[ROM] CM6-clean for development / testing

[Guest twrock]

Hmm, I just had another instance of the phone call speaker not working when I picked up a call. The other person could hear me.

I tried enabling and disabling airplane mode. Didn't help.

Rebooted, and the sound returned.

According to our previous discussion on this topic, this has happened to someone else with TXG, so it looks like a CM6 problem.

[Guest twrock]

md5, but I might have missed one.

I'm wonderig though, because initially wifi was working... that makes me think the set is complete.

So in the end, do you think this is "hopeless"? Is there anything else to try?

Just for my own understanding, are the /system files in the .32 install zip directly from the CM7.2 rom, or some other source?

[Guest dr.flo]

So in the end, do you think this is "hopeless"? Is there anything else to try?

Just for my own understanding, are the /system files in the .32 install zip directly from the CM7.2 rom, or some other source?

I don't know if it is really hopeless....

Sources of system files are

/bin -> cm72

/lib/hw -> u8150 froyo

/wifi -> um840 ported rom with new wifi driver

bin / lib are for sure not related to wifi. They just needed replacements reflecting changes in kernel. If you leave them out, you get slow graphics, rotation not working, and compcache not working.

You could try inserting the whole system/wifi folder of the current um840 port. I only picked the files that differ. I believe I didn't miss one, but who knows.

As far as I know, pier used wpa_supplicant and wpa_supplicant.conf from cm6, later replacing wpa_supplicant with cm7 version. That's the same setup that we have here, so I don't know what to tweak.

You can look at wpa_supplicant.conf of um840 and verify that it's really the same as ours. It might be that pier has mixed cm7 and cm6 elements to get working wifi in um840. If so, we could try to do the same.

Places influencing wifi are

- wpa_supplicant

- wpa_supplicant.conf

- whole system/wifi folder

- ramdisk (init.rc, init.qcom.rc)

- libhardware_legacy.so (but um840 uses cm6 version as far as I remember)

Edited March 5, 2012 by dr.flo

[Guest twrock]

I don't know if it is really hopeless....

Sources of system files are

/bin -> cm72

/lib/hw -> u8150 froyo

/wifi -> um840 ported rom with new wifi driver

bin / lib are for sure not related to wifi. They just needed replacements reflecting changes in kernel. If you leave them out, you get slow graphics, rotation not working, and compcache not working.

You could try inserting the whole system/wifi folder of the current um840 port. I only picked the files that differ. I believe I didn't miss one, but who knows.

As far as I know, pier used wpa_supplicant and wpa_supplicant.conf from cm6, later replacing wpa_supplicant with cm7 version. That's the same setup that we have here, so I don't know what to tweak.

You can look at wpa_supplicant.conf of um840 and verify that it's really the same as ours. It might be that pier has mixed cm7 and cm6 elements to get working wifi in um840. If so, we could try to do the same.

Places influencing wifi are

- wpa_supplicant

- wpa_supplicant.conf

- whole system/wifi folder

- ramdisk (init.rc, init.qcom.rc)

- libhardware_legacy.so (but um840 uses cm6 version as far as I remember)

Thanks. Lots of info there for me to digest.

So far, the only thing I noticed is that your .32 kernel update zip did not have one file, /sysytem/wifi/data.patch.hw2_0.bin, that was in UM840. If I'm reading things correctly, your script deletes it, but then doesn't replace it. I don't have any idea if that matters at all. But am building a rom for testing that will include that file anyway.

Also, there are quite a few files in CM6 /system/wifi that are not in UM840. But I'm assuming they are there for a reason.

I'm sure it has anything to do with wifi, but what about /system/lib/modules/2.6.32.9-perf/zram.ko? I'm curious about what it does.

There was already a ramzswap.ko file there that your script deletes.

[Guest dr.flo]

So far, the only thing I noticed is that your .32 kernel update zip did not have one file, /sysytem/wifi/data.patch.hw2_0.bin, that was in UM840. If I'm reading things correctly, your script deletes it, but then doesn't replace it.

Really? Then that's a mistake, the file should be there.

I'm sure it has anything to do with wifi, but what about /system/lib/modules/2.6.32.9-perf/zram.ko? I'm curious about what it does.

There was already a ramzswap.ko file there that your script deletes.

