[Buildroot] [PATCH 1/2] linux-headers: bump 3.14.x and 4.{4, 5}.x series

[Buildroot] [PATCH 1/2] linux-headers: bump 3.14.x and 4.{4, 5}.x series

[Gustavo Zacarias]

Kernel headers 4.5 -> 4.5.4 broke netfilter uapi headers leading to
build failures for cases where glibc's net/if.h was included before
linux/if.h
This is fixed since 4.5.5 via linux kernel commit 4a91cb61 and fixes:
http://autobuild.buildroot.net/results/736/7362431f62ebb1b436ac41ea3ef85228c763e6f3/

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
---
 package/linux-headers/Config.in.host | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/package/linux-headers/Config.in.host b/package/linux-headers/Config.in.host
index 7c080d6..ac89a6b 100644
--- a/package/linux-headers/Config.in.host
+++ b/package/linux-headers/Config.in.host
@@ -214,13 +214,13 @@ config BR2_DEFAULT_KERNEL_HEADERS
 	default "3.4.112"	if BR2_KERNEL_HEADERS_3_4
 	default "3.10.101"	if BR2_KERNEL_HEADERS_3_10
 	default "3.12.59"	if BR2_KERNEL_HEADERS_3_12
-	default "3.14.69"	if BR2_KERNEL_HEADERS_3_14
+	default "3.14.70"	if BR2_KERNEL_HEADERS_3_14
 	default "3.18.33"	if BR2_KERNEL_HEADERS_3_18
 	default "3.19.8"	if BR2_KERNEL_HEADERS_3_19
 	default "4.0.9"		if BR2_KERNEL_HEADERS_4_0
 	default "4.1.24"	if BR2_KERNEL_HEADERS_4_1
 	default "4.2.8"		if BR2_KERNEL_HEADERS_4_2
 	default "4.3.6"		if BR2_KERNEL_HEADERS_4_3
-	default "4.4.10"	if BR2_KERNEL_HEADERS_4_4
-	default "4.5.4"		if BR2_KERNEL_HEADERS_4_5
+	default "4.4.11"	if BR2_KERNEL_HEADERS_4_4
+	default "4.5.5"		if BR2_KERNEL_HEADERS_4_5
 	default BR2_DEFAULT_KERNEL_VERSION if BR2_KERNEL_HEADERS_VERSION
-- 
2.7.3

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Gustavo Zacarias]

This was added in 0dece985 to deal with kernel headers 4.5 -> 4.5.4
breakage. Unfortunately the fix in 4.5.5 and 4.6+ doesn't deal correctly
with this causing more build breakage, so we'd rather "break" for the
small range of 4.5.x versions broken than for the rest of time.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
---
 package/connman/0001-xtables.patch | 61 --------------------------------------
 1 file changed, 61 deletions(-)
 delete mode 100644 package/connman/0001-xtables.patch

diff --git a/package/connman/0001-xtables.patch b/package/connman/0001-xtables.patch
deleted file mode 100644
index 277fce6..0000000
--- a/package/connman/0001-xtables.patch
+++ /dev/null
@@ -1,61 +0,0 @@
-Use some preprocessor tricks to route around incompatibilies seen
-between newer Linux kernels and glibc when both net/if.h and linux/if.h
-are included in the same source.
-
-Patch from gentoo, see https://bugs.gentoo.org/show_bug.cgi?id=577584
-
-Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
-
---- connman-1.31/src/iptables.c.old	2016-03-22 20:12:47.829460752 -0700
-+++ connman-1.31/src/iptables.c	2016-03-22 21:33:36.835384724 -0700
-@@ -28,11 +28,11 @@
- #include <stdio.h>
- #include <string.h>
- #include <unistd.h>
- #include <sys/errno.h>
- #include <sys/socket.h>
--#include <xtables.h>
-+#include "connman_xtables.h"
- #include <inttypes.h>
- 
- #include <linux/netfilter_ipv4/ip_tables.h>
- 
- #include "connman.h"
---- connman-1.31/src/firewall.c.old	2016-03-22 21:29:01.959472262 -0700
-+++ connman-1.31/src/firewall.c	2016-03-22 21:33:53.048144181 -0700
-@@ -23,11 +23,11 @@
- #include <config.h>
- #endif
- 
- #include <errno.h>
- 
--#include <xtables.h>
-+#include "connman_xtables.h"
- #include <linux/netfilter_ipv4/ip_tables.h>
- 
- #include "connman.h"
- 
- #define CHAIN_PREFIX "connman-"
---- /dev/null	2016-03-18 06:21:16.372989086 -0700
-+++ connman-1.31/include/connman_xtables.h	2016-03-22 21:32:21.349504786 -0700
-@@ -0,0 +1,20 @@
-+#ifndef CONNMAN_XTABLES_H
-+#define CONNMAN_XTABLES_H
-+
-+#include <linux/version.h>
-+#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 5, 0)
-+#include <xtables.h>
-+#else
-+#ifdef __USE_MISC
-+#define GENTOO_USE_MISC __USE_MISC
-+#undef __USE_MISC
-+#endif
-+
-+#include <xtables.h>
-+
-+#ifdef GENTOO_USE_MISC
-+#define __USE_MISC GENTOO_USE_MISC
-+#undef GENTOO_USE_MISC
-+#endif
-+#endif
-+#endif
-- 
2.7.3

