05-10-2012 04:01 PM
937 ... 3334353637 ...
tools
  1. tdm's Avatar
    tomorrow? i'm starting right now!

    edit: made a diff, it applied almost completely clean to my kernel. only things that didnt are ami304.c and .h

    edit 2: compiling! thanks to leslieanne's build server, it should be done soon

    edit 3: uploading...

    edit 4: tvall_kernel_v0.3.1.zip

    edit 5: ami304.c and .h didnt apply because of a commit with the message "Update to Sprint's/Zefie's AMI304 driver." which.. already patched them
    You gotta be careful with the AMI304 driver. They changed the ioctl api in a driver update, so the kernel driver and userspace need to match. I think that update was done with the Sprint GB release (but I'm not 100% sure).
    05-04-2012 09:48 AM
  2. tdm's Avatar
    Dechutes makes some great beers. If your ever back in the tri cities you have to let me buy you a beer at ice harbor brewery best beer ever. OK sorry to get off topic... so umm... how about that radio update!

    Sent from my Galaxy Nexus using Tapatalk 2 Beta-5
    I think I'll be down there week after next, actually. I'll PM you when I find out for sure.
    05-04-2012 09:49 AM
  3. mrg666's Avatar
    edit: made a diff, it applied almost completely clean to my kernel. only things that didnt are ami304.c and .h
    If there isn't any difference in the code, what kind of an improvement we are expecting from this kernel source?

    [HL]Edit for clarification for the others reading the thread: There is a difference, apparently, as TDM's diff shows later in the thread.[/HL]
    05-04-2012 10:11 AM
  4. cole2kb's Avatar
    If there isn't any difference in the code, what kind of an improvement we are expecting from this kernel source?
    ...and why is it breaking IHO for some of us, and the only fix being a different kernel?

    Come to think of it, I'm moreso just asking for curiosity's sake now...I think I'll be sticking with ICS.
    05-04-2012 10:39 AM
  5. mrg666's Avatar
    ...and why is it breaking IHO for some of us, and the only fix being a different kernel?
    I was referring to the recent ZV9 kernel source from LG. The different kernels that you have seen improvement with are 2.6.35 Gingerbread kernel. This LG source is still the same 2.6.32.9 Froyo source we have been using with IHO.

    In case there is a problem with the v3.7 kernel binary, I am testing the kernel binary which is built on my machine using the source from Bob's repo. I would appreciate if anybody can test and give feedback. I think I will use these binaries in the next MiRaGe build.

    http://www.mediafire.com/?9ogmh5j5hqswxa6
    MD5sum: E69FAF8FA222D06ACB4116C64DDCB893

    PS: kernel binary updated with v3.7 source to be consistent and used gcc 4.7 instead of gcc 4.5

    [HL]Edit: TDM's diff was done properly after tvall's diff and the differences are shown below.[/HL]
    cole2kb and Dannemand like this.
    05-04-2012 10:54 AM
  6. brotherswing's Avatar
    I'll test it for you. Would you prefer I report back here, another thread, or a PM?

    Sent from the BACKside of Tapatalk 2
    mrg666 likes this.
    05-04-2012 10:59 AM
  7. mrg666's Avatar
    I'll test it for you. Would you prefer I report back here, another thread, or a PM?

    Sent from the BACKside of Tapatalk 2
    You can post here or PM me. Whichever is easier for you. Thanks!
    05-04-2012 11:11 AM
  8. Dannemand's Avatar
    @mrg: thanks a lot. I'll test this later today. Do you have an MD5 for it? Small risk, I know. Just don't want to think I'm testing a new kernel when I'm actually testing a bad download

    Question: did you build it for a specific MiRaGe version?
    brotherswing likes this.
    05-04-2012 12:39 PM
  9. thewraith420's Avatar
    Looks like a flashable zip to me. Should work on any IHO build...
    Correct me if I'm wrong

    Sent from my LG-VM670 using Tapatalk 2
    05-04-2012 01:37 PM
  10. tdm's Avatar
    Here is the ZV9 kernel diff. 25kb instead of 125mb.
    mrg666 likes this.
    05-04-2012 01:42 PM
  11. mrg666's Avatar
    @mrg: thanks a lot. I'll test this later today. Do you have an MD5 for it? Small risk, I know. Just don't want to think I'm testing a new kernel when I'm actually testing a bad download

    Question: did you build it for a specific MiRaGe version?
    It can be used with any IHO build. Also I have added the MD5sum and updated the binary. So you need to download again.
    Dannemand likes this.
    05-04-2012 01:50 PM
  12. mrg666's Avatar
    Here is the ZV9 kernel diff. 25kb instead of 125mb.
    Thanks! There is more difference than just one file as I see. I will start adding the differences one file at a time and testing at each step.
    Dannemand likes this.
    05-04-2012 01:57 PM
  13. tdm's Avatar
    Thanks! There is more difference than just one file as I see. I will start adding the differences one file at a time and testing at each step.
    Looks like the differences are:

    * New screen support.
    * Camera changes.
    * Diagnostics/testmode changes.
    * AMI304 driver update.
    * RTC alarm changes.

