linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Gleixner <tglx@linutronix.de>
To: Greg KH <greg@kroah.com>
Cc: Chuck Ebbert <cebbert@redhat.com>, Ingo Molnar <mingo@elte.hu>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Andi Kleen <andi@firstfloor.org>,
	Gerd Hoffmann <kraxel@redhat.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Stable Team <stable@kernel.org>
Subject: Re: [stable] The SMP alternatives code breaks exception fixup?
Date: Thu, 7 Feb 2008 13:16:04 +0100 (CET)	[thread overview]
Message-ID: <alpine.LFD.1.00.0802071254590.9204@apollo.tec.linutronix.de> (raw)
In-Reply-To: <20080206231020.GB12395@kroah.com>

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1017 bytes --]

On Wed, 6 Feb 2008, Greg KH wrote:

> On Tue, Jan 22, 2008 at 03:19:33PM -0500, Chuck Ebbert wrote:
> > On 01/22/2008 07:35 AM, Thomas Gleixner wrote:
> > > 
> > > That's a long standing bug in both the PI futex and the standard futex
> > > code. Needs to go to stable as well.
> > > 
> > 
> > Here's the 2.6.23 version:
> > 
> > 
> > Subject: x86: fix missing exception entry for SMP alternatives in futex macros
> > From: Thomas Gleixner <tglx@linutronix.de>
> 
> I don't see this in Linus's tree, am I just missing it?  Do you have a
> git commit id?

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=9d55b9923a1b7ea8193b8875c57ec940dc2ff027
 
I did a simpler version by just replacing LOCK_PREFIX with lock. The
reason is, that we would need a separate implementation for
__futex_atomic_op1() as well, due to:

case FUTEX_OP_ADD:
	__futex_atomic_op1(LOCK_PREFIX "xaddl %0, %2", ret, oldval,

And it's not really worth the trouble.

Backport to 2.6.24 and 2.6.23 attached.

Thanks,
	tglx

[-- Attachment #2: Type: TEXT/PLAIN, Size: 2973 bytes --]

Subject: x86: replace LOCK_PREFIX in futex.h
From: Thomas Gleixner <tglx@linutronix.de>
Date: Thu, 07 Feb 2008 13:05:19 +0100

The exception fixup for the futex macros __futex_atomic_op1/2 and
futex_atomic_cmpxchg_inatomic() is missing an entry when the lock
prefix is replaced by a NOP via SMP alternatives.

Chuck Ebert tracked this down from the information provided in:
https://bugzilla.redhat.com/show_bug.cgi?id=429412

A possible solution would be to add another fixup after the
LOCK_PREFIX, so both the LOCK and NOP case have their own entry in the
exception table, but it's not really worth the trouble.

Simply replace LOCK_PREFIX with lock and keep those untouched by SMP
alternatives.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
                                                                
Index: linux-2.6.23/include/asm-i386/futex.h
===================================================================
--- linux-2.6.23.orig/include/asm-i386/futex.h	2008-02-07 13:08:03.000000000 +0100
+++ linux-2.6.23/include/asm-i386/futex.h	2008-02-07 13:09:39.000000000 +0100
@@ -28,7 +28,7 @@
 "1:	movl	%2, %0\n\
 	movl	%0, %3\n"					\
 	insn "\n"						\
-"2:	" LOCK_PREFIX "cmpxchgl %3, %2\n\
+"2:	lock cmpxchgl %3, %2\n\
 	jnz	1b\n\
 3:	.section .fixup,\"ax\"\n\
 4:	mov	%5, %1\n\
@@ -68,7 +68,7 @@ futex_atomic_op_inuser (int encoded_op, 
 #endif
 		switch (op) {
 		case FUTEX_OP_ADD:
-			__futex_atomic_op1(LOCK_PREFIX "xaddl %0, %2", ret,
+			__futex_atomic_op1("lock xaddl %0, %2", ret,
 					   oldval, uaddr, oparg);
 			break;
 		case FUTEX_OP_OR:
@@ -111,7 +111,7 @@ futex_atomic_cmpxchg_inatomic(int __user
 		return -EFAULT;
 
 	__asm__ __volatile__(
-		"1:	" LOCK_PREFIX "cmpxchgl %3, %1		\n"
+		"1:	lock cmpxchgl %3, %1			\n"
 
 		"2:	.section .fixup, \"ax\"			\n"
 		"3:	mov     %2, %0				\n"
Index: linux-2.6.23/include/asm-x86_64/futex.h
===================================================================
--- linux-2.6.23.orig/include/asm-x86_64/futex.h	2008-02-07 13:08:03.000000000 +0100
+++ linux-2.6.23/include/asm-x86_64/futex.h	2008-02-07 13:09:39.000000000 +0100
@@ -27,7 +27,7 @@
 "1:	movl	%2, %0\n\
 	movl	%0, %3\n"					\
 	insn "\n"						\
-"2:	" LOCK_PREFIX "cmpxchgl %3, %2\n\
+"2:	"lock cmpxchgl %3, %2\n\
 	jnz	1b\n\
 3:	.section .fixup,\"ax\"\n\
 4:	mov	%5, %1\n\
@@ -62,7 +62,7 @@ futex_atomic_op_inuser (int encoded_op, 
 		__futex_atomic_op1("xchgl %0, %2", ret, oldval, uaddr, oparg);
 		break;
 	case FUTEX_OP_ADD:
-		__futex_atomic_op1(LOCK_PREFIX "xaddl %0, %2", ret, oldval,
+		__futex_atomic_op1("lock xaddl %0, %2", ret, oldval,
 				   uaddr, oparg);
 		break;
 	case FUTEX_OP_OR:
@@ -101,7 +101,7 @@ futex_atomic_cmpxchg_inatomic(int __user
 		return -EFAULT;
 
