edit: The stock kernel can be built following this thread.
I'm trying to build the stock kernel from source (the one from huaweidevice). I'm running Ubuntu 10.10 64-bit.
I've gotten this far:
- I've extracted config from phone /proc/config.gz and copied it to kernel/.config
- I went through this guide: source.android.com source initializing.html
- I created some symbolic links to get rid of file not found errors when compiling:
kernel-2.6.32-U8800-Froyo/kernel/net/netfilter$ ln -s xt_connmark.c xt_CONNMARK.c
kernel-2.6.32-U8800-Froyo/kernel/net/netfilter$ ln -s xt_HL.c xt_hl.c
kernel-2.6.32-U8800-Froyo/kernel/net/netfilter$ ln -s xt_MARK.c xt_mark.c
kernel-2.6.32-U8800-Froyo/kernel/net/ipv4/netfilter$ ln -s ipt_ECN.c ipt_ecn.c
kernel-2.6.32-U8800-Froyo/kernel/include/linux/netfilter$ ln -s xt_mark.h xt_MARK.h
kernel-2.6.32-U8800-Froyo/kernel/include/linux/netfilter_ipv4$ ln -s ipt_ecn.h ipt_ECN.h
kernel-2.6.32-U8800-Froyo/kernel/include/linux/netfilter$ ln -s xt_DSCP.h xt_dscp.h
kernel-2.6.32-U8800-Froyo/kernel/include/linux/netfilter_ipv4$ ln -s ipt_ttl.h ipt_TTL.h
kernel-2.6.32-U8800-Froyo/kernel/include/linux/netfilter_ipv6$ ln -s ip6t_hl.h ip6t_HL.h
Now for the cross compilation I've tried 3 different toolchains:
- The ones from developer.android.com sdk ndk index.html
- and codesourcery.com sgpp lite arm portal package8734/public/arm-none-eabi/arm-2011.03-42-arm-none-eabi-i686-pc-linux-gnu.tar.bz2
then I've tried compiling with:
make -j8 ARCH=arm CROSS_COMPILE=.../android-ndk-r5b/toolchains/arm-linux-androideabi-4.4.3/prebuilt/linux-x86/bin/arm-linux-androideabi-
and
make -j8 ARCH=arm CROSS_COMPILE=.../android-ndk-r5b/toolchains/arm-eabi-4.4.0/prebuilt/linux-x86/bin/arm-eabi-
and
make -j8 ARCH=arm CROSS_COMPILE=.../arm-2011.03/bin/arm-none-eabi-
they all give me output similar to this (this is with the 4.4.0 toolchain):
Code:
CHK include/linux/version.h
SYMLINK include/asm -> include/asm-arm
make[1]: `include/asm-arm/mach-types.h' is up to date.
CHK include/linux/utsrelease.h
CALL scripts/checksyscalls.sh
CHK include/linux/compile.h
CC net/netfilter/xt_MARK.o
CC net/ipv4/netfilter/ipt_ecn.o
CC net/netfilter/xt_NFQUEUE.o
CC net/ipv4/netfilter/ipt_NETMAP.o
CC net/netfilter/xt_comment.o
CC net/netfilter/xt_connlimit.o
net/netfilter/xt_MARK.c: In function 'mark_tg':
net/netfilter/xt_MARK.c:32: error: dereferencing pointer to incomplete type
net/netfilter/xt_MARK.c:32: error: dereferencing pointer to incomplete type
net/netfilter/xt_MARK.c: At top level:
net/netfilter/xt_MARK.c:41: error: invalid application of 'sizeof' to incomplete type 'struct xt_mark_tginfo2'
make[2]: *** [net/netfilter/xt_MARK.o] Error 1
make[2]: *** Waiting for unfinished jobs....
CC net/ipv4/netfilter/ipt_REDIRECT.o
CC net/ipv4/netfilter/ipt_REJECT.o
cc1: warnings being treated as errors
net/ipv4/netfilter/ipt_ecn.c:29: error: 'struct ipt_ECN_info' declared inside parameter list
net/ipv4/netfilter/ipt_ecn.c:29: error: its scope is only this definition or declaration, which is probably not what you want
net/ipv4/netfilter/ipt_ecn.c: In function 'set_ect_ip':
net/ipv4/netfilter/ipt_ecn.c:33: error: 'XT_DSCP_MASK' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:33: error: (Each undeclared identifier is reported only once
net/ipv4/netfilter/ipt_ecn.c:33: error: for each function it appears in.)
