Could someone please help?

[DZ]

Sorry. I'm new to this group and the whole process of Android-x86 building. I have setup my build system and tested it with an Android-x86-4.4-r2 source tree that someone in this group was kind enough to upload to github to help someone else out. I would now like to move on to the latest kitkat source of Android-x86-4.4-r3. I have been trying to sync it for a week and keep encountering sync errors. From other recent posts it seems that other people are having issues downloading source from the site. Sorry if this is a dumb question but could I ask who is responsible for maintaining the site and what the problem is? I feel like I'm trying over and over in the dark without really knowing what is the status of the site and whether I am trying in vain. I also saw a message encouraging people in the US to use the SourceForge site but that had problems too. I'm really getting discouraged with the Android-x86. I would appreciate some info and advice. Thank you, in advance, for any assistance.

[pstglia]

Hi

Em sábado, 19 de setembro de 2015 11:29:17 UTC, DZ escreveu:

Sorry. I'm new to this group and the whole process of Android-x86 building. I have setup my build system and tested it with an Android-x86-4.4-r2 source tree that someone in this group was kind enough to upload to github to help someone else out. I would now like to move on to the latest kitkat source of Android-x86-4.4-r3. I have been trying to sync it for a week and keep encountering sync errors. From other recent posts it seems that other people are having issues downloading source from the site. Sorry if this is a dumb question but could I ask who is responsible for maintaining the site and what the problem is? I feel like I'm trying over and over in the dark without really knowing what is the status of the site and whether I am trying in vain. I also saw a message encouraging people in the US to use the SourceForge site but that had problems too. I'm really getting discouraged with the Android-x86. I would appreciate some info and advice. Thank you, in advance, for any assistance.

Quoting Chih-Wei post:

"The git server is provided by TANet Hualien Center 

and maintained by some volunteers from Taiwan 

Linux user group (TLUG). It provides many services 

for TLUG for over 10 years and serves our project 

over 5 years. " 

Due servers age sometimes there are problems. The ideal was a new server.

But I believe Chih-Wei is trying to contact those with physical access to this server and will put it online.

You can check Chih-Wei original post here

https://groups.google.com/forum/#!searchin/android-x86/taiwan$20group/android-x86/7iLBuR662hU/dTo5I57XoJcJ

Regards,

Pstglia

[DZ]

Thanks for replying. Do you happen to know why the Android-x86 repository has not been moved to github? This would seem to make sense but perhaps there are good reasons not too.

Also, if you happen to have an unmodified copy of 4.4-r3 would you be able to upload it to github (or somewhere else) temporarily to help those of us who have been struggling to get it? I'm a newbie and I suspect other newbies are struggling and getting discouraged. It would help a lot to have a temporary solution. Thanks, in advance, for any help you can provide.

[pstglia]

Thanks for replying. Do you happen to know why the Android-x86 repository has not been moved to github? This would seem to make sense but perhaps there are good reasons not too.

Chih-Wei answered this another thread.

 

Also, if you happen to have an unmodified copy of 4.4-r3 would you be able to upload it to github (or somewhere else) temporarily to help those of us who have been struggling to get it? I'm a newbie and I suspect other newbies are struggling and getting discouraged. It would help a lot to have a temporary solution. Thanks, in advance, for any help you can provide.

I don't have an untouched one. But I have one with small mods that can be reverted (I I'm not wrong have just changed kernel - last 10 commits)

I've just uploaded manifest to github (https://github.com/pstglia). If interested:

mkdir your_src_dir

cd your_src_dir

repo init -u https://github.com/pstglia/manifest

repo sync

Note: Is there the possibility that I did something wrong when changing manifest (default.xml). Be warned

Regards,

Pstglia

[DZ]

Thanks for uploading. I'll give it a try!