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Thomas Petazzoni]

Hello,

On Thu, 19 May 2016 10:25:37 -0300, Gustavo Zacarias wrote:
> This was added in 0dece985 to deal with kernel headers 4.5 -> 4.5.4
> breakage. Unfortunately the fix in 4.5.5 and 4.6+ doesn't deal correctly
> with this causing more build breakage, so we'd rather "break" for the
> small range of 4.5.x versions broken than for the rest of time.
> 
> Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
> ---
>  package/connman/0001-xtables.patch | 61 --------------------------------------
>  1 file changed, 61 deletions(-)
>  delete mode 100644 package/connman/0001-xtables.patch

Does this fixes some autobuilder failures?

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Gustavo Zacarias]

On 19/05/16 11:06, Thomas Petazzoni wrote:

> Hello,
>
> On Thu, 19 May 2016 10:25:37 -0300, Gustavo Zacarias wrote:
>> This was added in 0dece985 to deal with kernel headers 4.5 -> 4.5.4
>> breakage. Unfortunately the fix in 4.5.5 and 4.6+ doesn't deal correctly
>> with this causing more build breakage, so we'd rather "break" for the
>> small range of 4.5.x versions broken than for the rest of time.
>>
>> Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
>> ---
>>   package/connman/0001-xtables.patch | 61 --------------------------------------
>>   1 file changed, 61 deletions(-)
>>   delete mode 100644 package/connman/0001-xtables.patch
>
> Does this fixes some autobuilder failures?

Hi.
Not right now since 4.6 is in -next, however it will if the stable 
headers are bumped to 4.5.5.
Regards.

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Thomas Petazzoni]

Hello,

On Thu, 19 May 2016 11:06:57 -0300, Gustavo Zacarias wrote:

> Not right now since 4.6 is in -next, however it will if the stable 
> headers are bumped to 4.5.5.

Well, your patch 1/2 does:

-	default "4.4.10"	if BR2_KERNEL_HEADERS_4_4
-	default "4.5.4"		if BR2_KERNEL_HEADERS_4_5
+	default "4.4.11"	if BR2_KERNEL_HEADERS_4_4
+	default "4.5.5"		if BR2_KERNEL_HEADERS_4_5

So it should fix the thing, right?

However, it means I'd have to rebuild once again the Buildroot external
toolchains, at least the ones that use 4.5 kernel headers, right?

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Gustavo Zacarias]

On 19/05/16 11:14, Thomas Petazzoni wrote:

> Hello,
>
> On Thu, 19 May 2016 11:06:57 -0300, Gustavo Zacarias wrote:
>
>> Not right now since 4.6 is in -next, however it will if the stable
>> headers are bumped to 4.5.5.
>
> Well, your patch 1/2 does:
>
> -	default "4.4.10"	if BR2_KERNEL_HEADERS_4_4
> -	default "4.5.4"		if BR2_KERNEL_HEADERS_4_5
> +	default "4.4.11"	if BR2_KERNEL_HEADERS_4_4
> +	default "4.5.5"		if BR2_KERNEL_HEADERS_4_5
>
> So it should fix the thing, right?
>
> However, it means I'd have to rebuild once again the Buildroot external
> toolchains, at least the ones that use 4.5 kernel headers, right?
>
> Thomas

Hi.
Well, no, after bumping to 4.5.5 connman will go broken with the hacky 
patch because it's incompatible with the kernel fix upstream.
It's a shitty situation, and yes, you'll have to rebuild, sorry.
This is what happens when kids break the headers.
Regards.