net/ipv4/netfilter/ipt_ecn.c:33: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:40: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c: At top level:
net/ipv4/netfilter/ipt_ecn.c:48: error: 'struct ipt_ECN_info' declared inside parameter list
net/ipv4/netfilter/ipt_ecn.c: In function 'set_ect_tcp':
net/ipv4/netfilter/ipt_ecn.c:58: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:58: error: 'IPT_ECN_OP_SET_ECE' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:59: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:60: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:60: error: 'IPT_ECN_OP_SET_CWR' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:61: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:69: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:70: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:71: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:72: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c: In function 'ecn_tg':
net/ipv4/netfilter/ipt_ecn.c:84: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:84: error: 'IPT_ECN_OP_SET_IP' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:85: error: passing argument 2 of 'set_ect_ip' from incompatible pointer type
net/ipv4/netfilter/ipt_ecn.c:29: note: expected 'const struct ipt_ECN_info *' but argument is of type 'const struct ipt_ECN_info *'
net/ipv4/netfilter/ipt_ecn.c:88: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:88: error: 'IPT_ECN_OP_SET_ECE' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:88: error: 'IPT_ECN_OP_SET_CWR' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:90: error: passing argument 2 of 'set_ect_tcp' from incompatible pointer type
net/ipv4/netfilter/ipt_ecn.c:48: note: expected 'const struct ipt_ECN_info *' but argument is of type 'const struct ipt_ECN_info *'
net/ipv4/netfilter/ipt_ecn.c: In function 'ecn_tg_check':
net/ipv4/netfilter/ipt_ecn.c:101: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:101: error: 'IPT_ECN_OP_MASK' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:103: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:106: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:106: error: 'XT_DSCP_MASK' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:108: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:111: error: dereferencing pointer to incomplete type
net/ipv4/netfilter/ipt_ecn.c:111: error: 'IPT_ECN_OP_SET_ECE' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c:111: error: 'IPT_ECN_OP_SET_CWR' undeclared (first use in this function)
net/ipv4/netfilter/ipt_ecn.c: At top level:
net/ipv4/netfilter/ipt_ecn.c:124: error: invalid application of 'sizeof' to incomplete type 'struct ipt_ECN_info'
make[3]: *** [net/ipv4/netfilter/ipt_ecn.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[1]: *** [net/netfilter] Error 2
make[1]: *** Waiting for unfinished jobs....
make[2]: *** [net/ipv4/netfilter] Error 2
make[1]: *** [net/ipv4] Error 2
make: *** [net] Error 2
sorry for the messed up links, I'm not allowed to post links (yet).
Any ideas how to get forward?
The reason I'm doing this is that I want to fix the bug that the internal mic is disabled when a stereo headphone / or a 3.5mm jack - 2xrca cable is plugged in. I think I can fix it from kernel-2.6.32-U8800-Froyo/kernel/arch/arm/mach-msm/rpc_server_handset.c, but first I have to get my build environment working
copy missing files instead of link
xt_MARK.c and xt_mark.c are not the same, at least in my source
try copy the missing file from android source
http://android.git.kernel.org/?p=kernel/msm.git;a=tree
wmi543 said:
xt_MARK.c and xt_mark.c are not the same, at least in my source
try copy the missing file from android source
Click to expand...
Click to collapse
It worked I copied all the files from the git instead of ln -s.
"Kernel: arch/arm/boot/zImage is ready"
Thanks for the quick reply.
Then I had to remove comment out CONFIG_FB_MSM_MDDI_NT35582_WVGA=y line from .config and answer no to two questions when running make.
I used the 4.4.0 toolchain
Now I have the zImage but don't know how it'll work because the CONFIG_FB_MSM_MDDI_NT35582_WVGA probably has something to do with the display type.
BTW I've taken a backup with genokolar's recovery img. When I try this new kernel and if I can't boot, I should be able to get it to boot again by doing a restore from the earlier backup?
Now I'll start looking into packing the kernel to an .img (android-dls.com wiki/index.php?title=HOWTO:_Unpack,_Edit,_and_Re-Pack_Boot_Images#Structure_of_boot_and_recovery_images)..
OK. I copied the boot.img from the phone in bootloader mode, unpacked it with unpack-bootimg.pl. Then I bsdiffed (binary diff) the resulting boot.img-kernel.gz with the Image file I built, the patch file is only 389K so looks like they are quite similar. The only differences may be because of the display thing I changed in the Makefile.
Might try tomorrow to replace the unpacked kernel with the Image file and see if it works. What is the worst case scenario? I should always be able to restore from the old backup if it doesn't boot?
Maybe I'll also try to get the display thing to build right before trying it on the phone
edit: I read somewhere the .gz extension in the unpacked kernel file is a bug with the unpacking script, so "boot.img-kernel.gz" is actually the unpacked kernel file (so it should be the same as compiled Image) ?
Looks like the unpack-bootimg.pl and repack-bootimg.pl are not working right for the U8800 boot.img. I just tried to unpack the original boot.img and repack it, I got a blue screen when trying to boot. I could restore the original just by going to bootloader mode and copying back the original.
Any ideas how to unpack and repack the boot.img of the U8800?
You can repack the boot/recovery images with the following:
Code:
./mkbootfs RAMDISK DIRECTORY | ./minigzip > ramdisk-boot.img
./mkbootimg --kernel KERNEL --ramdisk ramdisk-boot.img --base 0x00200000 --pagesize 4096 --cmdline 'console=ttyDCC0 androidboot.hardware=huawei' -o boot.img
I just modified the repack perl script accordingly and added lines so that it will mount cust_backup and transfer the image across for me. Saves heaps of time.
stockwell said:
You can repack the boot/recovery images with the following:
Code:
./mkbootfs RAMDISK DIRECTORY | ./minigzip > ramdisk-boot.img
./mkbootimg --kernel KERNEL --ramdisk ramdisk-boot.img --base 0x00200000 --pagesize 4096 --cmdline 'console=ttyDCC0 androidboot.hardware=huawei' -o boot.img
I just modified the repack perl script accordingly and added lines so that it will mount cust_backup and transfer the image across for me. Saves heaps of time.
Click to expand...