[DZ]

I downloaded and tried to build with "lunch 6" and got this error:

make[2]: *** [drivers] Error 2
make[2]: Leaving directory `/home/zwikel/pstglia/out/target/product/x86/obj/kernel'
make[1]: *** [sub-make] Error 2
make[1]: Leaving directory `/home/zwikel/pstglia/kernel'
make: *** [out/target/product/x86/kernel] Error 2
make: *** Waiting for unfinished jobs....
external/llvm/include/llvm/IR/IRBuilder.h: In member function 'virtual llvm::Value* (anonymous namespace)::SparcV9ABIInfo::EmitVAArg(llvm::Value*, clang::QualType, clang::CodeGen::CodeGenFunction&) const':
external/llvm/include/llvm/IR/IRBuilder.h:928:66: warning: 'Stride' may be used uninitialized in this function [-Wmaybe-uninitialized]
external/clang/lib/CodeGen/TargetInfo.cpp:5313:12: note: 'Stride' was declared here
external/clang/lib/CodeGen/TargetInfo.cpp:5475:1: warning: 'ArgAddr' may be used uninitialized in this function [-Wmaybe-uninitialized]
external/clang/lib/CodeGen/TargetInfo.cpp:5312:16: note: 'ArgAddr' was declared here
rm out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_dri_common_intermediates/xmlpool/nl.po out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_dri_common_intermediates/xmlpool/fr.po out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_dri_common_intermediates/xmlpool/es.po out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_dri_common_intermediates/xmlpool/sv.po out/target/product/x86/obj/STATIC_LIBRARIES/libmesa_dri_common_intermediates/xmlpool/de.po

Should I remove your mods and try again? If so, how would I remove them? THanks

[pstglia]

If the error messages are related to arizona, wm5102 or sst, this could be my commits (although I can compile fine here). The lines you posted contains warnings from other stuff.

Anyway, to revert a commit there are some methods. One of them is "git revert" with the range of commits you want to undo

Note: I use to build using
make -j5 iso_img
or
make -j5 efi_img

[DZ]

I use make -j4 iso_img. I'll repo sync again to make sure I'm not missing anything and then run make again.

[DZ]

I resynced and tried doing a make 5 and also tried a make 6. Keep getting compile errors. I guess I'll have to give up and wait for 4.4-r3 to be available on the Android-x86 server (not sure when that will be!). Thank you for your help.

[Elie De Brauwer]

I could reproduce DZ's failure, but a bit higher is the interesting information:

/opt/android/workspace/android-x86-github-pstglia-workspace/kernel/drivers/mfd/wm5102-tables.c:142:5: error: redefinition of ‘wm5102_patch’

 int wm5102_patch(struct arizona *arizona)

     ^

In file included from /opt/android/workspace/android-x86-github-pstglia-workspace/kernel/drivers/mfd/wm5102-tables.c:16:0:

/opt/android/workspace/android-x86-github-pstglia-workspace/kernel/include/linux/mfd/arizona/core.h:199:19: note: previous definition of ‘wm5102_patch’ was here

 static inline int wm5102_patch(struct arizona *arizona)

                   ^

/opt/android/workspace/android-x86-github-pstglia-workspace/kernel/scripts/Makefile.build:258: recipe for target 'drivers/mfd/wm5102-tables.o' failed

make[4]: *** [drivers/mfd/wm5102-tables.o] Error 1

make[4]: *** Waiting for unfinished jobs....

After this I did a 'git checkout bdf7a59a2adbc83bfa52c67994024d0c25f904f9'  in the kernel folder which should be the last commit before your experiments. Perhaps it compiled fine on your end due to And at that point it required one additional fix which can be fixed by two slashes:

/opt/android/workspace/android-x86-github-pstglia-workspace/kernel/drivers/net/wireless/wl/src/wl/sys/wl_linux.c:2051:3: error: macro "__DATE__" might prevent reproducible builds [-Werror=date-time]

   __DATE__, __TIME__, EPI_VERSION_STR);

 

And at that point I succeeded in building the kernel, and my build is continuing. 

[pstglia]

Thanks for the info Elie. Let us know if you can complete the build reverting these commits

Regards,

Pstglia

[pstglia]

I tested building kernel here, and have the same errors as you have. I probably added something in a last session I didn't remember.

/home/paulo/Downloads/android-x86_novo/kernel/drivers/mfd/wm5102-tables.c:142:5: error: redefinition of 'wm5102_patch'

 int wm5102_patch(struct arizona *arizona)

     ^

In file included from /home/paulo/Downloads/android-x86_novo/kernel/drivers/mfd/wm5102-tables.c:16:0:

/home/paulo/Downloads/android-x86_novo/kernel/include/linux/mfd/arizona/core.h:199:19: note: previous definition of 'wm5102_patch' was here

 static inline int wm5102_patch(struct arizona *arizona)

                   ^

make[4]: *** [drivers/mfd/wm5102-tables.o] Error 1

make[3]: *** [drivers/mfd] Error 2

make[3]: *** Waiting for unfinished jobs....

