sim: Ensure to call sim_pin_check

In current code, sim_pin_check() is called inside sim_efpl_read_cb(). However, there may be a chance it would never be called, thus the

---
 src/sim.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

src/sim.c +0 -0

index a450b30..9bc9906 100644

src/sim.c +2 -1

 	ofono_sim_read(sim, SIM_EFPL_FILEID,
 			OFONO_SIM_FILE_STRUCTURE_TRANSPARENT,
 			sim_efpl_read_cb, sim);
+
+	sim_pin_check(sim);
 }
 
 static void sim_op_error(struct ofono_sim *sim)
--

1.7.0.4

I believe the right place to call the sim_pin_check() will be before the if condition(if (sim->language_prefs == NULL)) in sim_efpl_read_cb(), like shown below

     if (efpl) {
               g_slist_foreach(efpl, (GFunc)g_free, NULL); g_slist_free(efpl);
       }

       sim_pin_check(sim);

     if (sim->language_prefs == NULL)
           return;

     ofono_dbus_signal_array_property_changed(conn, path,

OFONO_SIM_MANAGER_INTERFACE,
                                              "PreferredLanguages", DBUS_TYPE_STRING, &sim->language_prefs);

Let me know your views on this.

Quoted 1 line

static void sim_iccid_read_cb(int ok, int length, int record, @@

-1454,6 +1452,8 @@ static void sim_initialize(struct ofono_sim *sim)

Quoted 10 lines

src/sim.c +2 -1

 	ofono_sim_read(sim, SIM_EFPL_FILEID,
 			OFONO_SIM_FILE_STRUCTURE_TRANSPARENT,
 			sim_efpl_read_cb, sim);
+
+	sim_pin_check(sim);
 }
 
 static void sim_op_error(struct ofono_sim *sim)
--

1.7.0.4

As per the 3GPP 31.102 section 5.1.1.2, CHV1 verification procedure will be done after reading of EFli and EFpl. Earlier( before "sim: Reorder SIM initialization" patch) src/sim.c had the sim_pin_check()called from sim_efphase_read_cb() where it called after issuing EFli and EFpl file read request. The patch "sim: Reorder SIM initialization" changed it to call from sim_efpl_read_cb() which is as per the specification.If the reading of EFli and EFpl fails, then sim_pin_check is not getting called because we are returning from "if (sim->language_prefs == NULL)". In your patch, you are moving it to be called after the EFpl read request which is same as what we had before "sim: Reorder SIM initialization" patch. I believe the right place to call the sim_pin_check() will be before the if condition(if (sim->language_prefs == NULL)) in sim_efpl_read_cb(), like shown below

     if (efpl) {
		g_slist_foreach(efpl, (GFunc)g_free, NULL);
	      g_slist_free(efpl);
	}

	sim_pin_check(sim);

     if (sim->language_prefs == NULL)
           return;

     ofono_dbus_signal_array_property_changed(conn, path,

OFONO_SIM_MANAGER_INTERFACE,

                                              "PreferredLanguages",
                                              DBUS_TYPE_STRING,
                                              &sim->language_prefs);

Let me know your views on this.

Your solution is better than mine, and might be Denis' intention of modification. Honestly, I ever thought about the change like your proposal (Actually it was the same as ppessi's patch), but I had a concern on current code. Now we read EFli and EFpl in parallel. Is it guaranteed the callback of EFli will be called before the callback of EFpl? IMO, it can't always guarantee this. If I'm correct, your solution is just half way to the final target. Maybe we need to work out a better one:)

-----Original Message-----
From: Denis Kenzior [mailto:denkenz(a)gmail.com]
Sent: Wednesday, August 25, 2010 11:07 PM
To: ofono(a)ofono.org
Cc: Gu, Yang
Subject: Re: [PATCH] sim: Ensure to call sim_pin_check

Hi Yang,

Quoted 1 line

Your solution is better than mine, and might be Denis' intention of modification.

Honestly, I ever thought about the change like your proposal (Actually it was the same as ppessi's patch), but I had a concern on current code. Now we read EFli and EFpl in parallel. Is it guaranteed the callback of EFli will be called before the callback of EFpl? IMO, it can't always guarantee this. If I'm correct, your solution is just half way to the final target. Maybe we need to work out a better one:)

Please note that the ofono_sim_read functions do not work in parallel. All file operations are added to a queue and executed sequentially. Hence we can rely on a certain order here.