[PATCH 0/4] backports: fixes for wireless-testing

[PATCH 0/4] backports: fixes for wireless-testing

[Arend van Spriel]

This series fixes backport issues found backporting the wireless-testing
tree that moved to 4.11-rc1.

Arend van Spriel (4):
  patches: update crypto-ccm.patch for v4.11-rc1
  backports: add backport for idr_remove
  backports: introduce new definition for DECLARE_EWMA macro
  backports: add dummy include/linux/sched/signal.h

 backport/backport-include/linux/average.h      | 65 +++++++++++++++++---------
 backport/backport-include/linux/idr.h          | 10 ++++
 backport/backport-include/linux/sched/signal.h | 10 ++++
 patches/crypto-ccm.patch                       |  9 ++--
 4 files changed, 69 insertions(+), 25 deletions(-)
 create mode 100644 backport/backport-include/linux/sched/signal.h

--
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

[PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Arend van Spriel]

Since v4.11-rc1 the patch file for crypto/ccm.c, which is incorporated
in the compat module, no longer applies. Update the patch file to fix
the issue.

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
 patches/crypto-ccm.patch | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/patches/crypto-ccm.patch b/patches/crypto-ccm.patch
index 8bdea3d..7a44d37 100644
--- a/patches/crypto-ccm.patch
+++ b/patches/crypto-ccm.patch
@@ -1,7 +1,7 @@
 --- a/compat/crypto-ccm.c
 +++ b/compat/crypto-ccm.c
-@@ -13,13 +13,44 @@
- #include <crypto/internal/aead.h>
+@@ -14,13 +14,44 @@
+ #include <crypto/internal/hash.h>
  #include <crypto/internal/skcipher.h>
  #include <crypto/scatterwalk.h>
 +#include <crypto/algapi.h>
@@ -46,7 +46,7 @@
  
  struct ccm_instance_ctx {
  	struct crypto_skcipher_spawn ctr;
-@@ -897,7 +928,7 @@ static struct crypto_template crypto_rfc
+@@ -1001,7 +1032,7 @@ static int cbcmac_create(struct crypto_template *tmpl, struct rtattr **tb)
  	.module = THIS_MODULE,
  };
  
@@ -55,7 +55,7 @@
  {
  	int err;
  
-@@ -923,18 +954,9 @@ out_undo_base:
+@@ -1033,19 +1064,10 @@ static int __init crypto_ccm_module_init(void)
  	goto out;
  }
  
@@ -65,6 +65,7 @@
  	crypto_unregister_template(&crypto_rfc4309_tmpl);
  	crypto_unregister_template(&crypto_ccm_tmpl);
  	crypto_unregister_template(&crypto_ccm_base_tmpl);
+ 	crypto_unregister_template(&crypto_cbcmac_tmpl);
  }
 -
 -module_init(crypto_ccm_module_init);
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Johannes Berg]

On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
> Since v4.11-rc1 the patch file for crypto/ccm.c, which is
> incorporated
> in the compat module, no longer applies. Update the patch file to fix
> the issue.

I think I want to get rid of this ccm thing entirely, any objections?
As I remember, it doesn't actually work properly anyway.

I suspect that if we really can't do anything else, we'd have to
provide a backport for all of crypto/, but I don't want to go there and
would rather mandate that all the crypto things are enabled in the base
kernel.

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Arend Van Spriel]

On 10-3-2017 7:34, Johannes Berg wrote:
> On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
>> Since v4.11-rc1 the patch file for crypto/ccm.c, which is
>> incorporated
>> in the compat module, no longer applies. Update the patch file to fix
>> the issue.
> 
> I think I want to get rid of this ccm thing entirely, any objections?
> As I remember, it doesn't actually work properly anyway.

No objection here. Does the same apply to skcipher stuff?

> I suspect that if we really can't do anything else, we'd have to
> provide a backport for all of crypto/, but I don't want to go there and
> would rather mandate that all the crypto things are enabled in the base
> kernel.

But that can only work when crypto api in base kernel and backport
kernel are the same, right?

