See whether this patch would help. https://review.source.android.com/Gerrit#change,6751 -- Chen
On Tue, Mar 31, 2009 at 11:27 PM, Brian <[email protected]> wrote: > > I got more detailed error messages. > > Console: switching to colour frame buffer device 100x37 > ----------------- [cut here] ------------------- > kernel BUG at kernel/timer.c: 488! > invalid opcode: 0000 [#1] PREEMPT SMP > Modules linked in: fbcon i915 drm atl2 > > pid:0, comm: swapper: Not tainted (2.6.25 ... > > ... > > Kernel panic - not syncing: Fatal except in interrupt > > Thanks, > Brian > > > On Mar 31, 8:02 am, Brian <[email protected]> wrote: >> I tried vmware 6.5 workstation on windows 32 bit, vmware 6.5 >> workstation on fedora core 8 64bit. >> Both did not work. >> >> I got where the kernel panic's happening. >> kernel 2.6.25 kernel/timer.c:488 >> >> int mod_timer(...) >> { >> BUG_ON(!timer->function) <--- here >> >> What can be a fix? Any suggestion? >> >> Thanks,Brian >> >> On Mar 30, 10:54 pm,Brian<[email protected]> wrote: >> >> > Hi Chen, >> >> > I managed to compile kernel 2.6.25. >> > When I ran it without vga=788, I see a shell prompt. >> > The screen does not go to black one. >> > But I keep getting the following message. >> > [timestamp] android_power: wakeup (0->0) at ... >> > (Btw, how can I turn off this message?) >> >> > I still have a problem with vga=788. >> > This time I'm getting a kernel panic, which I'm attaching. >> > Has anyone seen this problem? What's a proper approach to address this >> > problem? >> >> > Thanks,Brian >> >> > Process netcfg (pid: 2031, ti=cf17c000 task=cf3001b0 task.ti=cf17c000) >> > Stack: ... >> >> > Call Trace: >> > [<c01f12f2>] ? cursor_timer_handler+0x28/0x2b >> > ? run_timer_softirq+0x11d/0x176 >> > ? cursor_timer_handler+0x0/0x2b >> > ? __do_softirq+0x70/0xd5 >> > ? do_softirq+0x35/0x3a >> > ? local_bh_enable_ip+0x38/0x4f >> > ? _spin_unlock_bh+0x17/0x19 >> > ? dev_set_rx_mode+0x38/0x3c >> > ? dev_open+0x60/0x7e >> > ? dev_change_flags+0x9f/0x154 >> > ? devinet_ioctl+0x2da/0x5f8 >> > ? inet_ioctl+0x93/0xaf >> > ? sock_ioctl+0x1af/0x1d1 >> > ? sock_ioctl+0x0/0x1d1 >> > ? vfs_ioctl+0x22/0x69 >> > ? do_vfs_ioctl+0x222/0x235 >> > ? sys_socketcall+0x51/0x168 >> > ? sys_ioctl+0x40/0x5a >> > ? syscall_call+0x7/0xb >> > Code: .... >> > EIP: [<c012b22a>] mod_timer+0x9/0x27 SS:ESP 0068:cf17de18 >> > Kernel panic - not syncing: Fatal exception in interrupt >> >> > On Mar 30, 8:49 am,Brian<[email protected]> wrote: >> >> > > Thanks, Jerry! >> >> > > I'm using kernel 2.6.27, and the modules have version 2.6.25 as you >> > > can see below. >> > > There are inconsistencies in the cupcake branch. >> >> > > I will either get kernel 2.6.25 or try to compile 2.6.27 modules. >> >> > > Thanks,Brian >> >> > > ./kernel/include/config/kernel.release:2.6.27-00120-g453dc2b-dirty >> >> > > $ modinfo ./out/target/product/eee_701/rootimg/lib/modules/i915.ko >> > > filename: ./out/target/product/eee_701/rootimg/lib/modules/ >> > > i915.ko >> > > license: GPL and additional rights >> > > description: Intel Graphics >> > > author: Tungsten Graphics, Inc. >> > > license: GPL >> > > srcversion: 73F81EDC4CAE593AA695C49 >> > > depends: drm,cfbfillrect,cfbimgblt,cfbcopyarea >> > > vermagic: 2.6.25-00101-gb6922fa SMP preempt mod_unload PENTIUMM >> > > parm: modeset:int >> > > parm: fbpercrtc:int >> > > parm: i915_rightof:int >> >> > > On Mar 30, 8:39 am, Jerry Fan <[email protected]> wrote: >> >> > > > HiBrian, >> > > > just type modinfo /path/to/anymodule.ko >> > > > compare modinfo output on any kernel ko files in your Android kernel >> > > > against >> > > > i915 from eeepc. In modinfo output, u will read vermagic at last >> > > > line. You >> > > > need to make sure every protperty in i915 vermagic matches vermagic in >> > > > your >> > > > android kernel. Any signle property mismatches will result in the >> > > > problem >> > > > you have. >> >> > > > On Mon, Mar 30, 2009 at 11:15 PM,Brian<[email protected]> wrote: >> >> > > > > Thanks, Jerry! >> >> > > > > (I'm very sorry. I somehow got a wrong name.) >> >> > > > > Please see the previous email. >> >> > > > > Thanks, >> > > > >Brian >> >> > > > > On Mar 30, 8:14 am,Brian<[email protected]> wrote: >> > > > > > Thanks, Jeff. >> >> > > > > > It seems like that the modules are copied from vendor/asus/eee_701/ >> > > > > > modules >> > > > > > How can I check magic numbers? >> >> > > > > > I'm attaching files in the directory. There is the module.versions >> > > > > > file and drm-patches. Maybe you can pinpoint what's going on with >> > > > > > these files better. >> >> > > > > > Thanks, >> > > > > >Brian >> >> > > > > > vendor/asus/eee_701/modules$ ls >> > > > > > atl2.ko cfbcopyarea.ko cfbimgblt.ko drm-patches font.ko >> > > > > > module.versions >> > > > > > bitblit.ko cfbfillrect.ko drm.ko fbcon.ko i915.ko >> > > > > > softcursor.ko >> >> > > > > > vendor/asus/eee_701/modules$ cat module.versions >> > > > > > Modules: >> > > > > > atl2.ko >> >> > > > > > Source: >> > > > > > http://people.redhat.com/csnook/atl2/atl2-2.0.5.tar.bz2 >> >> > > > > > ---------------------------- >> >> > > > > > Modules: >> > > > > > bitblit.ko >> > > > > > cfbcopyarea.ko >> > > > > > cfbfillrect.ko >> > > > > > cfbimgblt.ko >> > > > > > fbcon.ko >> > > > > > font.ko >> > > > > > softcursor.ko >> >> > > > > > Source: >> > > > > > git://android.git.kernel.org/kernel/common.git android-2.6.25 >> >> > > > > > commit b6922faa7d617f11440ab8096b8797ccdcebd10d >> > > > > > Author: Nick Pelly <[email protected]> >> > > > > > Date: Mon Nov 10 21:59:54 2008 -0800 >> >> > > > > > android_paranoid_network: Use in_egroup_p to check gid's. >> >> > > > > > groups_search is not exported from sys.c which causes >> > > > > > problems >> > > > > > for module >> > > > > > builds. in_egroup_p is exported and is just what we need. >> >> > > > > > ----------------------------- >> >> > > > > > Modules: >> > > > > > drm.ko >> > > > > > i915.ko >> >> > > > > > Source: >> > > > > > git://anongit.freedesktop.org/git/mesa/drm modesetting-101 >> > > > > > with drm-patches/* applied. >> >> > > > > > commit e664967bde876b1a1cd205885dafa88a744a6c3d >> > > > > > Author: Dima Zavin <[email protected]> >> > > > > > Date: Thu Jun 26 14:38:42 2008 -0700 >> >> > > > > > intel_fb: Add a hack to allocate the virtual screen 2x the >> > > > > > height >> > > > > > of the actual mode height, to allow a simplisti >> >> > > > > > commit 41bebc2b10235f631f057e4325ea87027a1f2d2b >> > > > > > Author: Dima Zavin <[email protected]> >> > > > > > Date: Thu Jun 26 14:48:58 2008 -0700 >> >> > > > > > intel_fb: pan_display should call mode_set_base, and not >> > > > > > set_config. >> >> > > > > > Signed-off-by: Dima Zavin <[email protected]> >> >> > > > > > commit