linux-libc-headers: fix duplicate IFF_LOWER_UP DORMANT ECHO on musl

PATCH Superseded André Draszik 5 participants 10 messages Jun 22, 2017 – Jun 26, 2017

lore

Share & tools

lore

André Draszik

author

2017-06-22 15:14 UTC #0 lore raw

PATCH 1/1 linux-libc-headers: fix duplicate IFF_LOWER_UP DORMANT ECHO on musl

From: André Draszik <[email protected]>

musl _does_ define IFF_LOWER_UP DORMANT ECHO so we should prevent redefinition of these when on musl.

As per the included patch, this can be triggered by (from connman 6to4.c):
include <errno.h>
include <stdio.h>
include <stdlib.h>
include <string.h>
include <sys/socket.h>
include <netinet/in.h>
include <arpa/inet.h>
include <net/if.h>
include <linux/ip.h>
include <linux/if_tunnel.h>
include <linux/netlink.h>
include <linux/rtnetlink.h>
include <sys/ioctl.h>
include <unistd.h>

In file included from ../git/src/6to4.c:34:0:
.../usr/include/linux/if.h:97:2: error: expected identifier before numeric constant
IFF_LOWER_UP = 1<<16, /* __volatile__ */
^

This is because at that time, IFF_LOWER_UP has been converted to 0x10000 already:
enum net_device_flags {
0x10000 = 1<<16,
0x20000 = 1<<17,
0x40000 = 1<<18,

};

By defining __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO to 0, we avoid the duplicated definition.

Signed-off-by André Draszik <[email protected]>

---
 ....h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch | 62 ++++++++++++++++++++++
 .../linux-libc-headers/linux-libc-headers_4.10.bb  |  1 +
 2 files changed, 63 insertions(+)
 create mode 100644 meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch

meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +62 -0

new file mode 100644
index 0000000000..fd1bb9edb8
@@ -0,0 +1,62 @@
+From 08a04c025395b0580913285045f2af36ee0985db Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Andr=C3=A9=20Draszik?= <[email protected]>
+Date: Thu, 15 Jun 2017 16:55:33 +0100
+Subject: [PATCH] libc-compat.h: musl _does_ define IFF_LOWER_UP DORMANT ECHO
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+To trigger this (from connman 6to4.c):
+    include <errno.h>
+    include <stdio.h>
+    include <stdlib.h>
+    include <string.h>
+    include <sys/socket.h>
+    include <netinet/in.h>
+    include <arpa/inet.h>
+    include <net/if.h>
+    include <linux/ip.h>
+    include <linux/if_tunnel.h>
+    include <linux/netlink.h>
+    include <linux/rtnetlink.h>
+    include <sys/ioctl.h>
+    include <unistd.h>
+
+In file included from ../git/src/6to4.c:34:0:
+/scratch/yocto/build-tgm-r3-poky-fpp-tgm.pyro/tmp/work/tgm_r3-poky-linux-musl/connman/1.33-git4+AUTOINC+aa5b3dc12b-r0/recipe-sysroot/usr/include/linux/if.h:97:2: error: expected identifier before numeric constant
+  IFF_LOWER_UP   = 1<<16, /* __volatile__ */
+  ^
+
+This is because at that time, IFF_LOWER_UP has been converted
+to 0x10000 already:
+enum net_device_flags {
+ 0x10000 = 1<<16,
+ 0x20000 = 1<<17,
+ 0x40000 = 1<<18,
+
+};
+
+Upstream-Status: Pending
+Signed-off-by: André Draszik <[email protected]>
+Acked-by: Stephane Ayotte <[email protected]>
+---
+ include/uapi/linux/libc-compat.h | 3 ++-
+ 1 file changed, 2 insertions(+), 1 deletion(-)
+
+diff --git a/include/uapi/linux/libc-compat.h b/include/uapi/linux/libc-compat.h
+index f11ec0e..2003ae4 100644
+--- a/include/uapi/linux/libc-compat.h
++++ b/include/uapi/linux/libc-compat.h
+@@ -70,7 +70,8 @@
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS 0
+ /* For the future if glibc adds IFF_LOWER_UP, IFF_DORMANT and IFF_ECHO */
+ #ifndef __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO
+-#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 1
++/* musl has these defined */
++#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 0
+ #endif /* __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO */
+ 
+ #else /* _NET_IF_H */
+-- 
+2.11.0
+

meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.10.bb +1 -0

index 108446aa34..29262789a8 100644
@@ -4,6 +4,7 @@ SRC_URI_append_libc-musl = "\
     file://0001-libc-compat.h-fix-some-issues-arising-from-in6.h.patch \
     file://0002-libc-compat.h-prevent-redefinition-of-struct-ethhdr.patch \
     file://0003-remove-inclusion-of-sysinfo.h-in-kernel.h.patch \
+    file://0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch \
    "
 
 SRC_URI[md5sum] = "b5e7f6b9b2fe1b6cc7bc56a3a0bfc090"

Signature · 1 line

2.11.0

Khem Raj

2017-06-22 22:27 UTC #1 lore raw

I would rather fix the package instead of kernel headers unless kernel community accepts this patch

Quoted 148 lines · André Draszik

From: André Draszik <[email protected]>
musl _does_ define IFF_LOWER_UP DORMANT ECHO so we should prevent redefinition of these when on musl.
As per the included patch, this can be triggered by (from connman 6to4.c):
include <errno.h>
include <stdio.h>
include <stdlib.h>
include <string.h>
include <sys/socket.h>
include <netinet/in.h>
include <arpa/inet.h>
include <net/if.h>
include <linux/ip.h>
include <linux/if_tunnel.h>
include <linux/netlink.h>
include <linux/rtnetlink.h>
include <sys/ioctl.h>
include <unistd.h>
In file included from ../git/src/6to4.c:34:0:
.../usr/include/linux/if.h:97:2: error: expected identifier before numeric constant
IFF_LOWER_UP = 1<<16, /* __volatile__ */
^
This is because at that time, IFF_LOWER_UP has been converted to 0x10000 already:
enum net_device_flags {
0x10000 = 1<<16,
0x20000 = 1<<17,
0x40000 = 1<<18,
};
By defining __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO to 0, we avoid the duplicated definition.
Signed-off-by André Draszik <[email protected]>
---
 ....h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch | 62
++++++++++++++++++++++
 .../linux-libc-headers/linux-libc-headers_4.10.bb  |  1 +
 2 files changed, 63 insertions(+)
 create mode 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
diff --git a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file mode 100644
index 0000000000..fd1bb9edb8
--- /dev/null
+++
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +4 -0
@@ -0,0 +1,62 @@
+From 08a04c025395b0580913285045f2af36ee0985db Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Andr=C3=A9=20Draszik?= <[email protected]>
+Date: Thu, 15 Jun 2017 16:55:33 +0100
+Subject: [PATCH] libc-compat.h: musl _does_ define IFF_LOWER_UP DORMANT
ECHO
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +22 -0
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+To trigger this (from connman 6to4.c):
