Re: Android on N810 File System Problem

[Dirk Behme <dirk.behme@googlemail.com>]

Jonathan Herriott wrote:
> I'm using vanilla Kernel 2.6.21.0 with the patch from
> http://android-on-n8xx.googlecode.com/files/linux-2.6.21_rx-34_android-m5-rc14.bz2.
>  I was told in another thread that I need to use the m3 userspace
> files, but when trying to use those, I have the same issue.
> 
> I'm assuming you want the strace when running init:
> 
> \h:\w\$ strace -f -ff -tt -s 200 /init
> 04:02:38.896362 execve("/init", ["/init"], [/* 59 vars */]) = 0
> 04:02:38.909790 gettid()                = 1464
> 04:02:38.912689 syscall_983045(0xbea0d614, 0, 0x20, 0, 0xbe9ee000,
> 0xbea0d6d0, 0xbea0d720, 0xf0005, 0xbea0d720, 0, 0x80b8, 0x80b4, 0,
> 0xbea0d608, 0x17c77, 0x17f2c, 0x60000010, 0xbea0d614, 0, 0, 0xc764, 0,
> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) = 0
> 04:02:38.916168 socket(PF_FILE, SOCK_STREAM, 0) = 3
> 04:02:38.919250 connect(3, {sa_family=AF_FILE,
> path=@property_service}, 19) = -1 ECONNREFUSED (Connection refused)
> 04:02:38.923248 close(3)                = 0
> 04:02:38.926208 sigaction(SIGCHLD, {0x8211, [], SA_NOCLDSTOP}, NULL,
> 0xc0fb8) = 0
> 04:02:38.929718 umask(0)                = 022
> 04:02:38.932373 ioctl(1, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost
> isig icanon echo ...}) = 0
> 04:02:38.936523 dup(1)                  = 3
> 04:02:38.939422 write(3, "init: HOW ARE YOU GENTLEMEN\n", 28init: HOW
> ARE YOU GENTLEMEN
> ) = 28
> 04:02:38.943328 close(0)                = 0
> 04:02:38.945983 close(1)                = 0
> 04:02:38.948577 close(2)                = 0
> 04:02:38.951202 open("/dev/null", O_RDWR|O_LARGEFILE) = 0
> 04:02:38.954345 dup2(0, 1)              = 1
> 04:02:38.956939 dup2(0, 2)              = 2
> 04:02:38.959625 write(3, "init: reading config file\n", 26init:
> reading config file
> ) = 26
> 04:02:38.963592 brk(0)                  = 0x20000
> 04:02:38.966217 brk(0x20000)            = 0x20000
> 04:02:38.968994 brk(0x21000)            = 0x21000
> 04:02:38.971832 open("/etc/init.rc", O_RDONLY|O_LARGEFILE) = 4
> 04:02:38.975006 lseek(4, 0, SEEK_END)   = 5746
> 04:02:38.991088 lseek(4, 0, SEEK_SET)   = 0
> 04:02:38.993286 brk(0x22000)            = 0x22000
> 04:02:38.998413 read(4, "## Global environment setup\n##\nenv {\n
> PATH /sbin:/system/sbin:/system/bin\n    LD_LIBRARY_PATH /system/lib
> \n    ANDROID_BOOTLOGO 1\n    ANDROID_ROOT /system\n
> ANDROID_ASSETS /system/app\n  ANDROID_"..., 5746) = 5746
> 04:02:39.002288 close(4)                = 0
> 04:02:39.003845 brk(0x23000)            = 0x23000
> 04:02:39.005310 mkdir("/proc", 0755)    = -1 EEXIST (File exists)
> 04:02:39.006622 mkdir("/dev", 0755)     = -1 EEXIST (File exists)
> 04:02:39.007843 mkdir("/dev/pts", 0755) = -1 EEXIST (File exists)
> 04:02:39.009368 mkdir("/sys", 0755)     = -1 EEXIST (File exists)
> 04:02:39.011230 mkdir("/d", 0755)       = -1 EEXIST (File exists)
> 04:02:39.012451 mount("/dev/pts", "/dev/pts", "devpts", 0, NULL) = -1
> EBUSY (Device or resource busy)
> 04:02:39.017883 mount("/proc", "/proc", "proc", 0, NULL) = -1 EBUSY
> (Device or resource busy)
> 04:02:39.026428 mount("/sys", "/sys", "sysfs", 0, NULL) = -1 EBUSY
> (Device or resource busy)
> 04:02:39.033416 mount("/d", "/d", "debugfs", 0, NULL) = -1 EBUSY
> (Device or resource busy)
> 04:02:39.038238 mount("/tmp", "/tmp", "tmpfs", 0, NULL) = 0
> 04:02:39.043457 open("/proc/cmdline", O_RDONLY|O_LARGEFILE) = 4
> 04:02:39.046417 read(4, "root=1f03 rootfstype=jffs2 ro\n", 1023) = 30
> 04:02:39.048187 close(4)                = 0
> 04:02:39.049346 chmod("/proc/cmdline", 0400) = 0
> 04:02:39.050689 open("/system_properties", O_RDWR|O_CREAT|O_EXCL|
> O_LARGEFILE, 0600) = 4
> 04:02:39.052368 ftruncate(4, 32768)     = 0
> 04:02:39.053588 open("/system_properties", O_RDONLY|O_LARGEFILE) = 5
> 04:02:39.056884 unlink("/system_properties") = 0
> 04:02:39.058624 mmap2(NULL, 32768, PROT_READ|PROT_WRITE, MAP_SHARED,
> 4, 0) = -1 EINVAL (Invalid argument)
> 04:02:39.059997 close(5)                = 0
> 04:02:39.061096 close(4)                = 0
> 04:02:39.063110 open("/etc/default.prop", O_RDONLY|O_LARGEFILE) = 4
> 04:02:39.064422 lseek(4, 0, SEEK_END)   = 130
> 04:02:39.065521 lseek(4, 0, SEEK_SET)   = 0
> 04:02:39.066619 read(4, "# default system properties\n# these may be
> overridden by /data/local.prop\n\nnet.bt.name = Android\n\n# end
> default system properties\n", 130) = 130
> 04:02:39.069152 close(4)                = 0
> 04:02:39.070251 --- SIGSEGV (Segmentation fault) @ 0 (0) ---
> 04:02:39.072601 +++ killed by SIGSEGV +++
> Process 1464 detached