 	__asm__ __volatile__(
-		"1:	" LOCK_PREFIX "cmpxchgl %3, %1		\n"
+		"1:	lock cmpxchgl %3, %1			\n"
 
 		"2:	.section .fixup, \"ax\"			\n"
 		"3:	mov     %2, %0				\n"

[-- Attachment #3: Type: TEXT/PLAIN, Size: 2913 bytes --]

Subject: x86: replace LOCK_PREFIX in futex.h
From: Thomas Gleixner <tglx@linutronix.de>
Date: Thu, 07 Feb 2008 13:06:02 +0100

The exception fixup for the futex macros __futex_atomic_op1/2 and
futex_atomic_cmpxchg_inatomic() is missing an entry when the lock
prefix is replaced by a NOP via SMP alternatives.

Chuck Ebert tracked this down from the information provided in:
https://bugzilla.redhat.com/show_bug.cgi?id=429412

A possible solution would be to add another fixup after the
LOCK_PREFIX, so both the LOCK and NOP case have their own entry in the
exception table, but it's not really worth the trouble.

Simply replace LOCK_PREFIX with lock and keep those untouched by SMP
alternatives.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Index: linux-2.6.24/include/asm-x86/futex_32.h
===================================================================
--- linux-2.6.24.orig/include/asm-x86/futex_32.h	2008-02-07 13:05:39.000000000 +0100
+++ linux-2.6.24/include/asm-x86/futex_32.h	2008-02-07 13:07:20.000000000 +0100
@@ -28,7 +28,7 @@
 "1:	movl	%2, %0\n\
 	movl	%0, %3\n"					\
 	insn "\n"						\
-"2:	" LOCK_PREFIX "cmpxchgl %3, %2\n\
+"2:	lock cmpxchgl %3, %2\n\
 	jnz	1b\n\
 3:	.section .fixup,\"ax\"\n\
 4:	mov	%5, %1\n\
@@ -68,7 +68,7 @@ futex_atomic_op_inuser (int encoded_op, 
 #endif
 		switch (op) {
 		case FUTEX_OP_ADD:
-			__futex_atomic_op1(LOCK_PREFIX "xaddl %0, %2", ret,
+			__futex_atomic_op1("lock xaddl %0, %2", ret,
 					   oldval, uaddr, oparg);
 			break;
 		case FUTEX_OP_OR:
@@ -111,7 +111,7 @@ futex_atomic_cmpxchg_inatomic(int __user
 		return -EFAULT;
 
 	__asm__ __volatile__(
-		"1:	" LOCK_PREFIX "cmpxchgl %3, %1		\n"
+		"1:	lock cmpxchgl %3, %1			\n"
 
 		"2:	.section .fixup, \"ax\"			\n"
 		"3:	mov     %2, %0				\n"
Index: linux-2.6.24/include/asm-x86/futex_64.h
===================================================================
--- linux-2.6.24.orig/include/asm-x86/futex_64.h	2008-02-07 13:05:39.000000000 +0100
+++ linux-2.6.24/include/asm-x86/futex_64.h	2008-02-07 13:06:45.000000000 +0100
@@ -27,7 +27,7 @@
 "1:	movl	%2, %0\n\
 	movl	%0, %3\n"					\
 	insn "\n"						\
-"2:	" LOCK_PREFIX "cmpxchgl %3, %2\n\
+"2:	"lock cmpxchgl %3, %2\n\
 	jnz	1b\n\
 3:	.section .fixup,\"ax\"\n\
 4:	mov	%5, %1\n\
@@ -62,7 +62,7 @@ futex_atomic_op_inuser (int encoded_op, 
 		__futex_atomic_op1("xchgl %0, %2", ret, oldval, uaddr, oparg);
 		break;
 	case FUTEX_OP_ADD:
-		__futex_atomic_op1(LOCK_PREFIX "xaddl %0, %2", ret, oldval,
+		__futex_atomic_op1("lock xaddl %0, %2", ret, oldval,
 				   uaddr, oparg);
 		break;
 	case FUTEX_OP_OR:
@@ -101,7 +101,7 @@ futex_atomic_cmpxchg_inatomic(int __user
 		return -EFAULT;
 
 	__asm__ __volatile__(
-		"1:	" LOCK_PREFIX "cmpxchgl %3, %1		\n"
+		"1:	lock cmpxchgl %3, %1			\n"
 
 		"2:	.section .fixup, \"ax\"			\n"
 		"3:	mov     %2, %0				\n"

      parent reply	other threads:[~2008-02-07 12:17 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-01-21 20:47 The SMP alternatives code breaks exception fixup? Chuck Ebbert
2008-01-21 21:25 ` Chuck Ebbert
2008-01-22  5:26 ` Andi Kleen
2008-01-22 12:35   ` Thomas Gleixner
2008-01-22 20:19     ` Chuck Ebbert
2008-02-06 23:10       ` [stable] " Greg KH
2008-02-06 23:36         ` Linus Torvalds
2008-02-06 23:43           ` Ingo Molnar
2008-02-07  0:00             ` Greg KH
2008-02-07  6:55             ` Greg KH
2008-02-07 12:16         ` Thomas Gleixner [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=alpine.LFD.1.00.0802071254590.9204@apollo.tec.linutronix.de \
    --to=tglx@linutronix.de \
    --cc=akpm@linux-foundation.org \
    --cc=andi@firstfloor.org \
    --cc=cebbert@redhat.com \
    --cc=greg@kroah.com \
    --cc=kraxel@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=stable@kernel.org \
    --cc=torvalds@linux-foundation.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).