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Thomas Petazzoni]

Hello,

On Thu, 19 May 2016 11:18:23 -0300, Gustavo Zacarias wrote:

> Well, no, after bumping to 4.5.5 connman will go broken with the hacky 
> patch because it's incompatible with the kernel fix upstream.

Yes, but if we:

 1/ Apply your patch 1/2 which bumps to 4.5.5

 2/ Apply your patch 2/2 which removes the patch

 3/ Rebuild the pre-built Buildroot external toolchains that used 4.5.4
    headers

Then, the connman build failures should go away, right?

Thanks,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Gustavo Zacarias]

On 19/05/16 11:51, Thomas Petazzoni wrote:

> Hello,
>
> On Thu, 19 May 2016 11:18:23 -0300, Gustavo Zacarias wrote:
>
>> Well, no, after bumping to 4.5.5 connman will go broken with the hacky
>> patch because it's incompatible with the kernel fix upstream.
>
> Yes, but if we:
>
>   1/ Apply your patch 1/2 which bumps to 4.5.5
>
>   2/ Apply your patch 2/2 which removes the patch
>
>   3/ Rebuild the pre-built Buildroot external toolchains that used 4.5.4
>      headers
>
> Then, the connman build failures should go away, right?

Hi.
Yes, exactly.
The only caveat would be people with toolchain headers 4.5.0 - 4.5.4 or 
any eventual external toolchain that ships that way (which by definition 
would be "broken").
Regards.

[Buildroot] [PATCH 2/2] connman: remove xtables build fix patch

[Thomas Petazzoni]

Hello,

On Thu, 19 May 2016 11:56:19 -0300, Gustavo Zacarias wrote:

> The only caveat would be people with toolchain headers 4.5.0 - 4.5.4 or 
> any eventual external toolchain that ships that way (which by definition 
> would be "broken").

Yes, understood. But I believe that's a reasonable trade-off. Building
a toolchain with 4.5 headers (and therefore running 4.5) is really being
bleeding-edge, so it's pretty unlikely to see many users doing that.

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 1/2] linux-headers: bump 3.14.x and 4.{4, 5}.x series

[Bernd Kuhls]

Am Thu, 19 May 2016 10:25:36 -0300 schrieb Gustavo Zacarias:

> Kernel headers 4.5 -> 4.5.4 broke netfilter uapi headers leading to
> build failures for cases where glibc's net/if.h was included before
> linux/if.h This is fixed since 4.5.5 via linux kernel commit 4a91cb61
> and fixes:
> http://autobuild.buildroot.net/
results/736/7362431f62ebb1b436ac41ea3ef85228c763e6f3/

Tested-by: Bernd Kuhls <bernd.kuhls@t-online.de>
(compile-tested collectd, fixes
http://autobuild.buildroot.net/results/ae1/
ae1e3a33a1c2b7433ca7a04659467e2017756b00/)

[Buildroot] [PATCH 1/2] linux-headers: bump 3.14.x and 4.{4, 5}.x series

[Thomas Petazzoni]

Hello,

On Thu, 19 May 2016 10:25:36 -0300, Gustavo Zacarias wrote:
> Kernel headers 4.5 -> 4.5.4 broke netfilter uapi headers leading to
> build failures for cases where glibc's net/if.h was included before
> linux/if.h
> This is fixed since 4.5.5 via linux kernel commit 4a91cb61 and fixes:
> http://autobuild.buildroot.net/results/736/7362431f62ebb1b436ac41ea3ef85228c763e6f3/
> 
> Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
> ---
>  package/linux-headers/Config.in.host | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)

Both patches applied to master, thanks.

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com