Delay when answering/placing calls post 4.3

It's official, my phone absolutely hates the MK3 update.

Turned out CNexus' suggestion of disabling noise filtering made things bearable for me. The issue is still there but not anywhere near as noticeable. Had it not been worse before, I might not have noticed how it is now.

Yea, I tried everything and it made no difference.

My wife took the OTA on her s3 and she doesn't have any issues. I thought about going stock and taking the OTA, or using Odin to flash it but I can't bring myself to 'knox' my phone.

Has anyone here tried the MK5 modem jdsingle76 posted ? I have read and heard that even though its from VM/Boost that it also works for Sprint - mite be an option ???

http://xdaforums.com/showthread.php?p=47904496

From my GSIII - jd's Stock/Rooted/Deodexed 4.3

TL;DR: I'm working on fixing this. The work I've done recently is rolled up in this kernel, and it sounds like it's helping with at least one person's issues.

I suspect this is a collection of major kernel bugs caused by the new Sprint a2220 (voice processor) firmware (or lack thereof). As far as I can tell, the problem basically boils down to this: the dialer is trying to configure the a2220 incorrectly and causes a bunch of errors; instead of recovering gracefully, the driver completely resets the chip, which typically takes between 5 and 15 seconds. The dialer (and the person at the other end of the call) have to wait for the chip to reboot before the call can properly begin.

There's at least two issues at play here: kernels don't pull in the correct firmware and the a2220 driver absolutely sucks. I fixed the firmware issue a while back (this commit), and I'm currently working on completely rewriting the driver (it's beyond saving) to handle error conditions better and not cause them as often.

If anyone is feeling adventurous, I'd appreciate help testing the new driver. Any feedback (helps? doesn't? crashes?) would be helpful.

EDIT: There's a new, better-behaved build here.
EDIT 2: My changes are merged into my kernel, available here.

So what you're basically saying is that Samsung screwed the pooch?

Just wanted to put this out there - I am running this new kernel by @decimalman and so far so good - it has drastically improved the dialer lag and I have had much better sound quality during calls with noise reduction active - I'm on a VM S3 running the MK5 build of 4.3 TW by @jdsingle76 -

From my GSIII - jd's Stock/Rooted/Deodexed 4.3

metalfan78 said:

So what you're basically saying is that Samsung screwed the pooch?

Click to expand...

Click to collapse

Not at all. I'm saying that Samsung is playing fast and loose with the GPL. The MK3 kernel source contains any number of tricks to keep it from building correctly unless Samsung is building it themselves. Without patching it, exFAT support is (silently) skipped, the a2220 driver uses AT&T's firmware instead of Sprint's, and there's likely a few more bugs affecting the a2220 and video drivers (and possibly others as well).

EDIT: To clarify, some of the source code uses magic #define macros and directory layouts in order to change how it's built. Without being built by Samsung's build tree, that magic is missing and the resulting kernel isn't functionally equivalent to the kernel Samsung provided. GPLv2 doesn't explicitly forbid this, so they're technically compliant.

Interesting

Yup, what he said could not be more spot on...


For example, take a look at my commit here: https://github.com/cnexus/kernel_tw_43/commit/75f9e924c28a9d88cfe9b9c9b477c0c68621a87a

That line that I removed was originally only pulling in exfat if some folder way up in Samsung build tree exists. And so, since we don't have the same build setup as Samsung, exfat gets skipped unless you remove that