Compcache. The module type and name has changed between kernel .29 and .32., the compcache script in bin autodetects which version is there. Autodetect was introduced in cm7, before that there was only support for ramzswap, which we don't have (and need) for kernel .32

[Guest twrock]

Compcache. The module type and name has changed between kernel .29 and .32., the compcache script in bin autodetects which version is there. Autodetect was introduced in cm7, before that there was only support for ramzswap, which we don't have (and need) for kernel .32

So sorry. Looks like once again the "old man" got it wrong. data.patch.hw2_0.bin is in your .32 kernel update. It just wasn't in your CM6.1.1 rom. I was looking in the wrong folder when I reached that conclusion. False alarm.

Any chance any of the other files in your CM6 that aren't in UM840 are causing trouble, similar to the extra file that was in /lib/hw previously?

But to clarify for me (I'm slow), /system/lib/modules/2.6.32.9-perf/zram.ko that your .32 kernel update installs, is needed for compcache to work, and ramzswap.ko is not needed at all, so your script correctly deletes ramzswap.ko. Yes? NVM, it's obvious.

BTW, my attempt to combine your .32 kernel update with the CM6.1.1 rom failed. Somehow I lost 3g data (gprs still worked though). So I will go back to installing those in two separate steps until I figure out whether or not things actually work.

Edited March 6, 2012 by twrock

[Guest twrock]

Flashed a slightly modified .32 kernel update for testing. After that, Windows can not find a driver for the USB mass storage device. I know that there were discussions about this regarding CM7, but the proposed solution for that rom won't work with this one. (http://forum.xda-developers.com/showpost.php?p=18945071&postcount=4) This rom doesn't have that setting. Was there some other solution I haven't found by searching?

Pier said this previously:

Has a solution been found for CM6?

[Guest twrock]

dr. flo, did you use Set MAC Address when you were testing this rom with the .32 kernel? Any way that app is related to the wifi problems?

[Guest PaBliLlo]

Anyone knows if with this ROM, the phone can be used like 3G modem? I've tried others CM6 and the Windows drivers doesn't worked. The PC-Studio and the syncro tool didn't found the phone connected to the USB port, since the ROM was updated.

I want to use the new ROM, but I need the both tools I've mentioned.

Thanks!!!

[Guest dr.flo]

After that, Windows can not find a driver for the USB mass storage device. ...

Pier said this previously: http://android.modac...ost__p__1888140

Has a solution been found for CM6?

This seems to be not driver related. Possibly some config options in board config during compilation that select old kernel compatibility... Did you ever try kernel .32 with TXG? Does usb mass storage work there? I'm just asking because in my roms I used "vold" from cm72, this is somehow related to usb / sd etc. Maybe vold from original cm6 works better. Vold from my build doesn't recognize more than one sd card partition, so I needed to replace it.

dr. flo, did you use Set MAC Address when you were testing this rom with the .32 kernel? Any way that app is related to the wifi problems?

This app patches files inside wifi directory, so it might be related. But as of now, there were no complaints with um840 rom and setmacaddress / wifi reboots. Either no one uses it, or it doesn't cause problems there.

I did use it before exchanging kernels, so the new wifi driver found a file that was already patched (it is in those that do not get replaced). I don't know whether the new wifi driver doesn't like this way of patching mac address. But then why are there no issues on um840 rom? Or cm72, which has the same wifi driver since latest update?

Anyone knows if with this ROM, the phone can be used like 3G modem? I've tried others CM6 and the Windows drivers doesn't worked. The PC-Studio and the syncro tool didn't found the phone connected to the USB port, since the ROM was updated.

I want to use the new ROM, but I need the both tools I've mentioned.

Thanks!!!

I don't know, do any of those features work with other cm6 / cm7 roms?

Edited March 6, 2012 by dr.flo

[Guest twrock]

This seems to be not driver related. Possibly some config options in board config during compilation that select old kernel compatibility... Did you ever try kernel .32 with TXG? Does usb mass storage work there? I'm just asking because in my roms I used "vold" from cm72, this is somehow related to usb / sd etc. Maybe vold from original cm6 works better. Vold from my build doesn't recognize more than one sd card partition, so I needed to replace it.

I haven't tried .32 with TXG. In theory, should it work just to flash your .32 kernel update on top of an installed TXG?