+    include <errno.h>
+    include <stdio.h>
+    include <stdlib.h>
+    include <string.h>
+    include <sys/socket.h>
+    include <netinet/in.h>
+    include <arpa/inet.h>
+    include <net/if.h>
+    include <linux/ip.h>
+    include <linux/if_tunnel.h>
+    include <linux/netlink.h>
+    include <linux/rtnetlink.h>
+    include <sys/ioctl.h>
+    include <unistd.h>
+
+In file included from ../git/src/6to4.c:34:0:
+/scratch/yocto/build-tgm-r3-poky-fpp-tgm.pyro/tmp/work/tgm_r3-poky-linux-musl/connman/1.33-git4+AUTOINC+aa5b3dc12b-r0/recipe-sysroot/usr/include/linux/if.h:97:2:
error: expected identifier before numeric constant
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +20 -0
+  IFF_LOWER_UP   = 1<<16, /* __volatile__ */
+  ^
+
+This is because at that time, IFF_LOWER_UP has been converted
+to 0x10000 already:
+enum net_device_flags {
+ 0x10000 = 1<<16,
+ 0x20000 = 1<<17,
+ 0x40000 = 1<<18,
+
+};
+
+Upstream-Status: Pending
+Signed-off-by: André Draszik <[email protected]>
+Acked-by: Stephane Ayotte <[email protected]>
+---
+ include/uapi/linux/libc-compat.h | 3 ++-
+ 1 file changed, 2 insertions(+), 1 deletion(-)
+
+diff --git a/include/uapi/linux/libc-compat.h
b/include/uapi/linux/libc-compat.h
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +16 -0
+index f11ec0e..2003ae4 100644
+--- a/include/uapi/linux/libc-compat.h
++++ b/include/uapi/linux/libc-compat.h
+@@ -70,7 +70,8 @@
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS 0
+ /* For the future if glibc adds IFF_LOWER_UP, IFF_DORMANT and IFF_ECHO */
+ #ifndef __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO
+-#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 1
++/* musl has these defined */
++#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 0
+ #endif /* __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO */
+
+ #else /* _NET_IF_H */
+--
+2.11.0
+
(quoted diff) +0 -0
linux-libc-headers_4.10.bb b/meta/recipes-kernel/linux-libc-headers/ linux-libc-headers_4.10.bb
index 108446aa34..29262789a8 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.10.bb +1 -0
@@ -4,6 +4,7 @@ SRC_URI_append_libc-musl = "\
     file://0001-libc-compat.h-fix-some-issues-arising-from-in6.h.patch \
     file://0002-libc-compat.h-prevent-redefinition-of-struct-ethhdr.patch
 \
     file://0003-remove-inclusion-of-sysinfo.h-in-kernel.h.patch \
+
 file://0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch \
    "
SRC_URI[md5sum] = "b5e7f6b9b2fe1b6cc7bc56a3a0bfc090"
--
2.11.0
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

