[Android-x86 ] Notice: Build broken is fixed

[Chih-Wei]

The android-x86 build broken due to upstream changes (donut merged) is
fixed.
Now you can get a workable tree and enjoy Donut on x86.

You may try to repo sync. But the changes are huge so I'm not sure
whether you can succeed or not. If you encounter conflicts but don't
know how to fix, I suggest you redo repo init and then sync a fresh
tree.

Besides, wifi support has been integrated. Ath5k, ath9k modules work
fine. Other drivers need further testing. As always, test results are
welcome.

--
Chih-Wei
Android-x86
http://code.google.com/p/android-x86/

[Yi Sun]

FYI-- for the wifi, if your driver needs a firmware, it may not work at all.  And we are working on to create a working sample so that people can add the firmware by their self.
Yi

--
Android-x86
http://code.google.com/p/android-x86/

[Chen Yang]

Cool. would you like to provide some more detail about the issue? Thanks.
--
Chen

[Yi Sun]

Chih-wei and I are trying to find a good way to organise the firmware directories.

--
Android-x86
http://code.google.com/p/android-x86/

[Chen Yang]

Just finished one round of proof build of the public AOSP code:
here is the summary:
1.missing one _BYTE_ORDER define in bionic/arch-x86/include/machine/_types.h

--- a/libc/arch-x86/include/machine/_types.h
+++ b/libc/arch-x86/include/machine/_types.h
@@ -119,5 +119,6 @@ typedef    void *            __wctype_t;
 /* Feature test macros */
 #define __HAVE_CPUINFO
 #define __HAVE_MUTEX
+#define _BYTE_ORDER _LITTLE_ENDIAN
 
 #endif    /* _I386__TYPES_H_ */

2. some problem on x86 by recent check in dalvik:
project dalvik/
diff --git a/vm/oo/Class.c b/vm/oo/Class.c
index a5d42eb..c47dff4 100644
--- a/vm/oo/Class.c
+++ b/vm/oo/Class.c
@@ -4276,13 +4276,13 @@ noverify:
              */
             assert(f->byteOffset >= CLASS_SMALLEST_OFFSET);
             assert((f->byteOffset & (CLASS_OFFSET_ALIGNMENT - 1)) == 0);
+#define CLASS_BIT_SHIFT_COUNT(byteOffset) (((unsigned int)(byteOffset) - CLASS_SMALLEST_OFFSET) /  CLASS_OFFSET_ALIGNMENT)
+        if(CLASS_BIT_SHIFT_COUNT(f->byteOffset) >= CLASS_BITS_PER_WORD) {
+        clazz->refOffsets = CLASS_WALK_SUPER;
+        break;
+        }
             u4 newBit = CLASS_BIT_FROM_OFFSET(f->byteOffset);
-            if (newBit != 0) {
-                clazz->refOffsets |= newBit;
-            } else {
-                clazz->refOffsets = CLASS_WALK_SUPER;
-                break;
-            }
+            clazz->refOffsets |= newBit;
             f++;
         }
     }

3. applypatch depends on libmtdutils.a, which doesn't exsit on x86. so maybe can disable it.
project build/
diff --git a/tools/applypatch/Android.mk b/tools/applypatch/Android.mk
index fe317ff..8f494b6 100644
--- a/tools/applypatch/Android.mk
+++ b/tools/applypatch/Android.mk
@@ -14,6 +14,7 @@
 
 ifneq ($(TARGET_SIMULATOR),true)
 
+ifeq ($(TARGET_ARCH),arm)
 LOCAL_PATH := $(call my-dir)
 include $(CLEAR_VARS)
 
@@ -48,4 +49,5 @@ LOCAL_STATIC_LIBRARIES += libz
 
 include $(BUILD_HOST_EXECUTABLE)
 
+endif   # TARGET_ARCH == arm
 endif  # !TARGET_SIMULATOR

4.  preloaded-classes/init.rc fixes originally known to x86.

hope it helps.
--
Chen

[Yi Sun]

Have you run the image? There should be more issues, the 10482 was not fixed yet (not sure about today's status), the init.rc has new entries need to add ....
Yi

--
Android-x86
http://code.google.com/p/android-x86/

[Chen Yang]

yes, I have run it.
10482 has 2 issues, my 2nd item fixed one issue. I don't know whether current framework has the right kind of implementation that may expose the 2nd issue of that. At least, from what I have run, I haven't found problems so far.
--
Chen

[Yi Sun]

Chihwei backed out both from android-x86

[penguins]

Hi,

I downloaded android source code for x86 today and tried to compile
it, but failed, here is the error that I received:

$TARGET_ARCH=x86 TARGET_PRODUCT=eee_701 DISABLE_DEXPREOPT=true make -
j2 installer_img
============================================
PLATFORM_VERSION_CODENAME=Donut
PLATFORM_VERSION=Donut
TARGET_PRODUCT=eee_701
TARGET_BUILD_VARIANT=eng
TARGET_SIMULATOR=
TARGET_BUILD_TYPE=release
TARGET_ARCH=x86
HOST_ARCH=x86
HOST_OS=linux
HOST_BUILD_TYPE=release
BUILD_ID=Donut
============================================
build/core/main.mk:458: *** No TARGET_CPU_ABI defined by board
config: . Stop.

