Re: [RFC] [PATCH] ConnMan backtrace support

Samuel Ortiz writes: > Add a generic signal handler in order to dump ConnMan backtrace when > crashing. This is really nice. Makes my life easier, thanks :) The backtrace is dumped to the syslog, right? That way it's easy to access. -- Kalle Valo __

Re: [RFC] [PATCH] ConnMan backtrace support

Hi Samuel, > Add a generic signal handler in order to dump ConnMan backtrace when > crashing. The implementation is based on glibc backtrace() routines and > thus can not resolve static function names. A little python wrapper over > addr2line fixes that by generating a full backtrace from a comple

Re: [RFC] [PATCH] ConnMan backtrace support

Hi Samuel, > > +static void signal_handler(int signo) > > +{ > > + switch (signo) { > > + case SIGSEGV: > > + case SIGBUS: > > + case SIGILL: > > + case SIGABRT: > > + connman_info("Aborting (signal %d)", signo); > > + __connman_log_backtrace(); > > + exit(1

Roaming statistics

Hi, here is another round of this patch. I hope I have addressed all Samual's remark. The first patch fixes only the simple, easy to fix things. The second one is still WIP. Also I plan to fix the update trigger, so that counters are only updated as requested. No code yet. v4: - Remove counter

[PATCH v4 1/2] Add support for roaming/home network statistic

From: Daniel Wagner Currently connman only has one set of counters for collecting statistics on online time and the amount of transfered bytes. For 3G connections we should destinguish between home network and roaming. This patch introduces two sets of counter values for home network and roaming

[PATCH v4 2/2] Handle several counters simultaniously

From: Daniel Wagner The current implementation does not handle multiple counters at the same time correctly. Only the first counter gets the updates, all others just an empty dictionary. For each counter the *_update counters have to be kept separate. counter.c allocs and frees these data struct

[PATCH] Support for EAP-TTLS.

EAP-TTLS and EAP-PEAP are similar in terms of options to pass to wpa_supplicant. So in case of TTLS, the PEAP path is used. --- doc/config-format.txt | 16 plugins/supplicant.c |3 ++- 2 files changed, 14 insertions(+), 5 deletions(-) diff --git a/doc/config-format.txt b/d

Re: [PATCH] Support for EAP-TTLS.

Hi Fabien, > EAP-TTLS and EAP-PEAP are similar in terms of options to pass to > wpa_supplicant. So in case of TTLS, the PEAP path is used. > --- > doc/config-format.txt | 16 > plugins/supplicant.c |3 ++- > 2 files changed, 14 insertions(+), 5 deletions(-) > > diff --git

[PATCH 1/1] Support for EAP-TTLS.

EAP-TTLS and EAP-PEAP are similar in terms of options to pass to wpa_supplicant. So in case of TTLS, the PEAP path is used. --- doc/config-format.txt | 16 plugins/supplicant.c |3 ++- 2 files changed, 14 insertions(+), 5 deletions(-) diff --git a/doc/config-format.txt b/d

Re: [PATCH 1/1] Support for EAP-TTLS.

Hi Fabien, > EAP-TTLS and EAP-PEAP are similar in terms of options to pass to > wpa_supplicant. So in case of TTLS, the PEAP path is used. > --- > doc/config-format.txt | 16 > plugins/supplicant.c |3 ++- > 2 files changed, 14 insertions(+), 5 deletions(-) patch has been

[RFC] [PATCH v2] ConnMan backtrace support

Add a generic signal handler in order to dump ConnMan backtrace when crashing. The implementation is based on glibc backtrace() routines and thus can not resolve static function names. A little python wrapper over addr2line fixes that by taking a full backtrace from a complete connman log file. --

Re: [RFC] [PATCH v2] ConnMan backtrace support

Hi Samuel, > Add a generic signal handler in order to dump ConnMan backtrace when > crashing. The implementation is based on glibc backtrace() routines and > thus can not resolve static function names. A little python wrapper over > addr2line fixes that by taking a full backtrace from a complete c

Re: [PATCH 3/6] Avoid races between Modem PropertyChanged/GetProperties

Hi Pekka, On Mon, Aug 16, 2010 at 02:40:59PM +0300, Pekka Pessi wrote: > Hi Samuel, > > 2010/8/14 Samuel Ortiz : > >> Create modem only once GetProperties succeed, thus avoiding races with > >> PropertyChanged. > > So you mean races between the ofono_connect() and the manager_watch() code > > pat

Re: [RFC] [PATCH v2] ConnMan backtrace support

Hi Marcel, On Tue, Aug 17, 2010 at 04:34:06PM +0200, Marcel Holtmann wrote: > > +static void backtrace_signal_handler(int signo) > > +{ > > + void *frames[64]; > > + char **symbols; > > + size_t n_ptrs; > > + unsigned int i; > > + > > + switch (signo) { > > + case SIGSEGV: > > + case

[PATCH v3] ConnMan backtrace support

Add a generic signal handler in order to dump ConnMan backtrace when crashing. The implementation is based on glibc backtrace() routines and thus can not resolve static function names. A little python wrapper over addr2line fixes that by taking a full backtrace from a complete connman log file. --