André Draszik

author

2017-06-23 10:46 UTC #2 lore raw

connman is not doing anything wrong here.

The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).

libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.

Cheers,
Andre'

Quoted 163 lines · Khem Raj

I would rather fix the package instead of kernel headers unless kernel community accepts this patch
Quoted 158 lines · André Draszik
From: André Draszik <[email protected]>
musl _does_ define IFF_LOWER_UP DORMANT ECHO so we should prevent redefinition of these when on musl.
As per the included patch, this can be triggered by (from connman 6to4.c):
    include <errno.h>
    include <stdio.h>
    include <stdlib.h>
    include <string.h>
    include <sys/socket.h>
    include <netinet/in.h>
    include <arpa/inet.h>
    include <net/if.h>
    include <linux/ip.h>
    include <linux/if_tunnel.h>
    include <linux/netlink.h>
    include <linux/rtnetlink.h>
    include <sys/ioctl.h>
    include <unistd.h>
In file included from ../git/src/6to4.c:34:0:
.../usr/include/linux/if.h:97:2: error: expected identifier before numeric
constant
IFF_LOWER_UP   = 1<<16, /* __volatile__ */
^
This is because at that time, IFF_LOWER_UP has been converted to 0x10000 already:
enum net_device_flags {
0x10000 = 1<<16,
0x20000 = 1<<17,
0x40000 = 1<<18,
};
By defining __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO to 0, we avoid the duplicated definition.
Signed-off-by André Draszik <[email protected]>
---
....h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch | 62
++++++++++++++++++++++
.../linux-libc-headers/linux-libc-headers_4.10.bb  |  1 +
 2 files changed, 63 insertions(+)
create mode 100644 meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
diff --git a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
new file mode 100644
index 0000000000..fd1bb9edb8
--- /dev/null
+++
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +4 -0
@@ -0,0 +1,62 @@
+From 08a04c025395b0580913285045f2af36ee0985db Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Andr=C3=A9=20Draszik?= <[email protected]>
+Date: Thu, 15 Jun 2017 16:55:33 +0100
+Subject: [PATCH] libc-compat.h: musl _does_ define IFF_LOWER_UP DORMANT
ECHO
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +22 -0
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+To trigger this (from connman 6to4.c):
+    include <errno.h>
+    include <stdio.h>
+    include <stdlib.h>
+    include <string.h>
+    include <sys/socket.h>
+    include <netinet/in.h>
+    include <arpa/inet.h>
+    include <net/if.h>
+    include <linux/ip.h>
+    include <linux/if_tunnel.h>
+    include <linux/netlink.h>
+    include <linux/rtnetlink.h>
+    include <sys/ioctl.h>
+    include <unistd.h>
+
+In file included from ../git/src/6to4.c:34:0:
+/scratch/yocto/build-tgm-r3-poky-fpp-tgm.pyro/tmp/work/tgm_r3-poky-
linux-musl/connman/1.33-git4+AUTOINC+aa5b3dc12b-r0/recipe- sysroot/usr/include/linux/if.h:97:2:
error: expected identifier before numeric constant + IFF_LOWER_UP = 1<<16, /* __volatile__ */
+ ^
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +18 -0
+
+This is because at that time, IFF_LOWER_UP has been converted
+to 0x10000 already:
+enum net_device_flags {
+ 0x10000 = 1<<16,
+ 0x20000 = 1<<17,
+ 0x40000 = 1<<18,
+
+};
+
+Upstream-Status: Pending
+Signed-off-by: André Draszik <[email protected]>
+Acked-by: Stephane Ayotte <[email protected]>
+---
+ include/uapi/linux/libc-compat.h | 3 ++-
+ 1 file changed, 2 insertions(+), 1 deletion(-)
+
+diff --git a/include/uapi/linux/libc-compat.h
b/include/uapi/linux/libc-compat.h
+index f11ec0e..2003ae4 100644
+--- a/include/uapi/linux/libc-compat.h
++++ b/include/uapi/linux/libc-compat.h
+@@ -70,7 +70,8 @@
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS 0
+ /* For the future if glibc adds IFF_LOWER_UP, IFF_DORMANT and IFF_ECHO */
+ #ifndef __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO +-#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 1 ++/* musl has these defined */
++#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 0
+ #endif /* __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO */
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +5 -0
+
+ #else /* _NET_IF_H */
+--
+2.11.0
+
(quoted diff) +0 -0
linux-libc-headers_4.10.bb b/meta/recipes-kernel/linux-libc-headers/ linux-libc-headers_4.10.bb
index 108446aa34..29262789a8 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.10.bb +0 -0
@@ -4,6 +4,7 @@ SRC_URI_append_libc-musl = "\
     file://0001-libc-compat.h-fix-some-issues-arising-from-in6.h.patch \
     file://0002-libc-compat.h-prevent-redefinition-of-struct- ethhdr.patch
\
     file://0003-remove-inclusion-of-sysinfo.h-in-kernel.h.patch \
+ file://0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch \
    "
SRC_URI[md5sum] = "b5e7f6b9b2fe1b6cc7bc56a3a0bfc090" --
2.11.0
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Khem Raj

2017-06-23 13:38 UTC #3 lore raw

Quoted 2 lines · André Draszik

connman is not doing anything wrong here.

yes I am aware of this

Quoted 6 lines

The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.

see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch

I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.