Click to collapse
I used the unpack script from http://forum.xda-developers.com/showthread.php?t=891333 then modified the line "#$bootimg = substr($bootimg,2048);" to "$bootimg = substr($bootimg,4096);". There's still a difference of 215 bytes with the repacked img, but it works.
Now I can test the kernel I built myself. Thank you for the help.
edit: Looks like the built kernel doesn't work, now I can start tackling that
CONFIG_FB_MSM_MDDI_NT35582_WVGA=y
if you see only blue screen when you boot your own kernel, refer to my experience, edit .config, change 2 lines as below instead of comment out CONFIG_FB_MSM_MDDI_NT35582_WVGA=y should fix it
CONFIG_FB_MSM_MDDI_AUTO_DETECT=y
CONFIG_FB_MSM_TRY_MDDI_CATCH_LCDC_PRISM=y
BTW, there is a tool "abootimg" which is quite handy for packing the kernel
http://gitorious.org/ac100/abootimg
wmi543 said:
if you see only blue screen when you boot your own kernel, refer to my experience, edit .config, change 2 lines as below instead of comment out CONFIG_FB_MSM_MDDI_NT35582_WVGA=y should fix it
CONFIG_FB_MSM_MDDI_AUTO_DETECT=y
CONFIG_FB_MSM_TRY_MDDI_CATCH_LCDC_PRISM=y
BTW, there is a tool "abootimg" which is quite handy for packing the kernel
http://gitorious.org/ac100/abootimg
Click to expand...
Click to collapse
Works! great! Thanks for the help. Hopefully soon I can release my fixed kernel
Sorry, I'm new to Android. What advantages are there if you get the stock kernel or whatever?
"Stock" means the kernel that comes with the phone when you buy it. I just wanted to build it so I could modify it. I created a modified kernel that allows you to use the internal mic with headphones or for example when you connect the phone to your car stereo. The modified kernel is in the other thread.
Sent from my u8800 using XDA App
So with custom kernel you can achieve more functionality just like normal linux? Good. Would be awesome.
I am currently trying to build CM9 for my A500 on my Mac (which is running OS X 10.8). It is failing while building the kernel, with the error:
Code:
Undefined symbols for architecture i386:
"android::renderscript::rsi_Allocation1DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, void const*, unsigned long, unsigned int)", referenced from:
...
android::renderscript::rsp_Allocation2DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
I'll describe the steps I followed to get to that point below, but first: It doesn't look like I'm the only one having this problem, but it doesn't look common - I found the same error in three pastebins: 1, 2, 3. The interesting thing is that 1 and 2 there are both on OS X (both contain 'out/host/darwin-x86/'), and they are for different devices.
Anyway, to get to that point I did the following:
(Following the CyanogenMod Wiki template for Mac OS X building)
1. Trashed my entire MacPorts tree using the commands provided in the CyanogenMod wiki.
2. Reinstalled all the ports I already had installed, along with those indicated in the CyanogenMod wiki. I used the supplied command, but then after the described error appeared the first time, I installed all the ports with the +universal directive as well.
3. Checked out the code for CM9 using:
Code:
repo init -u http:github.com/CyanogenMod/android.git -b ics --reference=/Volumes/Backup/AndroidDevelopment/cm9/mirror/CyanogenMod/
repo sync
4. Following the post by waydownsouth, ran:
Code:
. build/envsetup.sh
lunch cm_a500-userdebug
This synced the a500 bits from the repository
5. Plugged in the tablet, and entered the commands:
Code:
cd device/acer/a500
./extract-files.sh
A bunch of files downloaded, and all was well.
6. I returned to the root of my working directory, and ran:
Code:
brunch cm_a500-userdebug
and received the error:
Code:
It bombed with:
make: *** No rule to make target `vendor/cm/proprietary/RomManager.apk', needed by `out/target/product/a500/system/app/RomManager.apk'. Stop.
make: *** Waiting for unfinished jobs....
7. Got the prebuilts using the commands:
Code:
cd vendor/cm
./get-prebuilts
8. Returned to the root of the working directory again, and ran:
Code:
brunch cm_a500-userdebug
It looked like it was working until I received the error:
Code:
Result was:
target thumb C++: libwebcore <= external/webkit/Source/WebCore/xml/XPathP****r.cpp
out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/XPathGrammar.hpp: In member function 'WebCore::XPath::Expression* WebCore::XPath::P****r::p****Statement(const WTF::String&, WTF::PassRefPtr<WebCore::XPathNSResolver>, WebCore::ExceptionCode&)':
out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/XPathGrammar.hpp:106: error: too many arguments to function 'int WebCore::XPath::xpathyyp****()'
external/webkit/Source/WebCore/xml/XPathP****r.cpp:480: error: at this point in file
make: *** [out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/xml/XPathP****r.o] Error 1
9. I found that the above error was due to Bison 2.6, and I shopped around, and found a patch which appeared to address this issue. It didn't apply quite correctly (on file Source/WebCore/xml/XPathP****r.cpp), but I manually merged that file.
10. I shut down and restarted (because I was going from one place to another, and stuff in on an external drive) so I got my environment variables set etc. again by running:
Code:
9. At this point, I shut down. When I recommenced, I ran
. build/envsetup.sh
lunch cm_a500-userdebug
brunch cm_a500-userdebug
Webkit built correctly this time, but then I hit the error:
Code:
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:4:17: error: elf.h: No such file or directory
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c: In function ‘main’:
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:9: error: ‘EI_NIDENT’ undeclared (first use in this function)
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:9: error: (Each undeclared identifier is reported only once
...
make[4]: *** [scripts/mod/mk_elfconfig] Error 1
11. I looked around and found one thread which suggested I copy the elf.h file to /external/elfutils/libelf but that didn't appear to fix the problem, so deleted the elf.h file from that directory again.