Regards,
Arend
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Johannes Berg]

On Fri, 2017-03-10 at 10:49 +0100, Arend Van Spriel wrote:
> 
> On 10-3-2017 7:34, Johannes Berg wrote:
> > On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
> > > Since v4.11-rc1 the patch file for crypto/ccm.c, which is
> > > incorporated
> > > in the compat module, no longer applies. Update the patch file to
> > > fix
> > > the issue.
> > 
> > I think I want to get rid of this ccm thing entirely, any
> > objections?
> > As I remember, it doesn't actually work properly anyway.
> 
> No objection here. Does the same apply to skcipher stuff?

I think so, but not sure.

> > I suspect that if we really can't do anything else, we'd have to
> > provide a backport for all of crypto/, but I don't want to go there
> > and
> > would rather mandate that all the crypto things are enabled in the
> > base
> > kernel.
> 
> But that can only work when crypto api in base kernel and backport
> kernel are the same, right?

No, I meant adding all of crypto/ to copy-list and making sure *that*
works across kernels ...

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Arend Van Spriel]

On 10-3-2017 10:52, Johannes Berg wrote:
> On Fri, 2017-03-10 at 10:49 +0100, Arend Van Spriel wrote:
>>
>> On 10-3-2017 7:34, Johannes Berg wrote:
>>> On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
>>>> Since v4.11-rc1 the patch file for crypto/ccm.c, which is
>>>> incorporated
>>>> in the compat module, no longer applies. Update the patch file to
>>>> fix
>>>> the issue.
>>>
>>> I think I want to get rid of this ccm thing entirely, any
>>> objections?
>>> As I remember, it doesn't actually work properly anyway.
>>
>> No objection here. Does the same apply to skcipher stuff?
> 
> I think so, but not sure.
> 
>>> I suspect that if we really can't do anything else, we'd have to
>>> provide a backport for all of crypto/, but I don't want to go there
>>> and
>>> would rather mandate that all the crypto things are enabled in the
>>> base
>>> kernel.
>>
>> But that can only work when crypto api in base kernel and backport
>> kernel are the same, right?
> 
> No, I meant adding all of crypto/ to copy-list and making sure *that*
> works across kernels ...

Understand. There seem to have been significant changes in crypto land
making this a challenge. Still seems like right idea.

Regards,
Arend
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 1/4] patches: update crypto-ccm.patch for v4.11-rc1

[Johannes Berg]

> > No, I meant adding all of crypto/ to copy-list and making sure
> > *that*
> > works across kernels ...
> 
> Understand. There seem to have been significant changes in crypto
> land making this a challenge. Still seems like right idea.

I'm not really sure it's the right idea, it seems like it'll be quite
complicated ... :)

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

[PATCH 2/4] backports: add backport for idr_remove

[Arend van Spriel]

Since commit d3e709e63e97 ("idr: Return the deleted entry from
idr_remove") the function signature changed so provide a backport
which does a idr_find() and subsequent idr_remove().

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
Hi Johannes,

Not sure if the find/remove approach is a good one here. Better solutions
are welcome.

Regards,
Arend
---
 backport/backport-include/linux/idr.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/backport/backport-include/linux/idr.h b/backport/backport-include/linux/idr.h
index d7cf9f0..01aa2c2 100644
--- a/backport/backport-include/linux/idr.h
+++ b/backport/backport-include/linux/idr.h
@@ -61,4 +61,14 @@ static inline void idr_preload_end(void)
 	for (id = 0; ((entry) = idr_get_next(idp, &(id))) != NULL; ++id)
 #endif

+#if LINUX_VERSION_IS_LESS(4, 11, 0)
+static inline void *backport_idr_remove(struct idr *idr, int id)
+{
+	void *item = idr_find(idr, id);
+	idr_remove(idr, id);
+	return item;
+}
+#define idr_remove	backport_idr_remove
+#endif
+
 #endif /* __BACKPORT_IDR_H */
--
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

[PATCH 3/4] backports: introduce new definition for DECLARE_EWMA macro

[Arend van Spriel]

Since commit eb1e011a1474 ("average: change to declare precision,
not factor") the definition of DECLARE_EWMA has changed so we
need a backport for the new macro as it differs. Changing the
strategy by checking the kernel version and #undef the macro to
be sure we do not run into macro redefinition issue.