Quoted 172 lines

Cheers,
Andre'
Quoted 157 lines · Khem Raj
I would rather fix the package instead of kernel headers unless kernel community accepts this patch
Quoted 152 lines · André Draszik
From: André Draszik <[email protected]>
musl _does_ define IFF_LOWER_UP DORMANT ECHO so we should prevent redefinition of these when on musl.
As per the included patch, this can be triggered by (from connman 6to4.c):
include <errno.h>
include <stdio.h>
include <stdlib.h>
include <string.h>
include <sys/socket.h>
include <netinet/in.h>
include <arpa/inet.h>
include <net/if.h>
include <linux/ip.h>
include <linux/if_tunnel.h>
include <linux/netlink.h>
include <linux/rtnetlink.h>
include <sys/ioctl.h>
include <unistd.h>
In file included from ../git/src/6to4.c:34:0:
.../usr/include/linux/if.h:97:2: error: expected identifier before numeric
constant
IFF_LOWER_UP = 1<<16, /* __volatile__ */
^
This is because at that time, IFF_LOWER_UP has been converted to 0x10000 already:
enum net_device_flags {
0x10000 = 1<<16,
0x20000 = 1<<17,
0x40000 = 1<<18,
};
By defining __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO to 0, we avoid the duplicated definition.
Signed-off-by André Draszik <[email protected]>
---
 ....h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch | 62
++++++++++++++++++++++
 .../linux-libc-headers/linux-libc-headers_4.10.bb  |  1 +
 2 files changed, 63 insertions(+)
 create mode 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
diff --git a/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
new file mode 100644
index 0000000000..fd1bb9edb8
--- /dev/null
+++
b/meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc- compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +4 -0
@@ -0,0 +1,62 @@
+From 08a04c025395b0580913285045f2af36ee0985db Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Andr=C3=A9=20Draszik?= <[email protected]>
+Date: Thu, 15 Jun 2017 16:55:33 +0100
+Subject: [PATCH] libc-compat.h: musl _does_ define IFF_LOWER_UP DORMANT
ECHO
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +22 -0
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+To trigger this (from connman 6to4.c):
+    include <errno.h>
+    include <stdio.h>
+    include <stdlib.h>
+    include <string.h>
+    include <sys/socket.h>
+    include <netinet/in.h>
+    include <arpa/inet.h>
+    include <net/if.h>
+    include <linux/ip.h>
+    include <linux/if_tunnel.h>
+    include <linux/netlink.h>
+    include <linux/rtnetlink.h>
+    include <sys/ioctl.h>
+    include <unistd.h>
+
+In file included from ../git/src/6to4.c:34:0:
+/scratch/yocto/build-tgm-r3-poky-fpp-tgm.pyro/tmp/work/tgm_r3-poky-
linux-musl/connman/1.33-git4+AUTOINC+aa5b3dc12b-r0/recipe- sysroot/usr/include/linux/if.h:97:2:
error: expected identifier before numeric constant
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +20 -0
+  IFF_LOWER_UP   = 1<<16, /* __volatile__ */
+  ^
+
+This is because at that time, IFF_LOWER_UP has been converted
+to 0x10000 already:
+enum net_device_flags {
+ 0x10000 = 1<<16,
+ 0x20000 = 1<<17,
+ 0x40000 = 1<<18,
+
+};
+
+Upstream-Status: Pending
+Signed-off-by: André Draszik <[email protected]>
+Acked-by: Stephane Ayotte <[email protected]>
+---
+ include/uapi/linux/libc-compat.h | 3 ++-
+ 1 file changed, 2 insertions(+), 1 deletion(-)
+
+diff --git a/include/uapi/linux/libc-compat.h
b/include/uapi/linux/libc-compat.h
+index f11ec0e..2003ae4 100644
+--- a/include/uapi/linux/libc-compat.h
++++ b/include/uapi/linux/libc-compat.h
+@@ -70,7 +70,8 @@
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS 0
+ /* For the future if glibc adds IFF_LOWER_UP, IFF_DORMANT and IFF_ECHO */
+ #ifndef __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO +-#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 1 ++/* musl has these defined */
++#define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 0
+ #endif /* __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO */
meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +5 -0
+
+ #else /* _NET_IF_H */
+--
+2.11.0
+
(quoted diff) +0 -0
linux-libc-headers_4.10.bb b/meta/recipes-kernel/linux-libc-headers/ linux-libc-headers_4.10.bb
index 108446aa34..29262789a8 100644
meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.10.bb +0 -0
@@ -4,6 +4,7 @@ SRC_URI_append_libc-musl = "\
     file://0001-libc-compat.h-fix-some-issues-arising-from-in6.h.patch
 \
     file://0002-libc-compat.h-prevent-redefinition-of-struct-
ethhdr.patch
\
     file://0003-remove-inclusion-of-sysinfo.h-in-kernel.h.patch \
+
file://0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch \
    "
SRC_URI[md5sum] = "b5e7f6b9b2fe1b6cc7bc56a3a0bfc090"
--
2.11.0
Quoted 6 lines
Quoted 6 lines
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