You can checkpoint to previous commit like Elie suggested. Sorry for the confusion

[Elie De Brauwer]

Op dinsdag 22 september 2015 03:33:37 UTC+2 schreef pstglia:

Hi Paulo,

With the patches described (reverting to the last commit, and fixing the __DATE__ / __TIME__ macro's) my build succeed and seems to be complete. Thanks again for setting up the mirror, it saved my day ;).

E.

 

 

[DZ]

Thanks to both of you. Elie -- I'm unclear on what to change for the macros you mention. You mentioned two slashes but I don't understand where they go. Could you please explain? Thank you.

[rbg]

and / or the patch should have been conditional based on gcc version maybe!!!

re: https://groups.google.com/forum/#!searchin/android-x86/$20__DATE__$20$2F%7Csort:date/android-x86/nT9OidsPCa8/2SP0PcvnBQAJ

[DZ]

I did a 'git checkout bdf7a59a2adbc83bfa52c67994024d0c25f904f9' in the kernel folder as suggested and did a lunch 6, make -jn iso_img. This seemed to complete without error. However when I install using the iso, wireless and Bluetooth do not work whereas they did with the standard 4.4-r3 build. Are we sure the source corresponds to the same source used for the standard -r3 build? If so, them perhaps I did something wrong!

[pstglia]

Em terça-feira, 22 de setembro de 2015 23:07:03 UTC, DZ escreveu:

I did a 'git checkout bdf7a59a2adbc83bfa52c67994024d0c25f904f9'  in the kernel folder as suggested and did a lunch 6, make -jn iso_img. This seemed to complete without error. However when I install using the iso, wireless and Bluetooth do not work whereas they did with the standard 4.4-r3 build. Are we sure the source corresponds to the same source used for the standard -r3 build? If so, them perhaps I did something wrong!
 

It was synced after kk 4.4 r3 release. If you look at authored dates/commit dates from Chih-Wei on different repos ( external-bluetooth-bluez on Jul 17, kernel at Jul 16 ) they are near the released iso/img.

I believe it's paired, but I won't say I'm 100% sure :)

Try comparing kernel modules  loaded on regular release and not on your generated one.

[Elie De Brauwer]

Op dinsdag 22 september 2015 15:09:06 UTC+2 schreef DZ:

Hi,

Today I migrated Paulo's code to a local mirror for me. To fix the two issues I encountered  I did the following.

To revert the last 10 changes in the kernel repo:

$ git reset --hard bdf7a59a2adbc83bfa52c67994024d0c25f904f9

$ git push --force x86 HEAD:android-x86_4.4_r3_fork

(x86 being the name of the remote).

To fix the compilation issue due to the __DATE__ and __TIME__ errors, I created a patch file (data_macro.patch) which is placed in the kernel under drivers/net/wireless/wl and which gets applied in build/core/tasks/kernel.mk where the broadcom driver is downloaded and extracted. 

Attached you can find two patches which accomplish this, not 100% clean, but playing along with how these patches are currently handled.

my 2 cents

E.

[DZ]

So do both patch files go in the kernel under drivers/net/wireless/wl?

[DZ]

Odd. I recreated the repo, downloaded the source, rebuilt the iso, and did not get the error. Has something??

[Elie De Brauwer]

Op woensdag 23 september 2015 14:45:11 UTC+2 schreef DZ:

No one needs to be applied to the 'linux' git repo and the other needs to be applied to the 'build' git repo, obviously to the 4.4_r3 branches. You only need them if your kernel builds fails due to the presence of the _DATE_ macro as shown above.

E.

 

[Mauro Rossi]

Hi DZ, Elie,
Please check how the macros build errore with GCC 4.9.2 was solved in the thread mentioned by rbg
Wrt rbg doubt about checking GCC, version it was evaluated but immediately discarded as an option.
Motivations are in the aforementioned thread: removal of date and time macros is OK, also with previous GCC versions.
M.