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
 backport/backport-include/linux/average.h | 65 +++++++++++++++++++++----------
 1 file changed, 44 insertions(+), 21 deletions(-)

diff --git a/backport/backport-include/linux/average.h b/backport/backport-include/linux/average.h
index 5d80041..51702fb 100644
--- a/backport/backport-include/linux/average.h
+++ b/backport/backport-include/linux/average.h
@@ -2,45 +2,68 @@
 #define __BACKPORT_AVERAGE
 #include_next <linux/average.h>
 
-#ifndef DECLARE_EWMA
-#define DECLARE_EWMA(name, _factor, _weight)				\
+/*
+ * Exponentially weighted moving average (EWMA)
+ *
+ * This implements a fixed-precision EWMA algorithm, with both the
+ * precision and fall-off coefficient determined at compile-time
+ * and built into the generated helper funtions.
+ *
+ * The first argument to the macro is the name that will be used
+ * for the struct and helper functions.
+ *
+ * The second argument, the precision, expresses how many bits are
+ * used for the fractional part of the fixed-precision values.
+ *
+ * The third argument, the weight reciprocal, determines how the
+ * new values will be weighed vs. the old state, new values will
+ * get weight 1/weight_rcp and old values 1-1/weight_rcp. Note
+ * that this parameter must be a power of two for efficiency.
+ */
+#if LINUX_VERSION_IS_LESS(4, 11, 0)
+#undef DECLARE_EWMA
+#define DECLARE_EWMA(name, _precision, _weight_rcp)			\
 	struct ewma_##name {						\
 		unsigned long internal;					\
 	};								\
 	static inline void ewma_##name##_init(struct ewma_##name *e)	\
 	{								\
-		BUILD_BUG_ON(!__builtin_constant_p(_factor));		\
-		BUILD_BUG_ON(!__builtin_constant_p(_weight));		\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_factor);			\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_weight);			\
+		BUILD_BUG_ON(!__builtin_constant_p(_precision));	\
+		BUILD_BUG_ON(!__builtin_constant_p(_weight_rcp));	\
+		/*							\
+		 * Even if you want to feed it just 0/1 you should have	\
+		 * some bits for the non-fractional part...		\
+		 */							\
+		BUILD_BUG_ON((_precision) > 30);			\
+		BUILD_BUG_ON_NOT_POWER_OF_2(_weight_rcp);		\
 		e->internal = 0;					\
 	}								\
 	static inline unsigned long					\
 	ewma_##name##_read(struct ewma_##name *e)			\
 	{								\
-		BUILD_BUG_ON(!__builtin_constant_p(_factor));		\
-		BUILD_BUG_ON(!__builtin_constant_p(_weight));		\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_factor);			\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_weight);			\
-		return e->internal >> ilog2(_factor);			\
+		BUILD_BUG_ON(!__builtin_constant_p(_precision));	\
+		BUILD_BUG_ON(!__builtin_constant_p(_weight_rcp));	\
+		BUILD_BUG_ON((_precision) > 30);			\
+		BUILD_BUG_ON_NOT_POWER_OF_2(_weight_rcp);		\
+		return e->internal >> (_precision);			\
 	}								\
 	static inline void ewma_##name##_add(struct ewma_##name *e,	\
 					     unsigned long val)		\
 	{								\
 		unsigned long internal = ACCESS_ONCE(e->internal);	\
-		unsigned long weight = ilog2(_weight);			\
-		unsigned long factor = ilog2(_factor);			\
+		unsigned long weight_rcp = ilog2(_weight_rcp);		\
+		unsigned long precision = _precision;			\
 									\
-		BUILD_BUG_ON(!__builtin_constant_p(_factor));		\
-		BUILD_BUG_ON(!__builtin_constant_p(_weight));		\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_factor);			\
-		BUILD_BUG_ON_NOT_POWER_OF_2(_weight);			\
+		BUILD_BUG_ON(!__builtin_constant_p(_precision));	\
+		BUILD_BUG_ON(!__builtin_constant_p(_weight_rcp));	\
+		BUILD_BUG_ON((_precision) > 30);			\
+		BUILD_BUG_ON_NOT_POWER_OF_2(_weight_rcp);		\
 									\
 		ACCESS_ONCE(e->internal) = internal ?			\
-			(((internal << weight) - internal) +		\
-				(val << factor)) >> weight :		\
-			(val << factor);				\
+			(((internal << weight_rcp) - internal) +	\
+				(val << precision)) >> weight_rcp :	\
+			(val << precision);				\
 	}
-#endif /* DECLARE_EWMA */
+#endif /* LINUX_VERSION_IS_LESS(4, 11, 0) */
 