[email protected]

2017-06-23 14:17 UTC #4 lore raw

Hi,

I'm chipping in since I've been messing with these things a bit in upstream Linux kernel.

Quoted 6 lines · Khem Raj

Quoted 2 lines · André Draszik
connman is not doing anything wrong here.
yes I am aware of this

Quoted 8 lines

Quoted 6 lines
The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.
see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch

I'm aware of those review comments but I have not seen any patches posted which fix the problem in some other way. Thus I would propose to apply these patches as a workaround until upstream fixes the issues.

These header files do not change that often either.

Quoted 4 lines

I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.

Upstream-Status: Denied

would be a correct marker for now I guess.

Hope this helps,

-Mikko

Khem Raj

2017-06-23 23:21 UTC #5 lore raw

Quoted 18 lines · On Fri, Jun 23, 2017 at 7:17 AM, <[email protected]> wrote:

Hi,
I'm chipping in since I've been messing with these things a bit in upstream Linux kernel.
Quoted 12 lines · Khem Raj
Quoted 2 lines · André Draszik
connman is not doing anything wrong here.
yes I am aware of this
Quoted 6 lines
The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.

Quoted 8 lines

Quoted 2 lines
see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch
I'm aware of those review comments but I have not seen any patches posted which fix the problem in some other way. Thus I would propose to apply these patches as a workaround until upstream fixes the issues.
These header files do not change that often either.

problem is you become incompatible ABI forever that worries me. However if bruce is fine to carry this patch as part of linux-yocto I might relent. It still will be hassle where folks will have to apply this patch to there kernels if they are building musl based systems.

Quoted 9 lines

Quoted 5 lines
I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.
Upstream-Status: Denied
would be a correct marker for now I guess.

I would rather see some progress made to get it resolved :) we need to actually remove glibc'ness completely from kernel. and this will fix itself.

Quoted 4 lines

Hope this helps,
-Mikko

Denys Dmytriyenko

2017-06-23 23:41 UTC #6 lore raw

Quoted 22 lines · Khem Raj

Quoted 21 lines · On Fri, Jun 23, 2017 at 7:17 AM, <[email protected]> wrote:
Hi,
I'm chipping in since I've been messing with these things a bit in upstream Linux kernel.
Quoted 14 lines · Khem Raj
Quoted 2 lines · André Draszik
connman is not doing anything wrong here.
yes I am aware of this
Quoted 6 lines
The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.
see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch

Quoted 7 lines

Quoted 5 lines
I'm aware of those review comments but I have not seen any patches posted which fix the problem in some other way. Thus I would propose to apply these patches as a workaround until upstream fixes the issues.
These header files do not change that often either.
problem is you become incompatible ABI forever that worries me.

Quoted 3 lines

However if bruce is fine to carry this patch as part of linux-yocto I might relent. It still will be hassle where folks will have to apply this patch to there kernels if they are building musl based systems.

Don't forget that not everyone is using linux-yocto kernel!

Quoted 12 lines

Quoted 8 lines
Quoted 4 lines
I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.
Upstream-Status: Denied
would be a correct marker for now I guess.
I would rather see some progress made to get it resolved :) we need to actually remove glibc'ness completely from kernel. and this will fix itself.