Then I found another thread which suggested the way to solve the problem was to build the kernel separately, and in advance using the command:
Code:
make -j`sysctl -an hw.logicalcpu` ARCH=arm CROSS_COMPILE=./prebuilt/darwin-x86/toolchain/arm-eabi-4.4.3/bin/arm-eabi- HOSTCFLAGS="-I./external/elfutils/libelf"
(Note that I changed the paths here to match my system).
To save from having to recompile everything every time, I ran:
export USE_CCACHE=1 && ./prebuilt/darwin-x86/ccache/ccache -M 20G
Then I used the command identified above (i.e. make -j`sysctl -an hw.logicalcpu` ...).
It failed after quite some time with the error:
Code:
Build failed with:
Undefined symbols for architecture i386:
"android::renderscript::rsi_Allocation1DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, void const*, unsigned long, unsigned int)", referenced from:
android::renderscript::rspr_Allocation1DElementData(android::renderscript::Context*, android::renderscript::Fifo*, unsigned char*, unsigned long)in libRS.a(rsgApiReplay.o)
android::renderscript::rsp_Allocation1DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
"android::renderscript::rsi_Allocation2DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, unsigned int, RsAllocationCubemapFace, void const*, unsigned long, unsigned int)", referenced from:
android::renderscript::rspr_Allocation2DElementData(android::renderscript::Context*, android::renderscript::Fifo*, unsigned char*, unsigned long)in libRS.a(rsgApiReplay.o)
android::renderscript::rsp_Allocation2DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
ld: symbol(s) not found for architecture i386
collect2: ld returned 1 exit status
As I noted above, it seems like one or two other people have had this error also. At this point, I'm basically stumped. I've never really done much C/C++ dev, so I'm fairly unfamiliar with reading the output of the linker. Looking over it, I wouldn't have expected it to need i386 symbols for libRS.a (if that is indeed what it is looking for symbols of) because it is building for ARM, but I could just be completely misunderstanding the output.
Anyway, if anyone has suggestions as to what might be causing the problem or how to address it (or even wants to briefly explain what that linker error actually means in a clearer way) that would be great. I've fought with CM7 builds in the past (for the Moto Defy) and basically given up because I couldn't get them to build. I'd rather not just do that again.
Have you solved this issue? Any update?
Any update?
FFU5y said:
I am currently trying to build CM9 for my A500 on my Mac (which is running OS X 10.8). It is failing while building the kernel, with the error:
Code:
Undefined symbols for architecture i386:
"android::renderscript::rsi_Allocation1DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, void const*, unsigned long, unsigned int)", referenced from:
...
android::renderscript::rsp_Allocation2DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
I'll describe the steps I followed to get to that point below, but first: It doesn't look like I'm the only one having this problem, but it doesn't look common - I found the same error in three pastebins: 1, 2, 3. The interesting thing is that 1 and 2 there are both on OS X (both contain 'out/host/darwin-x86/'), and they are for different devices.
Anyway, to get to that point I did the following:
(Following the CyanogenMod Wiki template for Mac OS X building)
1. Trashed my entire MacPorts tree using the commands provided in the CyanogenMod wiki.
2. Reinstalled all the ports I already had installed, along with those indicated in the CyanogenMod wiki. I used the supplied command, but then after the described error appeared the first time, I installed all the ports with the +universal directive as well.
3. Checked out the code for CM9 using:
Code:
repo init -u http:github.com/CyanogenMod/android.git -b ics --reference=/Volumes/Backup/AndroidDevelopment/cm9/mirror/CyanogenMod/
repo sync
4. Following the post by waydownsouth, ran:
Code:
. build/envsetup.sh
lunch cm_a500-userdebug
This synced the a500 bits from the repository
5. Plugged in the tablet, and entered the commands:
Code:
cd device/acer/a500
./extract-files.sh
A bunch of files downloaded, and all was well.
6. I returned to the root of my working directory, and ran:
Code:
brunch cm_a500-userdebug
and received the error:
Code:
It bombed with:
make: *** No rule to make target `vendor/cm/proprietary/RomManager.apk', needed by `out/target/product/a500/system/app/RomManager.apk'. Stop.
make: *** Waiting for unfinished jobs....
7. Got the prebuilts using the commands:
Code:
cd vendor/cm
./get-prebuilts
8. Returned to the root of the working directory again, and ran:
Code:
brunch cm_a500-userdebug
It looked like it was working until I received the error:
Code:
Result was:
target thumb C++: libwebcore <= external/webkit/Source/WebCore/xml/XPathP****r.cpp
out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/XPathGrammar.hpp: In member function 'WebCore::XPath::Expression* WebCore::XPath::P****r::p****Statement(const WTF::String&, WTF::PassRefPtr<WebCore::XPathNSResolver>, WebCore::ExceptionCode&)':
out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/XPathGrammar.hpp:106: error: too many arguments to function 'int WebCore::XPath::xpathyyp****()'
external/webkit/Source/WebCore/xml/XPathP****r.cpp:480: error: at this point in file
make: *** [out/target/product/a500/obj/STATIC_LIBRARIES/libwebcore_intermediates/Source/WebCore/xml/XPathP****r.o] Error 1
9. I found that the above error was due to Bison 2.6, and I shopped around, and found a patch which appeared to address this issue. It didn't apply quite correctly (on file Source/WebCore/xml/XPathP****r.cpp), but I manually merged that file.
