Send connman mailing list submissions to
        [email protected]

To subscribe or unsubscribe via the World Wide Web, visit
        https://lists.01.org/mailman/listinfo/connman
or, via email, send a message with subject or body 'help' to
        [email protected]

You can reach the person managing the list at
        [email protected]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of connman digest..."


Today's Topics:

   1. Re: [PATCH 1/7] mozjs: Fix error handling for
      wrongly-formatted hostname (Patrik Flykt)
   2. Re: proxy configuration (Patrik Flykt)
   3. Re: [PATCH 1/7] mozjs: Fix error handling for
      wrongly-formatted hostname (Marcel Holtmann)
   4. [PATCH] Remove libstdc++ from non-V8 builds (David Woodhouse)


----------------------------------------------------------------------

Message: 1
Date: Fri, 19 Aug 2016 10:15:57 +0300
From: Patrik Flykt <[email protected]>
To: [email protected], [email protected]
Subject: Re: [PATCH 1/7] mozjs: Fix error handling for
        wrongly-formatted hostname
Message-ID: <[email protected]>
Content-Type: text/plain; charset="UTF-8"

On Tue, 2016-08-16 at 17:01 +0100, [email protected] wrote:
> From: David Woodhouse <[email protected]>
> 
> In commit d9ea7b8a ("plugins: Do not try to resolve a wrongly
> formated
> hostname") we introduced a check on the string before we attempt to
> resolve
> it. In the mozjs case we simply return a NULL result instead of
> raising an
> exception when this function fails... except that the new sanity
> check was
> placed before the JS_SET_RVAL() call, and no return value was set at
> all.
> 
> Move it to the right place.

All seven patches applied. I cleaned up some whitespace damage also
from duktape.h and added duktape's .h files to Makefile.am.

Thanks!

        Patrik


------------------------------

Message: 2
Date: Fri, 19 Aug 2016 10:42:45 +0300
From: Patrik Flykt <[email protected]>
To: Thomas Green <[email protected]>, "[email protected]"
        <[email protected]>
Subject: Re: proxy configuration
Message-ID: <[email protected]>
Content-Type: text/plain; charset="UTF-8"


        Hi,

On Wed, 2016-08-17 at 14:13 +0000, Thomas Green wrote:
> No clues?
> ?
> From: connman [mailto:[email protected]] On Behalf Of
> Thomas Green
> Sent: Tuesday, August 16, 2016 8:59 AM
> To: [email protected]
> Subject: proxy configuration
> ?
> I need to understand how conman is finding and using proxies. We have
> seen in wireshark captures that the system is sending out a ?wpad.*?
> DNS query. This is used to find proxies on the network. We don?t want
> connman to find and setup proxies automatically.
> ?
> This is what I get when I dump the configuration for my current wired
> service:
> ?
> /net/connman/service/ethernet_0008720790f4_cable
> ? Type = ethernet
> ? Security = [? ]
> ? State = online
> ? Favorite = True
> ? Immutable = False
> ? AutoConnect = True
> ? Name = Wired
> ? Ethernet = [ Method=auto, Interface=eth0,
> Address=00:08:72:07:90:F4, MTU=1500 ]
> ? IPv4 = [ Method=dhcp, Address=10.20.136.11, Netmask=255.255.255.0,
> Gateway=10.20.136.1 ]
> ? IPv4.Configuration = [ Method=dhcp ]
> ? IPv6 = [? ]
> ? IPv6.Configuration = [ Method=auto, Privacy=disabled ]
> ? Nameservers = [ 10.20.35.16, 192.168.250.16, 192.168.253.16 ]
> ? Nameservers.Configuration = [? ]
> ? Timeservers = [ 10.20.136.1 ]
> ? Timeservers.Configuration = [? ]
> ? Domains = [ xxxxxx.COM ]
> ? Domains.Configuration = [? ]
> ? Proxy = [ Method=direct ]
> ? Proxy.Configuration = [? ]
> ? Provider = [? ]
> ?
> Notice the highlighted line indicates the proxy method is direct. ?I
> am not sure what this means and the documentation is not very
> helpful.

Hmm, looking at this for 7 minutes I was going to say that the method
should be set to direct to avoid wpad lookup.

In src/service.c,?check_proxy_setup(), wpad is tried if the service
proxy method is unknown. If it is direct, no wpad lookups should
happen. The proxy method is set by the service property 'Proxy' over D-
Bus.

But the proxy method cannot be set in a .config file, as no variable is
specified for this. So what now can happen is that a service manages to
connect before someone else manages to set the proxy method to 'direct'
over D-Bus? On subsequent connects the service should have been saved,
so this should only happen on initial startup (unless /var/lib/connman
is wiped over reboot or so).

Looks interesting, debug with at least CONNMAN_WEB_DEBUG=1, -d
src/wpad.c,src/wispr.c to see what is going on...