This app patches files inside wifi directory, so it might be related. But as of now, there were no complaints with um840 rom and setmacaddress / wifi reboots. Either no one uses it, or it doesn't cause problems there.

I did use it before exchanging kernels, so the new wifi driver found a file that was already patched (it is in those that do not get replaced). I don't know whether the new wifi driver doesn't like this way of patching mac address. But then why are there no issues on um840 rom? Or cm72, which has the same wifi driver since latest update?

Following your information, I recopied all the /system files from the sources you indicated (exception was U8150), BUT I also copied all files in the UM840 /system/wifi folder instead of just the four you had previously. I flashed that update. Now I'm using it to see if the wifi problem happens again. So far over the past 1.5 days, there has been no problem. I will use if for a few days more of testing before making any "tentative" conclusions. (The last time, wifi problems began on the first day of use.)

I noted that after flashing the .32 kernel update, the MAC address is no longer correct. I too had used SetMacAddress before flashing your kernel update. BUT, I have purposefully not used SetMacAddress after flashing .32 this time. I want to test for a while before using it, and then after using it, for comparison.

I did have some wifi trouble with the latest CM7, but I have no recollection if I had even tried to use SetMacAddress or not. So I'm not prepared to say whether or not there is a problem there.

[Guest dr.flo]

I cannot confirm usb mass storage not working. It works for me (both on ubunutu 10.04 x86, and windows 7 x64) without issues.

Tested on a clean install (cm6.1.1-u8220-v3 + kernel update flashed in one go, without booting into rom first).

The kernel update should also work on TXG. It might be needed to delete sensors.u8220.so there, to make it use sensors.default.so (I think the priority for those is u8220 -> msm7k -> default. That's what broke rotation when the wrong sensors.msm7k.so file was present).

[Guest dr.flo]

I just managed to break wifi two times by using setMacAddress. Environment: clean install as described in post just above.

It changes to "disconnected" state after using setMacAddress. Trying to turn wifi off results in a reboot.

It does not matter if wifi is on while doing setMacAddress or not. Conclusion: New wifi driver does not like what setMacAddress does.

wifi can be restored by restoring caldata.bin.* via adb push (or most likely via root explorer or similar, if you're not using adb).

Briefly scanning through the other threads I think setMacAddress hasn't been used with new wifi driver on any rom: on um840 it has been reported not to work, and in cm72 it is no longer included by default (is it?), which makes it less likely that someone tried it.

[Guest twrock]

I cannot confirm usb mass storage not working. It works for me (both on ubunutu 10.04 x86, and windows 7 x64) without issues.

Tested on a clean install (cm6.1.1-u8220-v3 + kernel update flashed in one go, without booting into rom first).

The kernel update should also work on TXG. It might be needed to delete sensors.u8220.so there, to make it use sensors.default.so (I think the priority for those is u8220 -> msm7k -> default. That's what broke rotation when the wrong sensors.msm7k.so file was present).

I'll try again with my latest build. I'm using Windows XP. Don't know if that makes a difference or not. It searches for a driver, but can't find one. Confirmed that Linux (Ubuntu) works fine.

So if Set MAC Address won't work with the new wifi driver, what option is there for people who must have a correct MAC address?

I have a new build of your CM6.1.1 with the .32 kernel (Styx). I think I'll start a new thread with a link. Definitely beta, since I'm not sure what problems I might have introduced. So far, Wifi is working fine for me.

[Guest jafp]

Briefly scanning through the other threads I think setMacAddress hasn't been used with new wifi driver on any rom: on um840 it has been reported not to work, and in cm72 it is no longer included by default (is it?), which makes it less likely that someone tried it.

In cm72 cm7.2_u8220_20120301.zip (last version) setmacaddress is included but not working,

Pier has this information "use SetMacAddress app from Market to manually set mac address. Do once." in the first topic of [ROM] CyanogenMod 7.2 | Kernel 2.6.32 | Camera works. In CM7.2 kernel 2.6.32 i have tried with the app from the market and i have the some problem as you descrive, i have tried several things, even install the rom again, but nothing.

I beleave that many people will not need this, but in my situation i will need because in my job, the wifi is protected by mac.