10. I shut down and restarted (because I was going from one place to another, and stuff in on an external drive) so I got my environment variables set etc. again by running:
Code:
9. At this point, I shut down. When I recommenced, I ran
. build/envsetup.sh
lunch cm_a500-userdebug
brunch cm_a500-userdebug
Webkit built correctly this time, but then I hit the error:
Code:
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:4:17: error: elf.h: No such file or directory
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c: In function ‘main’:
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:9: error: ‘EI_NIDENT’ undeclared (first use in this function)
/Volumes/Backup/AndroidDevelopment/cm9/working_copy_2/kernel/acer/t20-common/scripts/mod/mk_elfconfig.c:9: error: (Each undeclared identifier is reported only once
...
make[4]: *** [scripts/mod/mk_elfconfig] Error 1
11. I looked around and found one thread which suggested I copy the elf.h file to /external/elfutils/libelf but that didn't appear to fix the problem, so deleted the elf.h file from that directory again.
Then I found another thread which suggested the way to solve the problem was to build the kernel separately, and in advance using the command:
Code:
make -j`sysctl -an hw.logicalcpu` ARCH=arm CROSS_COMPILE=./prebuilt/darwin-x86/toolchain/arm-eabi-4.4.3/bin/arm-eabi- HOSTCFLAGS="-I./external/elfutils/libelf"
(Note that I changed the paths here to match my system).
To save from having to recompile everything every time, I ran:
export USE_CCACHE=1 && ./prebuilt/darwin-x86/ccache/ccache -M 20G
Then I used the command identified above (i.e. make -j`sysctl -an hw.logicalcpu` ...).
It failed after quite some time with the error:
Code:
Build failed with:
Undefined symbols for architecture i386:
"android::renderscript::rsi_Allocation1DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, void const*, unsigned long, unsigned int)", referenced from:
android::renderscript::rspr_Allocation1DElementData(android::renderscript::Context*, android::renderscript::Fifo*, unsigned char*, unsigned long)in libRS.a(rsgApiReplay.o)
android::renderscript::rsp_Allocation1DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
"android::renderscript::rsi_Allocation2DElementData(android::renderscript::Context*, void*, unsigned int, unsigned int, unsigned int, RsAllocationCubemapFace, void const*, unsigned long, unsigned int)", referenced from:
android::renderscript::rspr_Allocation2DElementData(android::renderscript::Context*, android::renderscript::Fifo*, unsigned char*, unsigned long)in libRS.a(rsgApiReplay.o)
android::renderscript::rsp_Allocation2DElementData(android::renderscript::Context*, void const*, unsigned long)in libRS.a(rsgApiReplay.o)
ld: symbol(s) not found for architecture i386
collect2: ld returned 1 exit status
As I noted above, it seems like one or two other people have had this error also. At this point, I'm basically stumped. I've never really done much C/C++ dev, so I'm fairly unfamiliar with reading the output of the linker. Looking over it, I wouldn't have expected it to need i386 symbols for libRS.a (if that is indeed what it is looking for symbols of) because it is building for ARM, but I could just be completely misunderstanding the output.
Anyway, if anyone has suggestions as to what might be causing the problem or how to address it (or even wants to briefly explain what that linker error actually means in a clearer way) that would be great. I've fought with CM7 builds in the past (for the Moto Defy) and basically given up because I couldn't get them to build. I'd rather not just do that again.
Click to expand...
Click to collapse
sean.wang said:
Any update?
Click to expand...
Click to collapse
I actually gave up in the end, and took to running an Ubuntu 12.04 virtual machine for compiling. There just isn't enough people building on OSX to be able to readily work through the problems.
FFU5y said:
I actually gave up in the end, and took to running an Ubuntu 12.04 virtual machine for compiling. There just isn't enough people building on OSX to be able to readily work through the problems.
Click to expand...
Click to collapse
I believe this fixes this particular problem:
Code:
diff --git a/libs/rs/rsAllocation.cpp b/libs/rs/rsAllocation.cpp
index bbf2dbd..c4def9c 100644
--- a/libs/rs/rsAllocation.cpp
+++ b/libs/rs/rsAllocation.cpp
@@ -429,13 +429,13 @@ void rsi_Allocation1DData(Context *rsc, RsAllocation va, uint32_t xoff, uint32_t
}
void rsi_Allocation2DElementData(Context *rsc, RsAllocation va, uint32_t x, uint32_t y, uint32_t lod, RsAllocationCubemapFace face,
- const void *data, uint32_t sizeBytes, size_t eoff) {
+ const void *data, size_t sizeBytes, uint32_t eoff) {
Allocation *a = static_cast<Allocation *>(va);
a->elementData(rsc, x, y, data, eoff, sizeBytes);
}
void rsi_Allocation1DElementData(Context *rsc, RsAllocation va, uint32_t x, uint32_t lod,
- const void *data, uint32_t sizeBytes, size_t eoff) {
+ const void *data, size_t sizeBytes, uint32_t eoff) {
Allocation *a = static_cast<Allocation *>(va);
a->elementData(rsc, x, data, eoff, sizeBytes);
}
The issue has been fixed in another way, as a part of a larger patch, in the upstream AOSP. So, this is most probably only useful if you try to build CM9 / AOSP 4.0.4 under MacOS X 10.7 (Lion) with a suitable version of XCode. My XCode is 4.5 (i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1).