Cheers,

        Patrik


------------------------------

Message: 3
Date: Fri, 19 Aug 2016 10:44:03 +0200
From: Marcel Holtmann <[email protected]>
To: Patrik Flykt <[email protected]>
Cc: David Woodhouse <[email protected]>, [email protected]
Subject: Re: [PATCH 1/7] mozjs: Fix error handling for
        wrongly-formatted hostname
Message-ID: <[email protected]>
Content-Type: text/plain; charset=us-ascii

Hi Patrik,

>> In commit d9ea7b8a ("plugins: Do not try to resolve a wrongly
>> formated
>> hostname") we introduced a check on the string before we attempt to
>> resolve
>> it. In the mozjs case we simply return a NULL result instead of
>> raising an
>> exception when this function fails... except that the new sanity
>> check was
>> placed before the JS_SET_RVAL() call, and no return value was set at
>> all.
>> 
>> Move it to the right place.
> 
> All seven patches applied. I cleaned up some whitespace damage also
> from duktape.h and added duktape's .h files to Makefile.am.

there is no point to fix anything in duktape/ directory. This should stay 
imports of duktape as they are.

Regards

Marcel



------------------------------

Message: 4
Date: Fri, 19 Aug 2016 13:16:45 +0100
From: David Woodhouse <[email protected]>
To: [email protected]
Subject: [PATCH] Remove libstdc++ from non-V8 builds
Message-ID: <[email protected]>

From: David Woodhouse <[email protected]>

Autohell will automatically use g++ for the final link if it sees any
C++ sources, even if it isn't actually building them today.

So... link plugins/v8.cc into its own intermediate library, so there
*aren't* any C++ sources for src/pacrunner and it doesn't do that.

Add -fno-exceptions to AM_CXXFLAGS, so that it doesn't actually *need*
to link against libstdc++, even when V8 support is being built.

Add -lm to the relevant LDADD variables when duktape is being built,
because it needs trunc() from libm and was magically getting it before.
---
 Makefile.am | 25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

diff --git a/Makefile.am b/Makefile.am
index ec76db6..5564b0a 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -19,6 +19,8 @@ builtin_sources =
 builtin_cflags =
 builtin_libadd =
 
+noinst_LIBRARIES =
+
 if CURL
 builtin_modules += curl
 builtin_sources += plugins/curl.c
@@ -37,17 +39,24 @@ endif
 if V8
 js_sources = src/js_funcs.c
 builtin_modules += v8
-builtin_sources += plugins/v8.cc
-builtin_libadd += @V8_LIBS@
+builtin_libadd += plugins/libv8plugin.a @V8_LIBS@
+
+# If we compile this into a v8.a library first, then autohell
+# won't use g++ to link pacrunner. And thus won't end up linking
+# against libstdc++ even in the builds where V8 is disabled.
+noinst_LIBRARIES += plugins/libv8plugin.a
+
+plugins_libv8plugin_a_SOURCES = plugins/v8.cc
+
 endif
 
 if DUKTAPE
 js_sources = src/js_funcs.c
 builtin_modules += duktape
 builtin_sources += plugins/duktape.c
-builtin_libadd += duktape/libduktape.a
+builtin_libadd += duktape/libduktape.a -lm
 
-noinst_LIBRARIES = duktape/libduktape.a
+noinst_LIBRARIES += duktape/libduktape.a
 
 duktape_libduktape_a_SOURCES = duktape/duk_config.h \
                                duktape/duktape.h \
@@ -142,15 +151,13 @@ unit_test_mozjs_LDADD = @MOZJS_LIBS@ @GLIB_LIBS@ 
@PTHREAD_LIBS@
 endif
 
 if V8
-unit_test_pacrunner_SOURCES += plugins/v8.cc
-
-unit_test_pacrunner_LDADD += @V8_LIBS@
+unit_test_pacrunner_LDADD += plugins/libv8plugin.a @V8_LIBS@
 endif
 
 if DUKTAPE
 unit_test_pacrunner_SOURCES += plugins/duktape.c
 
-unit_test_pacrunner_LDADD += duktape/libduktape.a
+unit_test_pacrunner_LDADD += duktape/libduktape.a -lm
 endif
 
 endif #UNIT
@@ -160,7 +167,7 @@ AM_CFLAGS = @GLIB_CFLAGS@ @PTHREAD_LIBS@ @DBUS_CFLAGS@ \
                                -DPACRUNNER_PLUGIN_BUILTIN \
                                -DPLUGINDIR=\""$(plugindir)"\"
 
-AM_CXXFLAGS = $(AM_CFLAGS)
+AM_CXXFLAGS = $(AM_CFLAGS) -fno-exceptions
 
 AM_CPPFLAGS = -I$(builddir)/src -I$(srcdir)/src -I$(srcdir)/gdbus
 
-- 
2.7.4



------------------------------

Subject: Digest Footer

_______________________________________________
connman mailing list
[email protected]
https://lists.01.org/mailman/listinfo/connman


------------------------------

End of connman Digest, Vol 10, Issue 23
***************************************

Reply via email to