[PATCH 1/3] Use memcpy with dhcp unaligned access

From: Pekka Pessi --- gdhcp/client.c | 14 +++--- gdhcp/common.c |2 +- gdhcp/common.h | 17 +++-- 3 files changed, 15 insertions(+), 18 deletions(-) diff --git a/gdhcp/client.c b/gdhcp/client.c index a7bd88a..ba43b85 100644 --- a/gdhcp/client.c +++ b/gdhcp/client.c

[PATCH 2/3] Support ancient shells, do not use VAR+=VALUE

From: Pekka Pessi --- acinclude.m4 | 21 ++--- 1 files changed, 14 insertions(+), 7 deletions(-) diff --git a/acinclude.m4 b/acinclude.m4 index 84086fd..7e0cdc0 100644 --- a/acinclude.m4 +++ b/acinclude.m4 @@ -15,13 +15,20 @@ AC_DEFUN([COMPILER_FLAGS], [ CFLAGS

[PATCH 3/3] Always check for ns_initparse

From: Pekka Pessi ns_initparse is used by gresolv, too. --- configure.ac | 12 ++-- 1 files changed, 6 insertions(+), 6 deletions(-) diff --git a/configure.ac b/configure.ac index eeda72f..83aa29e 100644 --- a/configure.ac +++ b/configure.ac @@ -154,12 +154,7 @@ AM_CONDITIONAL(LOOPBAC

Re: [PATCH 2/3] Support ancient shells, do not use VAR+=VALUE

Hi Pekka, > acinclude.m4 | 21 ++--- > 1 files changed, 14 insertions(+), 7 deletions(-) > > diff --git a/acinclude.m4 b/acinclude.m4 > index 84086fd..7e0cdc0 100644 > --- a/acinclude.m4 > +++ b/acinclude.m4 > @@ -15,13 +15,20 @@ AC_DEFUN([COMPILER_FLAGS], [ > CFL

Re: [PATCH 1/3] Use memcpy with dhcp unaligned access

Hi Pekka, > --- > gdhcp/client.c | 14 +++--- > gdhcp/common.c |2 +- > gdhcp/common.h | 17 +++-- > 3 files changed, 15 insertions(+), 18 deletions(-) > > diff --git a/gdhcp/client.c b/gdhcp/client.c > index a7bd88a..ba43b85 100644 > --- a/gdhcp/client.c > +++ b/gdhc

Re: [PATCH 3/3] Always check for ns_initparse

Hi Pekka, > ns_initparse is used by gresolv, too. > --- > configure.ac | 12 ++-- > 1 files changed, 6 insertions(+), 6 deletions(-) patch has been applied. Thanks. Regards Marcel ___ connman mailing list connman@connman.net http://lists.

Re: [ofono-online-v2 1/6] Refactor Ofono SetProperty D-Bus calls.

Hi Pekka, On Mon, Aug 16, 2010 at 05:40:56PM +0300, pekka.pe...@nokia.com wrote: > From: Pekka Pessi Some nitpicks: First of all, fix your subjec to oFono, not Ofono, please. Then see below: > --- > plugins/ofono.c | 181 > +++ > 1 files c

Re: [PATCH 1/3] Use memcpy with dhcp unaligned access

Hi Marcel, > I really need an explanation on why. We are using exactly the same in > BlueZ and it works just fine there. As I mentioned on IRC, these are GCC > guru approved ways of doing unaligned access. Your gcc guru is absolutely right, there seems to be no problems with arm7v compiler and N9

Re: [ofono-online-v2 2/6] Log more of the ofono errors.

Hi Pekka, On Mon, Aug 16, 2010 at 05:40:57PM +0300, pekka.pe...@nokia.com wrote: > From: Pekka Pessi > > --- > plugins/ofono.c | 34 +- > 1 files changed, 29 insertions(+), 5 deletions(-) > > diff --git a/plugins/ofono.c b/plugins/ofono.c > index 8254b56..b15e

Re: [ofono-online-v2 5/6] Create device once IMSI is available

Hi Pekka, On Mon, Aug 16, 2010 at 05:41:00PM +0300, pekka.pe...@nokia.com wrote: > From: Pekka Pessi > > Try to get IMSI immediately after SimManager interface is available. > > Listen to SIM events in Ofono plugin, add device when IMSI I suppose you want to do that in order for ConnMan to be a

Re: [ofono-online-v2 4/6] Avoid race between ofono signals and methods.

Hi Pekka, On Mon, Aug 16, 2010 at 05:40:59PM +0300, pekka.pe...@nokia.com wrote: > From: Pekka Pessi > > Create modem only after Modem.GetProperties succeeds, thus avoiding races > with Modem.PropertyChanged signal. So I've commented on this one and my concernas about the Powered property not be

Re: [ofono-online-v2 6/6] Control modem "Online" property.

Hi Pekka, On Mon, Aug 16, 2010 at 05:41:01PM +0300, pekka.pe...@nokia.com wrote: > From: Pekka Pessi > > The Modem Online property is tied with the cellular device Powered property. Here again, I would appreciate more detailed explanations as to why we want to use the Online property instead of