UPDATE: Some information has been updated below as we learn more.

In response to Rogers’ admission that enabling GPS on the HTC Magic & Dream can stop 911 calls from completing properly, a member of the Howard Forums posted on 01/20/2010 that he contacted Rogers technical support on September 22, 2009 to warn them of the 911 issue, and his concern that it possibly would affect all HTC Magic phones.

Canadiancow appears to be a cellphone software developer, and posted an .mp3 file of what is apparently his call to Rogers. According to Canadiancow, Google released a fix for this issue on 09/17/2009. You can hear the full .mp3 recording of the call here.

The tech on the phone call claims to test the issue with a Magic at his end, and says he is able to get through to 911 with no problem. The two then debate which version of the OS build is being used. Whether or not the tech had GPS enabled when he made the 911 test call is not clarified, but Canadiancow informs the tech that Google has released a fix for this specific issue. In the end, the tech says he is hesitant to keep calling 911 to test the phone, and Canadiancow relents and says goodbye.

If this is true that Rogers was informed of this in September of 2009,  but nothing was done … that’s pretty damning of Rogers and HTC, who claim to be concerned for our safety. Imagine it is possible Rogers & HTC knew for 3.5 months that calls to 911 could crash with GPS on and didn’t do anything about it!

Apparently T-Mobile in the USA mentioned a 911 issue as one of the fixes rolled out with Android 1.6! The question now appears to be whether or not this is the same 911 issue.  In the T-Mobile forums, Community Manager, Will, posted the following on October 1st 2009 (part of it bolded by me):

The rollout of Android 1.6 (Donut) to T-Mobile’s G1 and myTouch 3G customers is now underway, and provides new features and software enhancements. The update also includes an important fix for isolated instances of system reboots after dialing 911. We have worked with Google and HTC on a solution, and the rollout of Android 1.6 is being accelerated to ensure the system reboot fix is rapidly delivered to our customers.

If this is the same 911 issue, and If HTC knew about it then, why didn’t they do anything about it up here? Did they tell Rogers? If they did, why didn’t we get 1.6 if this 911 issue was fixed in 1.6?

This may be about the new “enhanced 911″ which apparently only began recently in Canada? The CRTC mandated it to be up and running by Feb 1, 2010. When did Rogers upgrade to Enhanced 911? If the issue was known about before enhanced 911 rolled out here, why didn’t they have the fix ready for the rollout? And as some have suggested, did it to occur to anyone that Canadians with these phones might travel to an area with enhanced 911?

Let’s review… Rogers and HTC not only failed to release Android 1.6 (which had been available since Oct 1 2009), but techs, CSRs, and representatives of the Office of the President didn’t seem to know an update was available until we started complaining about it. After 2.5 months of wrangling with Rogers and HTC, HTC first announced there were NO plans to issue an update because they felt 1.5 was a “stable & reliable” platform. Their words! They also told us that the carrier, Rogers, would have to request an update. Rogers claims they requested an update in response to our complaints, but that HTC wouldn’t provide one. It wasn’t until we stepped up the heat across the net that we were offered an update on the Magic to 2.1 “by mid-year” and a free hardware upgrade for Dream owners to the Magic.

I’d like to know, if this is an Android software issue, does it also affect the LG Eve which is still on Android 1.5 as well? UPDATE: Great Big Dog reports via twitter that according to LG Rep Sheryl Davey this is a “Carrier” issue. Stil not sure if that means it does or does not affect the LG Eve?

Who knew what, and when?

We’ll be waiting for some answers from @RogersMary, but don’t expect Rogers to admit to anything. Now that Canadiancow has released the recording of the call, if it can be verified as true it might actually be something the media might want to investigate.

You can read the whole Howard Forums thread here: http://www.howardforums.com/showthread.php?t=1612180&page=1&pp=15

Thanks to Puleen Patel for the tip about the Howard Forums thread.

UPDATE (12:30pm): I contacted Canadiancow to ask him about his claim that Google released a patch for this 911 issue in Sept of 2009. In his original forum post, he linked to this page at the Android Open Source Project that highlights the code and the changes. Canadiancow replied with an explanation of what the code on that page means, and why it should be considered a “Google” fix. Here is his email:

That’s one change to the Android source code that was incorporated into Android on September 17 2009.  All changes to Android are submitted through patches, which can be small like this one, or large like a complete overhaul to the system.

If you look at that patch in particular, you can see the Author is someone named Mike Lockwood, but it was committed (or incorporated into AOSP) by The Android Open Source Project, which means someone at Google reviewed and accepted the patch.

The patch comment is:

AI 151388: Fix null pointer exceptions that can occur if the GPS is started due to E911

This is exactly the issue that’s affecting Rogers phones.

Below that, you can see “services/java/com/android/server/LocationManagerService.java”.  This is the source code file in Android that controls all location services.  If you click the “diff” link beside it, you can see exactly what changed in this patch.

The red code was removed, and the green code was added.

What I’m about the say is going to be very technical, and if you aren’t a developer, you probably won’t understand it, but in case you are:

There were only two changes.  In the first change, a check was added around “records.remove(this);” to ensure that “records” is not null.  If you try to execute “records.remove(this)” on “records” when it’s null, the code will crash.  This will have various consequences based on what code is crashing.  In this case, it’s a system service, and it causes the whole system to crash.  The second change was almost identical, except they added the check to ensure “urs” is not null before using it.

In other words, if the system tried to apply the “records.remove(this)” part of the code to “records” while “records” had no value, it would crash.

When I asked him whether or not this was a “carrier” issue, as claimed by LG, and whether or not this was the same issue T-Mobile fixed when they released Android 1.6, he replied:

It’s not a carrier issue.  It’s a bug in the Android source code.  Whether it affects the Eve is unknown.  I’m hoping someone who has one can try to make a test call with GPS enabled to see what happens.

So yes, it’s the same issue that affected the T-Mobile G1.  And yes, it was fixed it 1.6/Donut. But I could make a 1.5 ROM today that includes the fix, so I don’t know if other 1.5 devices are affected.

We would like to remind people that it is NOT a good idea to call 911 to “test” your phone. My brother is a 911 operator, and believe me, they don’t have the time for “test” calls. Hopefully Rogers & HTC fix this ASAP!

I’ve emailed Telus to ask if the same bug affects the HTC Hero. Waiting for a reply.