[Guest dr.flo]

...

i have tried with the app from the market and i have the some problem as you descrive, i have tried several things, even install the rom again, but nothing.

I beleave that many people will not need this, but in my situation i will need because in my job, the wifi is protected by mac.

Thank you for confirming this!

For you it would mean you have to use the older version of cm72 with wifi reconnection bug - setmacaddress was working there.

[Guest dr.flo]

I'll try again with my latest build. I'm using Windows XP. Don't know if that makes a difference or not. It searches for a driver, but can't find one. Confirmed that Linux (Ubuntu) works fine.

So if Set MAC Address won't work with the new wifi driver, what option is there for people who must have a correct MAC address?

I'll try with XP later today. But usb mass storage should not need drivers. Drivers are needed for the other interfaces like adb (on windows)

Currently there's no known alternative to change the mac address... I remember having read something about ar6000 and a "softmac file". Maybe it's worth googeling some more about what a softmac file is and where the driver would expect to find it...

edit: softmac file support is what we would want, but it has to be enabled in the driver, which is what we cannot do, since we don't have the sources....

edit2: seems our ar6000.ko has set softmac file to NULL by default, but it might still take it as an insmod parameter. Possibly, libhardware_legacy.so can be updated to pass arguments to the driver (as it does for bcm wifi driver!) A solution might be close....

Edited March 7, 2012 by dr.flo

[Guest dr.flo]

mass storage worked on windows xp for me.

softmac solution has been posted in styx thread...

[Guest PaBliLlo]

Anyone knows if with this ROM, the phone can be used like 3G modem? I've tried others CM6 and the Windows drivers doesn't worked. The PC-Studio and the syncro tool didn't found the phone connected to the USB port, since the ROM was updated.

I want to use the new ROM, but I need the both tools I've mentioned.

Thanks!!!

I don't know, do any of those features work with other cm6 / cm7 roms?

Never worked with cm6 or cm7. I've tried official ROMs (from Ancel, Movilnet, Tre and T-Mobile) based on Android 1.5 and 2.1, and the features worked fine. But using CM (based on 2.2+) I've lost syncro with Outlook and the 3G modem. Surely I need to update some Windows driver, but I didn't find it. Thanks!

Edited March 8, 2012 by PaBliLlo

[Guest desalesouche]

i could observe(if i remenber well) with setcpu in advanced option while testing rom and battery with antutu tester :

cm6.1.1-29k:sampling rate= -1 / upthreshold= -1 / ignore nice load= -1 / powersavebias= -1

cm6.1.1-32k::sampling rate= 20000 /upthreshold=76 / ignore nice load=0 / powersavebias=0

if somebody can explain difference? is it better with these change?

does flashing 32k zip on cm6.1.1-29k is making these change?

what if setting sampling rate to = 200000 tried by twrock thinking 20000 was a typo?

[Guest twrock]

i could observe(if i remenber well) with setcpu in advanced option while testing rom and battery with antutu tester :

cm6.1.1-29k:sampling rate= -1 / upthreshold= -1 / ignore nice load= -1 / powersavebias= -1

cm6.1.1-32k::sampling rate= 20000 /upthreshold=76 / ignore nice load=0 / powersavebias=0

if somebody can explain difference? is it better with these change?

does flashing 32k zip on cm6.1.1-29k is making these change?

what if setting sampling rate to = 200000 tried by twrock thinking 20000 was a typo?

The thing about .29 kernel is that it shows up strangely in SetCPU. The first time you tap on advanced options, you get that weird set of -1 numbers. But if you click back to another tab and then back to advanced, you will see something closer to what it should be.

BUT, the numbers won't match what the boot.img should be set at. That is what I find most strange. And when I run SetCPU with .32 kernel. the numbers match what is set in the boot.img. So I feel more comfortable that things are "right" with .32 kernel boot.img. Both boot.img's are set at 76 upthreshold, and if unmodified, 20000 sampling_rate. So in theory, they should give you the same numbers in SetCPU, but they don't.

What does it all mean? I don't know. :huh: Maybe nothing. But since there appears to be no advantage to running the .29 kernel, and that is the one that looks strange in SetCPU, I feel better using the .32 kernel.

Sampling_rate can be modified in the boot.img quite easily, so I typically just do it there for roms I modifiy.