| Message ID | 20150324223049.GA4134@vostro |
|---|---|
| State | Superseded |
| Headers | show |
Dear Alex Suykov, On Wed, 25 Mar 2015 00:30:49 +0200, Alex Suykov wrote: > Fixes > http://autobuild.buildroot.net/results/8c666ef11310868097601b6de1dd25779729b878/ > > if_ether.h issue discussion: http://www.openwall.com/lists/musl/2012/10/11/1 > > Signed-off-by: Alex Suykov <alex.suykov@gmail.com> > --- > package/aiccu/0004-if-ether-header.patch | 23 +++++++++++++++++++++++ > package/aiccu/0005-res-ninit.patch | 22 ++++++++++++++++++++++ > 2 files changed, 45 insertions(+) > create mode 100644 package/aiccu/0004-if-ether-header.patch > create mode 100644 package/aiccu/0005-res-ninit.patch Thanks. What is the upstreaming plan for those patches? Thomas
Wed, Mar 25, 2015 at 08:09:55PM +0100, Thomas Petazzoni wrote: > > package/aiccu/0004-if-ether-header.patch | 23 +++++++++++++++++++++++ > > package/aiccu/0005-res-ninit.patch | 22 ++++++++++++++++++++++ > > Thanks. What is the upstreaming plan for those patches? None I'd say, their last release is 8 years old and patches 0001-0003 have been around for quite some time as well. By the way, please disregard this patch. I just realized 0001 does exactly the same thing for uclibc that I did with 0005 for musl. It's better to merge them.
diff --git a/package/aiccu/0004-if-ether-header.patch b/package/aiccu/0004-if-ether-header.patch new file mode 100644 index 0000000..e0ecd74 --- /dev/null +++ b/package/aiccu/0004-if-ether-header.patch @@ -0,0 +1,23 @@ +musl does not allow using <netinet/*> together with <linux/*> headers. +Since there are both netinet/if_ether.h and linux/if_ether.h providing +the same definitions, use linux/if_ether.h whenever other linux/* +headers are included. + +Signed-off-by: Alex Suykov <alex.suykov@gmail.com> + +--- aiccu/common/common.h ++++ aiccu/common/common.h +@@ -91,11 +91,12 @@ + #include <sys/select.h> + + #include <net/if.h> +- #include <netinet/if_ether.h> + #ifdef linux + #include <netpacket/packet.h> ++ #include <linux/if_ether.h> + #include <linux/if_tun.h> + #else ++ #include <netinet/if_ether.h> + #ifdef _DFBSD + #include <net/tun/if_tun.h> + #else diff --git a/package/aiccu/0005-res-ninit.patch b/package/aiccu/0005-res-ninit.patch new file mode 100644 index 0000000..f25b53b --- /dev/null +++ b/package/aiccu/0005-res-ninit.patch @@ -0,0 +1,22 @@ +With _LINUX defined, getrrs uses res_ninit instead of res_init. + +Musl does not provide res_ninit. + +Since getrrs is only called from a single location before any threads +are spawned, it is actually safe to use res_init here. + +Signed-off-by: Alex Suykov <alex.suykov@gmail.com> + +--- aiccu/common/resolver.c ++++ aiccu/common/resolver.c +@@ -24,6 +24,10 @@ + #include <resolv.h> + #include <netdb.h> + ++#ifdef _LINUX ++#undef _LINUX ++#endif ++ + int getrrs(const char *label, int rrtype, void gotrec(unsigned int num, int type, const char *record)) + { + #ifdef _LINUX
Fixes http://autobuild.buildroot.net/results/8c666ef11310868097601b6de1dd25779729b878/ if_ether.h issue discussion: http://www.openwall.com/lists/musl/2012/10/11/1 Signed-off-by: Alex Suykov <alex.suykov@gmail.com> --- package/aiccu/0004-if-ether-header.patch | 23 +++++++++++++++++++++++ package/aiccu/0005-res-ninit.patch | 22 ++++++++++++++++++++++ 2 files changed, 45 insertions(+) create mode 100644 package/aiccu/0004-if-ether-header.patch create mode 100644 package/aiccu/0005-res-ninit.patch