Is this related?

Thanks.

P

On Aug 2, 6:12 pm, Yi Sun <beyo...@gmail.com> wrote:
> Chihwei backed out both from android-x86
>

> On 2009-8-2, at 16:54, Chen Yang <sunsety...@gmail.com> wrote:
>
>
>
> > yes, I have run it.
> > 10482 has 2 issues, my 2nd item fixed one issue. I don't know  
> > whether current framework has the right kind of implementation that  
> > may expose the 2nd issue of that. At least, from what I have run, I  
> > haven't found problems so far.
> > --
> > Chen
>

> > On Mon, Aug 3, 2009 at 2:10 AM, Yi Sun <beyo...@gmail.com> wrote:
> > Have you run the image? There should be more issues, the 10482 was  
> > not fixed yet (not sure about today's status), the init.rc has new  
> > entries need to add ....
> > Yi
>

> > On Sun, Aug 2, 2009 at 3:21 AM, Chen Yang <sunsety...@gmail.com>  

> > On Thu, Jul 30, 2009 at 3:10 PM, Chen Yang <sunsety...@gmail.com>  

> > wrote:
> > Cool. would you like to provide some more detail about the issue?  
> > Thanks.
> > --
> > Chen
>

> > On 7/30/09, Chih-Wei <cwhu...@linux.org.tw> wrote:
>
> > The android-x86 build broken due to upstream changes (donut merged) is
> > fixed.
> > Now you can get a workable tree and enjoy Donut on x86.
>
> > You may try to repo sync. But the changes are huge so I'm not sure
> > whether you can succeed or not. If you encounter conflicts but don't
> > know how to fix, I suggest you redo repo init and then sync a fresh
> > tree.
>
> > Besides, wifi support has been integrated. Ath5k, ath9k modules work
> > fine. Other drivers need further testing. As always, test results are
> > welcome.
>
> > --
> > Chih-Wei
> > Android-x86
> >http://code.google.com/p/android-x86/
>
> > --
> > Android-x86

> >http://code.google.com/p/android-x86/- Hide quoted text -
>
> - Show quoted text -

[Yi Sun]

You are using wrong command.

Take a look on the wiki page:
http://code.google.com/p/android-x86/wiki/GetSourceCode

--
Android-x86
http://code.google.com/p/android-x86/

[Chen Yang]

I missed one item in BoardConfig.mk
diff --git a/BoardConfig.mk b/BoardConfig.mk
index 54b0d4f..68000ce 100644
--- a/BoardConfig.mk
+++ b/BoardConfig.mk
@@ -8,6 +8,7 @@ TARGET_NO_BOOTLOADER := true
 TARGET_NO_RECOVERY := true
 
 #TARGET_NO_KERNEL := true
+TARGET_CPU_ABI := none
 
 TARGET_HARDWARE_3D := false

Some more items:
diff --git a/eee_701.mk b/eee_701.mk
index 508108c..1dd6558 100644
--- a/eee_701.mk
+++ b/eee_701.mk
@@ -4,7 +4,7 @@
 # For now, reuse the generic (phone) infrastructure.
 #
 
-$(call inherit-product, $(SRC_TARGET_DIR)/product/generic_with_google.mk)
+$(call inherit-product, $(SRC_TARGET_DIR)/product/generic.mk)
 
 PRODUCT_NAME := eee_701
 PRODUCT_DEVICE := eee_701
diff --git a/init.rc b/init.rc
index 00aa93c..3cdcedc 100644
--- a/init.rc
+++ b/init.rc
@@ -108,6 +108,8 @@ on boot
     setprop ro.FOREGROUND_APP_ADJ 0
     setprop ro.VISIBLE_APP_ADJ 1
     setprop ro.SECONDARY_SERVER_ADJ 2
+    setprop ro.BACKUP_APP_ADJ 2
+    setprop ro.HOME_APP_ADJ 4
     setprop ro.HIDDEN_APP_MIN_ADJ 7
     setprop ro.CONTENT_PROVIDER_ADJ 14
     setprop ro.EMPTY_APP_ADJ 15
@@ -117,9 +119,12 @@ on boot
     setprop ro.FOREGROUND_APP_MEM 1536
     setprop ro.VISIBLE_APP_MEM 2048
     setprop ro.SECONDARY_SERVER_MEM 4096
-    setprop ro.HIDDEN_APP_MEM 8192
-    setprop ro.CONTENT_PROVIDER_MEM 8704
-    setprop ro.EMPTY_APP_MEM 16384
+    setprop ro.BACKUP_APP_MEM 4096
+    setprop ro.HOME_APP_MEM 4096
+    setprop ro.HIDDEN_APP_MEM 5120
+    setprop ro.CONTENT_PROVIDER_MEM 5632
+    setprop ro.EMPTY_APP_MEM 6144
+
 