Andre McCurdy

2017-06-24 00:54 UTC #7 lore raw

Quoted 31 lines · Denys Dmytriyenko

Quoted 22 lines · Khem Raj
Quoted 21 lines · On Fri, Jun 23, 2017 at 7:17 AM, <[email protected]> wrote:
Hi,
I'm chipping in since I've been messing with these things a bit in upstream Linux kernel.
Quoted 14 lines · Khem Raj
Quoted 2 lines · André Draszik
connman is not doing anything wrong here.
yes I am aware of this
Quoted 6 lines
The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.
see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch
Quoted 7 lines
Quoted 5 lines
I'm aware of those review comments but I have not seen any patches posted which fix the problem in some other way. Thus I would propose to apply these patches as a workaround until upstream fixes the issues.
These header files do not change that often either.
problem is you become incompatible ABI forever that worries me.

Quoted 7 lines

+1
Quoted 3 lines
However if bruce is fine to carry this patch as part of linux-yocto I might relent. It still will be hassle where folks will have to apply this patch to there kernels if they are building musl based systems.
Don't forget that not everyone is using linux-yocto kernel!

Neither is linux-libc-headers, so getting these musl patches into the linux-yocto kernel isn't going to change much...

Quoted 17 lines

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines
I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.
Quoted 7 lines
Quoted 3 lines
Upstream-Status: Denied
would be a correct marker for now I guess.
I would rather see some progress made to get it resolved :) we need to actually remove glibc'ness completely from kernel. and this will fix itself.
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

[email protected]

2017-06-26 07:02 UTC #8 lore raw

In reply to Khem Raj

Quoted 26 lines · Khem Raj

Quoted 25 lines · On Fri, Jun 23, 2017 at 7:17 AM, <[email protected]> wrote:
Hi,
I'm chipping in since I've been messing with these things a bit in upstream Linux kernel.
Quoted 6 lines · Khem Raj
Quoted 2 lines · André Draszik
connman is not doing anything wrong here.
yes I am aware of this
Quoted 8 lines
Quoted 6 lines
The kernel is redefining IFF_LOWER_UP, because it thinks the libc doesn't define it yet (and glibc doesn't).
libc-compat.h is the way to solve these kind of issues. There also is https:
//lkml.org/lkml/2017/3/12/238 which is very similar. I'll pick that instead.
see the comment https://lkml.org/lkml/2017/3/16/121
that worries me for this patch
I'm aware of those review comments but I have not seen any patches posted which fix the problem in some other way. Thus I would propose to apply these patches as a workaround until upstream fixes the issues.

Quoted 6 lines

Quoted 1 line
These header files do not change that often either.
problem is you become incompatible ABI forever that worries me. However if bruce is fine to carry this patch as part of linux-yocto I might relent. It still will be hassle where folks will have to apply this patch to there kernels if they are building musl based systems.

API or ABI? But agreed this is a problem.

Quoted 6 lines

Quoted 6 lines
Quoted 5 lines
I am not questioning the correctness of patch too. But it would be better to get this patch accepted into kernel before applying to OE since these are kind of patches which you can get stuck with for life if upstream is not accepting it.

Quoted 7 lines

Quoted 3 lines
Upstream-Status: Denied
would be a correct marker for now I guess.
I would rather see some progress made to get it resolved :) we need to actually remove glibc'ness completely from kernel. and this will fix itself.

Yes. With the glibc'ness in the kernel headers I was just following existing examples and adding some more. Some of those changes got accepted and only later came the resistance to remove glibc'ness completely. No-one else has proposed patches. Maybe I should invent some hobby project time again for this...

-Mikko

André Draszik

author

2017-06-23 11:42 UTC #9 lore raw

PATCH v2 1/1 linux-libc-headers: fix duplicate IFF_LOWER_UP DORMANT ECHO on musl

In reply to André Draszik

From: André Draszik <[email protected]>

musl _does_ define IFF_LOWER_UP DORMANT ECHO so we should prevent redefinition of these when on musl.

