G1: Add an SMS quirk for CNMI mode
PATCH 1/1 G1: Add an SMS quirk for CNMI mode
From 853ec0707530ee9c35a41bea3f8a6cfd067c7317 Mon Sep 17 00:00:00 2001
From: Andres Salomon <[email protected]>
Date: Wed, 2 Sep 2009 19:35:40 -0400
Subject: [PATCH] G1: Add an SMS quirk for CNMI mode
The G1 doesn't support mode2, despite advertising it.
The G1 chokes w/ an "Error 303" when we specify NMI mode 2. Adding a quirk to drop that mode from the supported list (just use mode 1) allows the G1 to properly deal with SMS.
--- drivers/atmodem/sms.c | 19 ++++++++++++++----- drivers/atmodem/vendor.h | 1 + plugins/g1.c | 3 +++ 3 files changed, 18 insertions(+), 5 deletions(-)
drivers/atmodem/sms.c +14 -5
@@ -35,6 +35,7 @@#include <ofono/sms.h> #include "smsutil.h" #include "util.h" +#include "vendor.h" #include "gatchat.h" #include "gatresult.h"
@@ -570,17 +571,25 @@ static inline gboolean append_cnmi_element(char *buf, int *len, int cap, }static gboolean build_cnmi_string (char *buf,int *cnmi_opts, -gboolean cnma_enabled) +struct sms_data *data) { +const char *mode;int len =sprintf (buf,"AT+CNMI=" ); -/* Mode doesn't matter, but sounds like 2 is the sanest option */ -if (!append_cnmi_element (buf, &len, cnmi_opts[0 ],"2310" , FALSE)) +if (data->vendor ==OFONO_VENDOR_HTC_G1 ) +/* The G1 advertises support for mode 2, but returns an error +* if we attempt to actually use it. */ + mode ="1" ; +else +/* Sounds like 2 is the sanest mode */ + mode ="2310" ; + +if (!append_cnmi_element (buf, &len, cnmi_opts[0 ], mode, FALSE))return FALSE;/* Prefer to deliver SMS via +CMT if CNMA is supported */ if (!append_cnmi_element (buf, &len, cnmi_opts[1 ], - cnma_enabled ?"21" :"1" , FALSE)) + data->cnma_enabled ?"21" :"1" , FALSE))return FALSE;/* Always deliver CB via +CBM, otherwise don't deliver at all */
@@ -666,7 +675,7 @@ static void at_cnmi_query_cb(gboolean ok, GAtResult *result, gpointer user_data) goto out; } -if (build_cnmi_string (buf, cnmi_opts, data->cnma_enabled)) +if (build_cnmi_string (buf, cnmi_opts, data)) supported = TRUE;if (data->cnma_enabled)
drivers/atmodem/vendor.h +1 -0
@@ -21,4 +21,5 @@enum ofono_vendor {OFONO_VENDOR_GENERIC =0 , +OFONO_VENDOR_HTC_G1 =1 , };
plugins/g1.c +3 -0
@@ -47,6 +47,8 @@#include <ofono/ussd.h> #include <ofono/voicecall.h> +#include <drivers/atmodem/vendor.h> +/* Supply our own syntax parser */ enum G1_STATE_ {
@@ -234,6 +236,7 @@ static void g1_populate(struct ofono_modem *modem)ofono_call_meter_create (modem,0 ,"atmodem" , chat);ofono_call_barring_create (modem,0 ,"atmodem" , chat);ofono_ssn_create (modem,0 ,"atmodem" , chat); +ofono_sms_create (modem,OFONO_VENDOR_HTC_G1 ,"atmodem" , chat);ofono_phonebook_create (modem,0 ,"atmodem" , chat); mw =ofono_message_waiting_create (modem);
Signature · 1 line
1.6.3.3
Hi Andres,
Quoted 9 lines
The G1 doesn't support mode2, despite advertising it.
The G1 chokes w/ an "Error 303" when we specify NMI mode 2. Adding a quirk to drop that mode from the supported list (just use mode 1) allows the G1 to properly deal with SMS.--- drivers/atmodem/sms.c | 19 ++++++++++++++----- drivers/atmodem/vendor.h | 1 + plugins/g1.c | 3 +++ 3 files changed, 18 insertions(+), 5 deletions(-)
patch has been applied. Thanks.
Regards
Marcel
Bastian
drivers/atmodem/sms.c static inline gboolean append_cnmi_element(char *buf, int *len, int cap, · line 588
Quoted 12 lines
drivers/atmodem/sms.c +9
-2-/* Mode doesn't matter, but sounds like 2 is the sanest option */ -if (!append_cnmi_element (buf, &len, cnmi_opts[0 ],"2310" , FALSE)) +if (data->vendor ==OFONO_VENDOR_HTC_G1 ) +/* The G1 advertises support for mode 2, but returns an error +* if we attempt to actually use it. */ + mode ="1" ; +else +/* Sounds like 2 is the sanest mode */ + mode ="2310" ; + +if (!append_cnmi_element (buf, &len, cnmi_opts[0 ], mode, FALSE))return FALSE;
Instead of having vendor checks all over the place, wouldn't it be cleaner if the plugin had a way to provide/fix cnmi_opts?
Cheers,
Waldo
Hi Waldo,
Quoted 2 lines
Instead of having vendor checks all over the place, wouldn't it be cleaner if the plugin had a way to provide/fix
cnmi_opts?
For now we decided to do it this way, since providing arbitrary properties per atom seemed to be too much. The hope is that there are not many modems that blatantly lie about what they can do (like the G1) If quirks get out of hand we can revisit this.
In the meantime if you need to override the entire cnmi string, do something like:
if (data->vendor == OFONO_VENDOR_WALDOS_MODEM) {
sprintf(buf, "AT+CNMI=....");
supported = TRUE;
} else if (build_cnmi_string(buf, cnmi_opts, data))
supported = TRUE;Regards,
-Denis