Mmh, there are some reports on this list with similar SIGSEGV:

http://marc.info/?l=linux-omap&m=120393103218574&w=2
http://marc.info/?l=linux-omap&m=120400327301855&w=2
http://marc.info/?l=linux-omap&m=120834030203728&w=2

But It doesn't look like the archives have a detailed solution for 
this. Seems that changing some parameters (e.g. switching to other 
binder driver, SDK etc.) made the error go away.

I'm CCing Anil and colorant, maybe they remember what helped.

Dirk

> On Thu, Apr 24, 2008 at 10:42 PM, Dirk Behme <dirk.behme@googlemail.com> wrote:
> 
>>Jonathan,
>>
>> CCing OMAP mailing list, there are more people to help.
>>
>>
>> Jonathan Herriott wrote:
>>
>>
>>>Hi,
>>>
>>>I'm trying to get android installed on an N810.  I have successfully
>>>installed the kernel,
>>>
>>
>> Which kernel (2.6.x, x == ?)  do you use on N810? If I remember correctly,
>>with N810 people use an older kernel than the one coming with Android SDK.
>>
>>
>>
>>
>>>but I am having major issues with the file
>>>system.  I followed the tutorial you provided at
>>>http://elinux.org/Android_on_OMAP, but I get a segmentation fault when
>>>I try to init the kernel.  I made sure to use the same android SDK
>>>version that was used to patch the kernel.
>>>
>>>My steps for reproduction are:
>>>
>>>1) On my computer, unpack the ramdisk image to a folder.
>>>2) Untar system, dev, and data to the unpacked ramdisk image.
>>>3) Tar up the patched file system and scp it to /opt on the device.
>>>Ssh to the device and unpack the file system in /opt
>>>4) Run umask 000 && chroot /opt/android /init
>>>5) Receive a segmentation fault.
>>>
>>
>> Can you send us the complete segementation fault output?
>>
>> Regards
>>
>> Dirk
>>
>>
>>
> 
>