    If I had to guess, I'd say the RTC alarm changes are the most likely to affect stability. These would be:

    drivers/rtc/alarm.c
    drivers/rtc/hctosys.c
    (note drivers/rtc/rtc-lib.c is just whitespace)
    mrg666, Dannemand and brotherswing like this.
    05-04-2012 02:16 PM
  14. mrg666's Avatar
    Looks like the differences are:

    * New screen support.
    * Camera changes.
    * Diagnostics/testmode changes.
    * AMI304 driver update.
    * RTC alarm changes.

    If I had to guess, I'd say the RTC alarm changes are the most likely to affect stability. These would be:

    drivers/rtc/alarm.c
    drivers/rtc/hctosys.c
    (note drivers/rtc/rtc-lib.c is just whitespace)
    We already have the updated AMI304 in IHO kernel. I have checked in the repo and it is the same with ZV9 source; Blarf updated it.
    05-04-2012 02:22 PM
  15. tvall's Avatar
    So, wait, you're saying nothing has changed?

    Sent from my VM670 using Tapatalk 2
    No, I'm saying everything that changed patched in without issue except the ami304 files (because they were already updated). Tdm has a nice list of changes, and everything in that list has been patched into the kernel I built.

    shouldn't have slept till one, I wanted to be part of that conversation... that's what I get for staying up till 1 examining this
    edit: wait... i got like 11 or 12 hours of sleep... why am I so sleepy?
    cole2kb likes this.
    05-04-2012 02:37 PM
  16. cole2kb's Avatar
    Ahhh, I see. Thanks for the clarification, friend.

    Sent from my VM670 using Tapatalk 2
    05-04-2012 02:43 PM
  17. mrg666's Avatar
    Looks like the differences are:

    * New screen support.
    * Camera changes.
    * Diagnostics/testmode changes.
    * AMI304 driver update.
    * RTC alarm changes.

    If I had to guess, I'd say the RTC alarm changes are the most likely to affect stability. These would be:

    drivers/rtc/alarm.c
    drivers/rtc/hctosys.c
    (note drivers/rtc/rtc-lib.c is just whitespace)
    The 3.7 kernel with only rtc and diagnostic changes from ZV9 source added is below.
    Mirage_BobZ37_kernel_rtc_diag.zip
    MD5sum: 2063A60A037D850D04416BF655B60DD3
    It would be great if you can test and post your feedback. I have flashed to my phone and it still seems to be working. I can't make it reboot yet.
    05-04-2012 02:49 PM
  18. Dannemand's Avatar
    In case there is a problem with the v3.7 kernel binary, I am testing the kernel binary which is built on my machine using the source from Bob's repo. I would appreciate if anybody can test and give feedback. I think I will use these binaries in the next MiRaGe build.

    Mirage-BobZ37-gcc47_kernel.zip
    MD5sum: E69FAF8FA222D06ACB4116C64DDCB893

    PS: kernel binary updated with v3.7 source to be consistent and used gcc 4.7 instead of gcc 4.5
    I flashed this one onto my MiRaGe 04/27 (which I updated to yesterday). Boots fine, everything runs fine. No OC yet, wanted to test first.

    Running my repro test (15 sec airplane mode, SMS to my OV number, SMS to my GV number which forwards to my OV), it got a bit lagging after 2-3 cycles, but not so much that I would conclude anything.

    Then, lo and behold, it rebooted when I switched to airplane mode starting the 5th test cycle!

    As much as I wish it hadn't rebooted, I have to appreciate when it reproduces promptly instead of playing nice for 20-30 cycles, then rebooting on me after I've concluded that it's stable
    mrg666 and kwknott like this.
    05-04-2012 02:50 PM
  19. Dannemand's Avatar
    The 3.7 kernel with only rtc and diagnostic changes from ZV9 source added is below.
    Mirage_BobZ37_kernel_rtc_diag.zip
    MD5sum: 2063A60A037D850D04416BF655B60DD3
    It would be great if you can test and post your feedback. I have flashed to my phone and it still seems to be working. I can't make it reboot yet.
    Wow, you're cranking these out faster than I can test them! I'm on it...

    And thanks!
    mrg666 likes this.
    05-04-2012 02:52 PM
  20. mrg666's Avatar
    I flashed this one onto my MiRaGe 04/27 (which I updated to yesterday). Boots fine, everything runs fine. No OC yet, wanted to test first.

    Running my repro test (15 sec airplane mode, SMS to my OV number, SMS to my GV number which forwards to my OV), it got a bit lagging after 2-3 cycles, but not so much that I would conclude anything.