Hello XDA users!
I'm so confused with this thing. I've been trying to do that nearly a month. After a billion google queries and phone reboots (if my N4 could throw up, it would) I got nothing.
First of all, I can't compile the kernel.
Sys info: I'm on Arch Linux 3.12.1-3-ARCH 64 bit.
I have all the three google toolchains (arm-linux-androideabi-4.{6,7,8}) and currently I'm trying with 4.8.
With default config (I stress that) I get this: http://showterm.io/1a41b9af919653f53a890
I use google's repo (https://android.googlesource.com/kernel/msm.git) on the android-msm-mako-3.4-kitkat-mr0 branch (I use occam-krt16o build on my N4).
I know this must be noobish problem, but I have no idea what's the problem...
Thanks in advance!
I've got some news!
First of all, you need to edit scripts/gcc-wrapper.py to use python2.7 and not 3.
However, 4.8 toolchain spits some errors and I'm not really into casting spells just to compile with 4.8 instead of 4.7. If there is a major difference between them (performance improvements etc.) I'll look into it.
Now it's time to implement the framebuffer changes to the .config
for more info:
http://forum.xda-developers.com/showpost.php?p=10075806&postcount=3
http://forum.xda-developers.com/showpost.php?p=18256605&postcount=1
UPDATE: Well, I applied this patch (https://github.com/Tasssadar/kernel...a9d#diff-94babbe85a910bf17f882465a760c431R608), I can only get blank screen. Any ideas?
Bump.
Hi does anyone know how to get this to work.
I have an Alcatel OT918D ANDROID 2.3.5.
I have managed to compile my own kernel and boot it. But for adding the frame buffer option in the makefile I get this error during compilation:
**************************************************************************************************************
drivers/video/console/fbcon.c:638: note: expected 'struct file *' but argument is of type 'struct fb_info *'
drivers/video/console/fbcon.c:638: error: too few arguments to function 'info->fbops->fb_open'
drivers/video/console/fbcon.c: In function 'con2fb_release_oldinfo':
drivers/video/console/fbcon.c:670: warning: passing argument 1 of 'oldinfo->fbops->fb_release' from incompatible pointer type
drivers/video/console/fbcon.c:670: note: expected 'struct file *' but argument is of type 'struct fb_info *'
drivers/video/console/fbcon.c:670: error: too few arguments to function 'oldinfo->fbops->fb_release'
drivers/video/console/fbcon.c:673: warning: passing argument 1 of 'newinfo->fbops->fb_release' from incompatible pointer type
drivers/video/console/fbcon.c:673: note: expected 'struct file *' but argument is of type 'struct fb_info *'
drivers/video/console/fbcon.c:673: error: too few arguments to function 'newinfo->fbops->fb_release'
drivers/video/console/fbcon.c: In function 'fbcon_startup':
drivers/video/console/fbcon.c:868: warning: passing argument 1 of 'info->fbops->fb_open' from incompatible pointer type
drivers/video/console/fbcon.c:868: note: expected 'struct file *' but argument is of type 'struct fb_info *'
drivers/video/console/fbcon.c:868: error: too few arguments to function 'info->fbops->fb_open'
drivers/video/console/fbcon.c: In function 'fbcon_exit':
drivers/video/console/fbcon.c:3388: warning: passing argument 1 of 'info->fbops->fb_release' from incompatible pointer type
drivers/video/console/fbcon.c:3388: note: expected 'struct file *' but argument is of type 'struct fb_info *'
drivers/video/console/fbcon.c:3388: error: too few arguments to function 'info->fbops->fb_release'
make[3]: *** [drivers/video/console/fbcon.o] Error 1
make[2]: *** [drivers/video/console] Error 2
make[1]: *** [drivers/video] Error 2
make: *** [drivers] Error 2
make: *** Waiting for unfinished jobs....
