[PATCH v2] powerpc/32: fix csum_partial_copy_generic()

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

 arch/powerpc/lib/checksum_32.S | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/arch/powerpc/lib/checksum_32.S b/arch/powerpc/lib/checksum_32.S
index d90870a..0a57fe6 100644
--- a/arch/powerpc/lib/checksum_32.S
+++ b/arch/powerpc/lib/checksum_32.S

@@ -127,8 +127,9 @@ _GLOBAL(csum_partial_copy_generic)
 	stw	r7,12(r1)
 	stw	r8,8(r1)
 
-	andi.	r0,r4,1			/* is destination address even ? */
-	cmplwi	cr7,r0,0
+	rlwinm	r0,r4,3,0x8
+	rlwnm	r6,r6,r0,0,31	/* odd destination address: rotate one byte */
+	cmplwi	cr7,r0,0	/* is destination address even ? */
 	addic	r12,r6,0
 	addi	r6,r4,-4
 	neg	r0,r4

@@ -237,7 +238,7 @@ _GLOBAL(csum_partial_copy_generic)
 66:	addze	r3,r12
 	addi	r1,r1,16
 	beqlr+	cr7
-	rlwinm	r3,r3,8,0,31	/* swap bytes for odd destination */
+	rlwinm	r3,r3,8,0,31	/* odd destination address: rotate one byte */
 	blr
 
 /* read fault */

-- 
2.1.0

On Tue, Aug 02, 2016 at 10:07:05AM +0200, Christophe Leroy wrote:

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

Thanks.  It looks good to me now.


Segher

On Tue, Aug 02, 2016 at 10:07:05AM +0200, Christophe Leroy wrote:

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

This looks good to me now, thanks,


Segher

On Tue, 2016-08-02 at 10:07 +0200, Christophe Leroy wrote:

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

 arch/powerpc/lib/checksum_32.S | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Alessio, can you confirm whether this fixes the problem you reported?

-Scott

Scott,

On 4 August 2016 at 05:53, Scott Wood [off-list ref] wrote:

On Tue, 2016-08-02 at 10:07 +0200, Christophe Leroy wrote:

quoted

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

 arch/powerpc/lib/checksum_32.S | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Alessio, can you confirm whether this fixes the problem you reported?

No unfortunately.

Ciao,
Alessio

Alessio Igor Bogani [off-list ref] writes:

On 4 August 2016 at 05:53, Scott Wood [off-list ref] wrote:

quoted

On Tue, 2016-08-02 at 10:07 +0200, Christophe Leroy wrote:

quoted

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
 v2: updated comments as suggested by Segher

 arch/powerpc/lib/checksum_32.S | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

Alessio, can you confirm whether this fixes the problem you reported?

No unfortunately.

Thanks for testing.

I've dropped the patch for now, send me a new one that works.

cheers

Le 05/08/2016 à 08:57, Michael Ellerman a écrit :

Alessio Igor Bogani [off-list ref] writes:

quoted

On 4 August 2016 at 05:53, Scott Wood [off-list ref] wrote:

quoted

On Tue, 2016-08-02 at 10:07 +0200, Christophe Leroy wrote:

quoted

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()")
Cc: stable@vger.kernel.org

Signed-off-by: Christophe Leroy <redacted>
---
  v2: updated comments as suggested by Segher

  arch/powerpc/lib/checksum_32.S | 7 ++++---
  1 file changed, 4 insertions(+), 3 deletions(-)

Alessio, can you confirm whether this fixes the problem you reported?

No unfortunately.

Thanks for testing.

I've dropped the patch for now, send me a new one that works.

The purpose of this patch was not to address Alessio's issue, but to fix 
a huge issue on checksum calculation which induces breakdown of TCP 
connections.

I think it is worth commiting it upstream and on impacted stable 
releases, allthought we don't have yet identified the issue Alessio's has.

Christophe

Christophe Leroy [off-list ref] writes:

Le 05/08/2016 =C3=A0 08:57, Michael Ellerman a =C3=A9crit :

quoted

Alessio Igor Bogani [off-list ref] writes:

quoted

On 4 August 2016 at 05:53, Scott Wood [off-list ref] wrote:

quoted

On Tue, 2016-08-02 at 10:07 +0200, Christophe Leroy wrote:

quoted

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

The purpose of this patch was not to address Alessio's issue, but to fix=

=20

a huge issue on checksum calculation which induces breakdown of TCP=20
connections.

I think it is worth commiting it upstream and on impacted stable=20
releases, allthought we don't have yet identified the issue Alessio's has.

OK. I'll put it back into fixes on Monday if I haven't heard anything
further.

cheers

On Tue, 2016-02-08 at 08:07:05 UTC, Christophe Leroy wrote:

commit 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic()
based on copy_tofrom_user()") introduced a bug when destination
address is odd and initial csum is not null

In that (rare) case the initial csum value has to be rotated one byte
as well as the resulting value is

This patch also fixes related comments

Fixes: 7aef4136566b0 ("powerpc32: rewrite csum_partial_copy_generic() based on copy_tofrom_user()")
Signed-off-by: Christophe Leroy <redacted>

Applied to powerpc fixes, thanks.

https://git.kernel.org/powerpc/c/1bc8b816cb8058c31f61fe7844

cheers