    Then, lo and behold, it rebooted when I switched to airplane mode starting the 5th test cycle!

    As much as I wish it hadn't rebooted, I have to appreciate when it reproduces promptly instead of playing nice for 20-30 cycles, then rebooting on me after I've concluded that it's stable
    Thanks, this eliminates the possibility of a problem with the binary itself. The next kernel you are now testing is with actual code change in RTC and Diagnostic parts from LG's ZV9 kernel source. Fingers crossed ...
    Dannemand, Dave Begue and kwknott like this.
    05-04-2012 03:28 PM
  21. cole2kb's Avatar
    Big props to you guys for doing this. I'm at work or I'd be happy to flash temporarily to test as well. Thanks again.

    Sent from my VM670 using Tapatalk 2
    Dannemand likes this.
    05-04-2012 03:32 PM
  22. Dannemand's Avatar
    Thanks, this eliminates the possibility of a problem with the binary itself. The next kernel you are now testing is with actual code change in RTC and Diagnostic parts from LG's ZV9 kernel source. Fingers crossed ...
    So far this one looks quite promising, my friend! Of course nothing conclusive until after a few days, but here's what I tried so far:

    Flashed over the last one (same MiRaGe 04/27 install that rebooted before). Triple-wiped cache & dalvik first (always do). Boots fine, runs fine, shows as 2.6.32.9-v3.7 mirage@aurora.

    After 12 test cycles (ie 12 airplane modes and 24 roundtrip SMSs) it is still going strong, no reboots. That in itself is not conclusive. But a good sign is that I didn't see any of the SMS sending lags -- those are normally indicative that something is not right.

    I DID get quite a bit of delivery backups (multiple messages arriving ketchup-style after a few mins) but that could easily be just VMU network.

    I ALSO had one message that failed to send -- another frequent problem on the old Bobz kernel that wasn't there on GingerKernel or ZV9 stock. But it was a message trying to send just as a whole slew of messages were coming in. And it send fine on first retry attempt. So it could be complete coincidence.

    Significantly, ALL test messages arrived, and none were massively delayed. And I just now exchanged SMSs with my wife -- the importance of that function not to be underestimated

    Although it's very subjective, I must say this has more of that snappier feeling on radio tasks that characterized GingerKernel and ZV9 stock. So I think you may be onto it.

    I'll keep you posted over the next couple days, of course.

    Regardless of the outcome of this particular kernel, thanks a lot to you, tdm, cole2kb, tvall and others who've remained focused on this issue!
    Dave Begue, mrg666 and kwknott like this.
    05-04-2012 03:50 PM
  23. mrg666's Avatar
    So far this one looks quite promising, my friend! Of course nothing conclusive until after a few days, but here's what I tried so far:

    Flashed over the last one (same MiRaGe 04/27 install that rebooted before). Triple-wiped cache & dalvik first (always do). Boots fine, runs fine, shows as 2.6.32.9-v3.7 mirage@aurora.

    After 12 test cycles (ie 12 airplane modes and 24 roundtrip SMSs) it is still going strong, no reboots. That in itself is not conclusive. But a good sign is that I didn't see any of the SMS sending lags -- those are normally indicative that something is not right.

    I DID get quite a bit of delivery backups (multiple messages arriving ketchup-style after a few mins) but that could easily be just VMU network.

    I ALSO had one message that failed to send -- another frequent problem on the old Bobz kernel that wasn't there on GingerKernel or ZV9 stock. But it was a message trying to send just as a whole slew of messages were coming in. And it send fine on first retry attempt. So it could be complete coincidence.

    Significantly, ALL test messages arrived, and none were massively delayed. And I just now exchanged SMSs with my wife -- the importance of that function not to be underestimated

    Although it's very subjective, I must say this has more of that snappier feeling on radio tasks that characterized GingerKernel and ZV9 stock. So I think you may be onto it.

    I'll keep you posted over the next couple days, of course.

    Regardless of the outcome of this particular kernel, thanks a lot to you, tdm, cole2kb, tvall and others who've remained focused on this issue!
    This is good news so far. At least, it is not causing additional problems. I will update MiRaGe tomorrow with this kernel to make it available for wider testing.
    05-04-2012 04:00 PM
  24. cole2kb's Avatar
    Here's a question, could the new kernel updates cause issues for those that haven't updated to zv9?

    Sent from my VM670 using Tapatalk 2
    05-04-2012 04:02 PM
  25. kwknott's Avatar
    Does seem like new kernel has helped, mrg. Been texting back and forth with my wife for about 40 min and no reboots and messages are sending right away unlike before they took 10 to 20 seconds or more. Will keep testing and see what happens. Thanks everyone for jumping on this so fast!!
    mrg666 and Dannemand like this.
    05-04-2012 04:04 PM
937 ... 3334353637 ...
LINK TO POST COPIED TO CLIPBOARD