CC [M] ../mediatek/source/kernel/drivers/lttng/probes/net-extended-trace.o
CC [M] ../mediatek/source/kernel/drivers/lttng/probes/block-trace.o
CC ../mediatek/source/kernel/kernel/trace_mt65xx_mon.o
LD [M] ../mediatek/source/kernel/drivers/lttng/ltt-relay.o
CC ../mediatek/source/kernel/drivers/m4u/m4u_kernel.o
../mediatek/source/kernel/drivers/m4u/m4u_kernel.c: In function 'm4u_user_v2p':
../mediatek/source/kernel/drivers/m4u/m4u_kernel.c:66: warning: format '%d' expects type 'int', but argument 3 has type 'struct task_struct *'
************************************************************************************************************
Code:
hardware/qcom/media-caf/mm-video/vidc/vdec/inc/frameparser.h: In constructor 'frame_parse::frame_parse()':
hardware/qcom/media-caf/mm-video/vidc/vdec/inc/frameparser.h:94:17: warning: 'frame_parse::bytes_tobeparsed' will be initialized after [-Wreorder]
hardware/qcom/media-caf/mm-video/vidc/vdec/inc/frameparser.h:67:14: warning: 'H264_Utils* frame_parse::mutils' [-Wreorder]
hardware/qcom/media-caf/mm-video/vidc/vdec/src/frameparser.cpp:76:1: warning: when initialized here [-Wreorder]
hardware/qcom/media-caf/mm-video/vidc/vdec/src/frameparser.cpp: In member function 'int frame_parse::parse_sc_frame(OMX_BUFFERHEADERTYPE*, OMX_BUFFERHEADERTYPE*, OMX_U32*)':
hardware/qcom/media-caf/mm-video/vidc/vdec/src/frameparser.cpp:215:16: warning: enumeration value 'A5' not handled in switch [-Wswitch]
hardware/qcom/media-caf/mm-video/vidc/vdec/src/frameparser.cpp:365:14: warning: enumeration value 'A4' not handled in switch [-Wswitch]
hardware/qcom/media-caf/mm-video/vidc/vdec/src/frameparser.cpp:365:14: warning: enumeration value 'A5' not handled in switch [-Wswitch]
hardware/qcom/media-caf/mm-video/vidc/vdec/test/omx_vdec_test.cpp:56:0: warning: "LOG_TAG" redefined [enabled by default]
In file included from system/core/include/cutils/log.h:1:0,
from system/core/include/utils/Vector.h:24,
from frameworks/native/include/binder/IBinder.h:23,
from frameworks/native/include/binder/Binder.h:20,
from frameworks/native/include/binder/IInterface.h:21,
from frameworks/native/include/binder/IMemory.h:26,
from frameworks/native/include/binder/MemoryHeapBase.h:23,
from hardware/qcom/media-caf/mm-video/vidc/vdec/test/omx_vdec_test.cpp:51:
system/core/include/log/log.h:68:0: note: this is the location of the previous definition
hardware/qcom/media-caf/mm-video/vidc/vdec/test/omx_vdec_test.cpp:84:27: fatal error: linux/msm_mdp.h: No such file or directory
compilation terminated.
target thumb C++: libOmxVdec <= hardware/qcom/media-caf/mm-video/vidc/vdec/src/mp4_utils.cpp
make: *** [/home/zeroblade1984/Android/omni/out/target/product/m7ul/obj/EXECUTABLES/mm-vdec-omx-test_intermediates/test/omx_vdec_test.o] Error 1
make: *** Waiting for unfinished jobs....
In file included from hardware/qcom/media-caf/mm-video/vidc/vdec/src/h264_utils.cpp:49:0:
hardware/qcom/media-caf/mm-video/vidc/common/inc/extra_data_handler.h:37:31: fatal error: linux/msm_vidc_dec.h: No such file or directory
compilation terminated.
make: *** [/home/zeroblade1984/Android/omni/out/target/product/m7ul/obj/SHARED_LIBRARIES/libOmxVdec_intermediates/src/h264_utils.o] Error 1
In file included from hardware/qcom/media-caf/mm-video/vidc/vdec/src/mp4_utils.cpp:29:0:
hardware/qcom/media-caf/mm-video/vidc/vdec/inc/omx_vdec.h:53:27: fatal error: linux/msm_ion.h: No such file or directory
compilation terminated.
make: *** [/home/zeroblade1984/Android/omni/out/target/product/m7ul/obj/SHARED_LIBRARIES/libOmxVdec_intermediates/src/mp4_utils.o] Error 1
real 6m45.287s
user 13m46.486s
sys 4m52.261s
i have been wretched all day building Omni for M7 and i am plagued by the errors above..been googling around to see if there's a way but to no futile response..hope some can help with this..just repo sync'd everything before i built
I'm really confused, that looks like a mismatch between the kernel and media-caf (indicating the kernel needs updating), but...
m7ul is successfully building? http://dl.omnirom.org/m7ul/
Did you perhaps accidentally sync a kernel, device, or media-caf tree from CM????
Entropy512 said:
I'm really confused, that looks like a mismatch between the kernel and media-caf (indicating the kernel needs updating), but...
m7ul is successfully building? http://dl.omnirom.org/m7ul/
Did you perhaps accidentally sync a kernel, device, or media-caf tree from CM????
Click to expand...
Click to collapse
i had it building..yes seems you are right..i accidentlly synced CM kernel..lolz..i have one question though..how to i remove the advanced settings package??as i have cleaned the kernel source and updated a few stuff..also updated some blobs and from my testing point everything is working..except that off course the Advanced settings is left out gray since i removed most of the stuff and left the kernel stockish..
kairi_zeroblade said:
i accidentlly synced CM kernel..lolz.
Click to expand...
Click to collapse
How dafuq can you "accidentally" sync a wrong kernel? smh
chasmodo said:
How dafuq can you "accidentally" sync a wrong kernel? smh
Click to expand...
Click to collapse
i alread had the manifests setup before and YES i did some experimenting on my own..good thing Entropy reminded me of that mistake.. :silly:
So I'm just trying to build CM14(I'm well aware it's basically stock android at this point), and I'm getting quite far in the build process, but getting some errors. I was hoping someone a little more experienced might be able to give me some insight on what the issue is.
