public inbox for linux-sh@vger.kernel.org
 help / color / mirror / Atom feed
From: Takashi Yoshii <yoshii.takashi@renesas.com>
To: linux-sh@vger.kernel.org
Subject: Re: [bug report] dead lock (?) occur on ap325 board
Date: Wed, 28 Jan 2009 09:29:13 +0000	[thread overview]
Message-ID: <49802569.4000808@renesas.com> (raw)
In-Reply-To: <uprie4cmj.wl%morimoto.kuninori@renesas.com>

Kunihiro's fix looks good for me.
I have accidentally(?) fixed this one just same way as him.

I don't know why the original code does 
> 	__res |= !__ex_flag;
to combine the results (counter and exception flag).
Logical-OR to scaler value is logically:) wrong.
# And using movt explicitly results stupid code generated anyway.

Because this __ex_flag is from "t" of movco, which means
 0: interrupted by int or exp.
 1: done successfully.
, it works as "Low-active" signal, and is not easy to handle
as a C-language conditon.

So, I think 
        int __done, __res;
        ....
        if (unlikely(!done || __res != 0))

is easier to read.
This fixes the deadlock issue Kunihiro reported (by coincident:).

# I'm not sure if this is suitable to be marked as "earlyclobber",
# but I leave it as-is, since nothing bad will occur.
/yoshii

Fix conditon checking expression of __mutex_fastpath_*

Signed-off-by: Takashi YOSHII <yoshii.takashi@renesas.com>
---
 arch/sh/include/asm/mutex-llsc.h |   21 +++++++++------------
 1 files changed, 9 insertions(+), 12 deletions(-)

diff --git a/arch/sh/include/asm/mutex-llsc.h b/arch/sh/include/asm/mutex-llsc.h
index ee839ee..66f045f 100644
--- a/arch/sh/include/asm/mutex-llsc.h
+++ b/arch/sh/include/asm/mutex-llsc.h
@@ -21,38 +21,36 @@
 static inline void
 __mutex_fastpath_lock(atomic_t *count, void (*fail_fn)(atomic_t *))
 {
-	int __ex_flag, __res;
+	int __done, __res;
 
 	__asm__ __volatile__ (
 		"movli.l	@%2, %0	\n"
 		"add		#-1, %0	\n"
 		"movco.l	%0, @%2	\n"
 		"movt		%1	\n"
-		: "=&z" (__res), "=&r" (__ex_flag)
+		: "=&z" (__res), "=&r" (__done)
 		: "r" (&(count)->counter)
 		: "t");
 
-	__res |= !__ex_flag;
-	if (unlikely(__res != 0))
+	if (unlikely(!__done || __res != 0))
 		fail_fn(count);
 }
 
 static inline int
 __mutex_fastpath_lock_retval(atomic_t *count, int (*fail_fn)(atomic_t *))
 {
-	int __ex_flag, __res;
+	int __done, __res;
 
 	__asm__ __volatile__ (
 		"movli.l	@%2, %0	\n"
 		"add		#-1, %0	\n"
 		"movco.l	%0, @%2	\n"
 		"movt		%1	\n"
-		: "=&z" (__res), "=&r" (__ex_flag)
+		: "=&z" (__res), "=&r" (__done)
 		: "r" (&(count)->counter)
 		: "t");
 
-	__res |= !__ex_flag;
-	if (unlikely(__res != 0))
+	if (unlikely(!__done || __res != 0))
 		__res = fail_fn(count);
 
 	return __res;
@@ -61,19 +59,18 @@ __mutex_fastpath_lock_retval(atomic_t *count, int (*fail_fn)(atomic_t *))
 static inline void
 __mutex_fastpath_unlock(atomic_t *count, void (*fail_fn)(atomic_t *))
 {
-	int __ex_flag, __res;
+	int __done, __res;
 
 	__asm__ __volatile__ (
 		"movli.l	@%2, %0	\n\t"
 		"add		#1, %0	\n\t"
 		"movco.l	%0, @%2 \n\t"
 		"movt		%1	\n\t"
-		: "=&z" (__res), "=&r" (__ex_flag)
+		: "=&z" (__res), "=&r" (__done)
 		: "r" (&(count)->counter)
 		: "t");
 
-	__res |= !__ex_flag;
-	if (unlikely(__res <= 0))
+	if (unlikely(!__done || __res <= 0))
 		fail_fn(count);
 }
 
-- 1.6.0.6 

  parent reply	other threads:[~2009-01-28  9:29 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-01-23  9:04 [bug report] dead lock (?) occur on ap325 board morimoto.kuninori
2009-01-26  2:20 ` Paul Mundt
2009-01-26  5:01 ` morimoto.kuninori
2009-01-27  8:26 ` morimoto.kuninori
2009-01-27  8:32 ` Paul Mundt
2009-01-27  8:46 ` morimoto.kuninori
2009-01-27  8:49 ` Paul Mundt
2009-01-28  1:25 ` morimoto.kuninori
2009-01-28  9:29 ` Takashi Yoshii [this message]
2009-01-28 10:03 ` Paul Mundt
2009-01-29  0:34 ` yoshii.takashi
2009-01-29 23:54 ` morimoto.kuninori

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=49802569.4000808@renesas.com \
    --to=yoshii.takashi@renesas.com \
    --cc=linux-sh@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