 #endif /* __BACKPORT_AVERAGE */
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 3/4] backports: introduce new definition for DECLARE_EWMA macro

[Johannes Berg]

On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
> Since commit eb1e011a1474 ("average: change to declare precision,
> not factor") the definition of DECLARE_EWMA has changed so we
> need a backport for the new macro as it differs. Changing the
> strategy by checking the kernel version and #undef the macro to
> be sure we do not run into macro redefinition issue.

Good one. However, why not just add this file to copy-list and be done
with it? That way, the original one from the kernel will never be
included at all and we'll always have the right version. Not that I
think we'll change it much, but why not make it easier?

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 3/4] backports: introduce new definition for DECLARE_EWMA macro

[Arend Van Spriel]

On 10-3-2017 7:36, Johannes Berg wrote:
> On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
>> Since commit eb1e011a1474 ("average: change to declare precision,
>> not factor") the definition of DECLARE_EWMA has changed so we
>> need a backport for the new macro as it differs. Changing the
>> strategy by checking the kernel version and #undef the macro to
>> be sure we do not run into macro redefinition issue.
> 
> Good one. However, why not just add this file to copy-list and be done
> with it? That way, the original one from the kernel will never be
> included at all and we'll always have the right version. Not that I
> think we'll change it much, but why not make it easier?

That will work as long a linux/average.h only contains DECLARE_EWMA and
that is all the backported code cares about. So for now the copy-list
approach will work. Maybe add a comment in the copy-list documenting the
approach.

Regards,
Arend
--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 3/4] backports: introduce new definition for DECLARE_EWMA macro

[Johannes Berg]

> That will work as long a linux/average.h only contains DECLARE_EWMA
> and that is all the backported code cares about. So for now the copy-
> list approach will work. Maybe add a comment in the copy-list
> documenting the approach.
> 
I think it's highly unlikely it'll ever contain anything else.

Will you respin this, or did you want me to handle it?

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

[PATCH 4/4] backports: add dummy include/linux/sched/signal.h

[Arend van Spriel]

Simply including linux/sched.h for older kernels not carrying this
file. This new include file was introduced by commit 3f07c0144132
("sched/headers: Prepare for new header dependencies before moving
code to <linux/sched/signal.h>").

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
 backport/backport-include/linux/sched/signal.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
 create mode 100644 backport/backport-include/linux/sched/signal.h

diff --git a/backport/backport-include/linux/sched/signal.h b/backport/backport-include/linux/sched/signal.h
new file mode 100644
index 0000000..ec89925
--- /dev/null
+++ b/backport/backport-include/linux/sched/signal.h
@@ -0,0 +1,10 @@
+#ifndef _BACKPORT_LINUX_SCHED_SIGNAL_H
+#define _BACKPORT_LINUX_SCHED_SIGNAL_H
+
+#if LINUX_VERSION_IS_LESS(4, 11, 0)
+#include <linux/sched.h>
+#else
+#include_next <linux/sched/signal.h>
+#endif
+
+#endif /* _BACKPORT_LINUX_SCHED_SIGNAL_H */
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

Re: [PATCH 0/4] backports: fixes for wireless-testing

[Johannes Berg]

On Thu, 2017-03-09 at 12:57 +0000, Arend van Spriel wrote:
> This series fixes backport issues found backporting the wireless-
> testing
> tree that moved to 4.11-rc1.
> 
> Arend van Spriel (4):
>   patches: update crypto-ccm.patch for v4.11-rc1
>   backports: add backport for idr_remove
>   backports: introduce new definition for DECLARE_EWMA macro
>   backports: add dummy include/linux/sched/signal.h

Applied, except I replaced the average.h by just copying that file

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in