In file included from ../git/src/6to4.c:34:0:
.../usr/include/linux/if.h:97:2: error: expected identifier before numeric constant
IFF_LOWER_UP = 1<<16, /* __volatile__ */
^

This is because at that time, IFF_LOWER_UP has been converted to 0x10000 already:
enum net_device_flags {
0x10000 = 1<<16,
0x20000 = 1<<17,
0x40000 = 1<<18,

};

Backport a patch that addresses this.

Signed-off-by André Draszik <[email protected]>

---
 ....h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch | 46 ++++++++++++++++++++++
 .../linux-libc-headers/linux-libc-headers_4.10.bb  |  1 +
 2 files changed, 47 insertions(+)
 create mode 100644 meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch

meta/recipes-kernel/linux-libc-headers/linux-libc-headers/0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch new file +46 -0

new file mode 100644
index 0000000000..bde2132ebf
@@ -0,0 +1,46 @@
+Content-Type: text/plain; charset="utf-8"
+MIME-Version: 1.0
+Content-Transfer-Encoding: 7bit
+Subject: [v2,2/3] uapi glibc compat: fix build if libc defines IFF_ECHO
+From: Hauke Mehrtens <[email protected]>
+X-Patchwork-Id: 9686293
+Message-Id: <[email protected]>
+To: [email protected], [email protected]
+Cc: [email protected], [email protected], [email protected],
+ [email protected], [email protected],
+ [email protected], [email protected], Hauke Mehrtens <[email protected]>
+Date: Tue, 18 Apr 2017 23:00:35 +0200
+
+musl 1.1.15 defines IFF_ECHO and the other net_device_flags options.
+When a user application includes linux/if.h and net/if.h the compile
+will fail.
+
+Activate __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO only when
+it is needed. This should also make this work in case glibc will add
+these defines.
+
+Acked-by: Mikko Rapeli <[email protected]>
+Signed-off-by: Hauke Mehrtens <[email protected]>
+---
+Upstream-Status: Submitted [https://patchwork.kernel.org/patch/9686293/]
+Signed-off-by: André Draszik <[email protected]>
+Acked-by: Stephane Ayotte <[email protected]>
+ include/uapi/linux/libc-compat.h | 2 ++
+ 1 file changed, 2 insertions(+)
+
+diff --git a/include/uapi/linux/libc-compat.h b/include/uapi/linux/libc-compat.h
+index 43a81136ea6e..ce2fa8a4ced6 100644
+--- a/include/uapi/linux/libc-compat.h
++++ b/include/uapi/linux/libc-compat.h
+@@ -64,9 +64,11 @@
+ /* Everything up to IFF_DYNAMIC, matches net/if.h until glibc 2.23 */
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS 0
+ /* For the future if glibc adds IFF_LOWER_UP, IFF_DORMANT and IFF_ECHO */
++#ifndef IFF_ECHO
+ #ifndef __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO
+ #define __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO 1
+ #endif /* __UAPI_DEF_IF_NET_DEVICE_FLAGS_LOWER_UP_DORMANT_ECHO */
++#endif /* IFF_ECHO */
+ 
+ #else /* _NET_IF_H */
+

meta/recipes-kernel/linux-libc-headers/linux-libc-headers_4.10.bb +1 -0

index 108446aa34..29262789a8 100644
@@ -4,6 +4,7 @@ SRC_URI_append_libc-musl = "\
     file://0001-libc-compat.h-fix-some-issues-arising-from-in6.h.patch \
     file://0002-libc-compat.h-prevent-redefinition-of-struct-ethhdr.patch \
     file://0003-remove-inclusion-of-sysinfo.h-in-kernel.h.patch \
+    file://0001-libc-compat.h-musl-_does_-define-IFF_LOWER_UP-DORMAN.patch \
    "
 
 SRC_URI[md5sum] = "b5e7f6b9b2fe1b6cc7bc56a3a0bfc090"

Signature · 1 line

2.11.0