 # Write value must be consistent with the above properties.
     write /sys/module/lowmemorykiller/parameters/adj 0,1,2,7,14,15
@@ -136,6 +141,12 @@ on boot
     chown radio system /sys/android_power/acquire_full_wake_lock
     chown radio system /sys/android_power/acquire_partial_wake_lock
     chown radio system /sys/android_power/release_wake_lock
+    chown radio system /sys/power/state
+    chown radio system /sys/power/wake_lock
+    chown radio system /sys/power/wake_unlock
+    chmod 0660 /sys/power/state
+    chmod 0660 /sys/power/wake_lock
+    chmod 0660 /sys/power/wake_unlock
     chown system system /sys/class/timed_output/vibrator/enable
     chown system system /sys/class/leds/keyboard-backlight/brightness
     chown system system /sys/class/leds/lcd-backlight/brightness
@@ -193,8 +204,8 @@ service servicemanager /system/bin/servicemanager
     onrestart restart zygote
     onrestart restart media
 
-service mountd /system/bin/mountd
-    socket mountd stream 0660 root mount
+service vold /system/bin/vold
+    socket vold stream 0660 root mount
 
 service zygote /system/bin/app_process -Xzygote /system/bin --zygote --start-system-server
     socket zygote stream 666


Meanwhile, in order to build the installer_img, don't forget to enable external/e2fsprogs/Android.mk remove the # before the include.

Hope it helps.
--
Chen

[rajesh123]

HI All,

I am also experiencing this error when I am building android for ARM

=========================================
PLATFORM_VERSION_CODENAME=REL
PLATFORM_VERSION=1.6
TARGET_PRODUCT=ldp1
TARGET_BUILD_VARIANT=eng
TARGET_SIMULATOR=
TARGET_BUILD_TYPE=release
TARGET_ARCH=arm

HOST_ARCH=x86
HOST_OS=linux
HOST_BUILD_TYPE=release
BUILD_ID=Donut
============================================
build/core/main.mk:458: *** No TARGET_CPU_ABI defined by board
config: . Stop.

I note that the HOST ARCH is displayed as x86 but i am running from a
dual core AMD athlon.

please let me know if there are any solutions .

Regards
S.Rajesh Kumar

On Aug 5, 1:48 am, penguins <suf...@gmail.com> wrote:
> Hi,
>
> I downloaded android source code for x86 today and tried to compile
> it, but failed, here is the error that I received:
>
> $TARGET_ARCH=x86 TARGET_PRODUCT=eee_701 DISABLE_DEXPREOPT=true make -
> j2 installer_img
> ============================================
> PLATFORM_VERSION_CODENAME=Donut
> PLATFORM_VERSION=Donut
> TARGET_PRODUCT=eee_701
> TARGET_BUILD_VARIANT=eng
> TARGET_SIMULATOR=
> TARGET_BUILD_TYPE=release
> TARGET_ARCH=x86
> HOST_ARCH=x86
> HOST_OS=linux
> HOST_BUILD_TYPE=release
> BUILD_ID=Donut

> ============================================build/core/main.mk:458: *** No TARGET_CPU_ABI defined by board

> config: .  Stop.
>
> Is this related?
>
> Thanks.
>
> P
>
> On Aug 2, 6:12 pm, Yi Sun <beyo...@gmail.com> wrote:
>
> > Chihwei backed out both from android-x86
>
> > On 2009-8-2, at 16:54, Chen Yang <sunsety...@gmail.com> wrote:
>
> > > yes, I have run it.

> > > 10482 has 2 issues, my 2nd itemfixedone issue. I don't know  

> > > whether current framework has the right kind of implementation that  
> > > may expose the 2nd issue of that. At least, from what I have run, I  
> > > haven't found problems so far.
> > > --
> > > Chen
>
> > > On Mon, Aug 3, 2009 at 2:10 AM, Yi Sun <beyo...@gmail.com> wrote:
> > > Have you run the image? There should be more issues, the 10482 was  

> > > notfixedyet (not sure about today's status), the init.rc has new  

> > > entries need to add ....
> > > Yi
>
> > > On Sun, Aug 2, 2009 at 3:21 AM, Chen Yang <sunsety...@gmail.com>  
> > > wrote:

> > > Just finished one round of proofbuildof the public AOSP code:

> > > The android-x86buildbrokendue to upstream changes (donut merged) is

> > >fixed.
> > > Now you can get a workable tree and enjoy Donut on x86.
>
> > > You may try to repo sync. But the changes are huge so I'm not sure
> > > whether you can succeed or not. If you encounter conflicts but don't
> > > know how to fix, I suggest you redo repo init and then sync a fresh
> > > tree.
>
> > > Besides, wifi support has been integrated. Ath5k, ath9k modules work
> > > fine. Other drivers need further testing. As always, test results are
> > > welcome.
>
> > > --
> > > Chih-Wei
> > > Android-x86
> > >http://code.google.com/p/android-x86/
>
> > > --
> > > Android-x86

> > >http://code.google.com/p/android-x86/-Hide quoted text -

[Chih-Wei]

As the error message said, you need to define
TARGET_CPU_ABI in your product makefile,
usually BoardConfig.mk.

PS. Your question is unrelated to the original thread.
Please open another thread next time.