ffbf in linearize_symbol (sym=0x4103ec8c) at linearize.c:2154
> > #6  0x080492a3 in main (argc=Cannot access memory at address 0x274d) at sparse.c:266
> >
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
>

[-- Attachment #2: local-label --]
[-- Type: application/octet-stream, Size: 1678 bytes --]

Perform local label lookup

This patch fix the sparse breakage triggered by
rcu_read_lock() lockdep annotations.

Now sparse look up the local label in symbol node
name space as well, just like looking up a normal
symbol node. Now a lable symbol can be both
type SYM_LABEL or SYM_NODE with MOD_LABEL.

Singed-Off-By: Christopher Li <sparse@chrisli.org>

Index: sparse/parse.c
===================================================================
--- sparse.orig/parse.c	2007-10-19 13:52:37.000000000 -0700
+++ sparse/parse.c	2007-10-19 13:55:23.000000000 -0700
@@ -459,6 +459,16 @@ static struct symbol *lookup_or_create_s
 	return sym;
 }
 
+static struct symbol * local_label(struct token *token)
+{
+	struct symbol *sym = lookup_symbol(token->ident, NS_SYMBOL);
+
+	if (sym && sym->ctype.modifiers & MOD_LABEL)
+		return sym;
+
+	return NULL;
+}
+
 /*
  * NOTE! NS_LABEL is not just a different namespace,
  * it also ends up using function scope instead of the
@@ -466,6 +476,9 @@ static struct symbol *lookup_or_create_s
  */
 struct symbol *label_symbol(struct token *token)
 {
+	struct symbol *sym = local_label(token);
+	if (sym)
+		return sym;
 	return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
 }
 
Index: sparse/validation/local-label.c
===================================================================
--- sparse.orig/validation/local-label.c	2007-10-19 13:52:40.000000000 -0700
+++ sparse/validation/local-label.c	2007-10-19 13:52:40.000000000 -0700
@@ -0,0 +1,8 @@
+void f(unsigned long ip);
+static void g(void)
+{
+       if (1) {
+	     f(({ __label__ x; x: (unsigned long)&&x; }));
+       }
+       f(({ __label__ x; x: (unsigned long)&&x; }));
+}

  reply	other threads:[~2007-10-19 21:08 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-10-16 10:55 sparse breakage triggered by rcu_read_lock() lockdep annotations Alexey Dobriyan
2007-10-18 17:41 ` Josh Triplett
2007-10-19 19:44 ` Chris Li
2007-10-19 21:08   ` Christopher Li [this message]
2007-10-21  3:34     ` [PATCH] " Josh Triplett

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=70318cbf0710191408mfb372bah292028f7bafb6eec@mail.gmail.com \
    --to=sparse@chrisli.org \
    --cc=adobriyan@sw.ru \
    --cc=josh@freedesktop.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-sparse@vger.kernel.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).