Here's the most recent bit from the console:
Code:
/home/kronflux/android/out/host/linux-x86/bin/checkpolicy: loading policy configuration from /home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/policy_recovery.conf
host Java: dumpkey (/home/kronflux/android/out/host/common/obj/JAVA_LIBRARIES/dumpkey_intermediates/classes)
device/huawei/angler/sepolicy/file.te:30:ERROR 'duplicate declaration of type/attribute' at token ';' on line 21108:
type sysfs_thermal, sysfs_type, fs_type;
type sysfs_surfaceflinger, fs_type, sysfs_type;
checkpolicy: error(s) encountered while parsing configuration
system/sepolicy/Android.mk:140: recipe for target '/home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/sepolicy.recovery' failed
make: *** [/home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/sepolicy.recovery] Error 1
make: *** Waiting for unfinished jobs....
target Export Resources: framework-res (/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk)
nothing matches overlay file default_wallpaper.png, for flavor hdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xhdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xxhdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xxxhdpi-v4
frameworks/base/core/res/res/values/cm_symbols.xml:21: warning: <private-symbols> is deprecated. Use the command line flag --private-symbols instead.
frameworks/base/core/res/res/values/cm_symbols.xml:21: warning: private symbol package already specified. Ignoring...
warning: string 'allow_button' has no default translation.
warning: string 'deny_button' has no default translation.
warning: string 'gsm_alphabet_default_charset' has no default translation.
warning: string 'other_permissions' has no default translation.
warning: string 'permission_title' has no default translation.
res/drawable-nodpi-v4/default_wallpaper.png:0: error: Resource entry default_wallpaper is already defined.
res/drawable-nodpi-v4/default_wallpaper.jpg:0: Originally defined here.
device/huawei/angler/overlay/frameworks/base/core/res/res/values/config.xml:29: error: Resource at config_orientationSensorType appears in overlay but not in the base package; use <add-resource> to add.
build/core/package_internal.mk:389: recipe for target '/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk' failed
make: *** [/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk] Error 1
make: *** Deleting file '/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk'
make: Leaving directory '/home/kronflux/android'
#### make failed to build some targets (04:00:18 (hh:mm:ss)) ####
kronflux said:
So I'm just trying to build CM14(I'm well aware it's basically stock android at this point), and I'm getting quite far in the build process, but getting some errors. I was hoping someone a little more experienced might be able to give me some insight on what the issue is.
Here's the most recent bit from the console:
Code:
/home/kronflux/android/out/host/linux-x86/bin/checkpolicy: loading policy configuration from /home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/policy_recovery.conf
host Java: dumpkey (/home/kronflux/android/out/host/common/obj/JAVA_LIBRARIES/dumpkey_intermediates/classes)
device/huawei/angler/sepolicy/file.te:30:ERROR 'duplicate declaration of type/attribute' at token ';' on line 21108:
type sysfs_thermal, sysfs_type, fs_type;
type sysfs_surfaceflinger, fs_type, sysfs_type;
checkpolicy: error(s) encountered while parsing configuration
system/sepolicy/Android.mk:140: recipe for target '/home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/sepolicy.recovery' failed
make: *** [/home/kronflux/android/out/target/product/angler/obj/ETC/sepolicy.recovery_intermediates/sepolicy.recovery] Error 1
make: *** Waiting for unfinished jobs....
target Export Resources: framework-res (/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk)
nothing matches overlay file default_wallpaper.png, for flavor hdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xhdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xxhdpi-v4
nothing matches overlay file default_wallpaper.png, for flavor xxxhdpi-v4
frameworks/base/core/res/res/values/cm_symbols.xml:21: warning: <private-symbols> is deprecated. Use the command line flag --private-symbols instead.
frameworks/base/core/res/res/values/cm_symbols.xml:21: warning: private symbol package already specified. Ignoring...
warning: string 'allow_button' has no default translation.
warning: string 'deny_button' has no default translation.
warning: string 'gsm_alphabet_default_charset' has no default translation.
warning: string 'other_permissions' has no default translation.
warning: string 'permission_title' has no default translation.
res/drawable-nodpi-v4/default_wallpaper.png:0: error: Resource entry default_wallpaper is already defined.
res/drawable-nodpi-v4/default_wallpaper.jpg:0: Originally defined here.
device/huawei/angler/overlay/frameworks/base/core/res/res/values/config.xml:29: error: Resource at config_orientationSensorType appears in overlay but not in the base package; use <add-resource> to add.
build/core/package_internal.mk:389: recipe for target '/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk' failed
make: *** [/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk] Error 1
make: *** Deleting file '/home/kronflux/android/out/target/common/obj/APPS/framework-res_intermediates/package-export.apk'
make: Leaving directory '/home/kronflux/android'
#### make failed to build some targets (04:00:18 (hh:mm:ss)) ####
Click to expand...
Click to collapse
you need to go to overlay/frameworks/base/core/res/res/values/config.xml at you device tree and find "config_orientationSensorType", delete it or change it as comment with <!-- -->
Also you will find major probs at sepolicies when building
leskal said:
you need to go to overlay/frameworks/base/core/res/res/values/config.xml at you device tree and find "config_orientationSensorType", delete it or change it as comment with <!-- -->
Also you will find major probs at sepolicies when building
Click to expand...
Click to collapse
I ended up getting it to build last night with various errors and poking around to fix it.
Unfortunately it wouldn't boot, so who knows what else needs to be done. I guess I'll just have to wait for the pros to get it up and running, and stick with AOSP for now.
Thanks for the reply though.