fe69162447b585918a8a26ab47b44c56a0209a83 >> > > > > > Author: Dima Zavin <[email protected]> >> > > > > > Date: Thu Jun 26 14:48:41 2008 -0700 >> >> > > > > > intelfb: Also fail check_var if the virtual screen dimensions >> > > > > > exceed the fb width/height. >> >> > > > > > Signed-off-by: Dima Zavin <[email protected]> >> >> > > > > > commit e4713780e7907e51bfc2f43667a67eee332b1449 >> > > > > > Author: Dima Zavin <[email protected]> >> > > > > > Date: Thu Jun 26 14:48:18 2008 -0700 >> >> > > > > > intelfb: Force color depth changes to be applied to hw if the >> > > > > > mode >> > > > > > itself has not changed. >> >> > > > > > Signed-off-by: Dima Zavin <[email protected]> >> >> > > > > > commit 9c7137483d1ef47eaff99f98b3c4e552c726e5d9 >> > > > > > Author: Maarten Maathuis <[email protected]> >> > > > > > Date: Thu Aug 14 20:42:28 2008 +0200 >> >> > > > > > nouveau: fix typo >> >> > > > > > On Mar 30, 7:50 am, Jerry Fan <[email protected]> wrote: >> >> > > > > > > HiBrian, >> > > > > > > How did you compile your i915? Did you just copyi915.kofrom >> > > > > > > somewhere >> > > > > > > else, becasue I see some version problem. In addition, I didn't >> > > > > > > see any >> > > > > > > CONFIG_i915 stuff in your .config. So, if you copied i915 from >> > > > > somewhere >> > > > > > > else, u need to make sure module magic number matches the magic >> > > > > > > number >> > > > > of ur >> > > > > > > kernel. >> >> > > > > > > On Mon, Mar 30, 2009 at 10:31 PM,Brian<[email protected]> wrote: >> >> > > > > > > > I made a bit more progress. >> >> > > > > > > > In system/core/mountd/Android.mk, >> >> > > > > > > > #disabled - we are using vold now instead >> > > > > > > > #include $(BUILD_EXECUTABLE) >> >> > > > > > > > So, mountd is not compiled and is not included, but init.rc >> > > > > > > > still >> > > > > uses >> > > > > > > > mountd instead of vold. >> > > > > > > > The code base's changing. >> > > > > > > > I compiled mountd and added it to system/bin, and mountd >> > > > > > > > starts but >> > > > > > > > I'm not sure it's using some configuration file correctly. >> >> > > > > > > > The problem of app_process is still there. It does not start >> > > > > > > > probably >> > > > > > > > because of framebuffer issues. >> > > > > > > > I'm attaching my dmesg. How can I start app_process correctly? >> >> > > > > > > > Thanks, >> > > > > > > >Brian >> >> > > > > > > > <6>[ 3.914300] EXT3 FS on sda8, internal >> > > > > > > > journal >> > > > > > > > <6>[ 3.915900] EXT3-fs: recovery >> > > > > > > > complete. >> > > > > > > > <6>[ 3.917404] EXT3-fs: mounted filesystem with ordered data >> > > > > > > > mode. >> > > > > > > > <6>[ 3.921754] kjournald starting. Commit interval 5 >> > > > > > > > seconds >> > > > > > > > <6>[ 3.923860] EXT3 FS on sda5, internal >> > > > > > > > journal >> > > > > > > > <6>[ 3.930177] EXT3-fs: recovery >> > > > > > > > complete. >> > > > > > > > <6>[ 3.931682] EXT3-fs: mounted filesystem with ordered data >> > > > > > > > mode. >> > > > > > > > <4>[ 3.934621] atl2: disagrees about version of symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.936923] softcursor: disagrees about version of symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.939144] bitblit: disagrees about version of symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.940580] font: disagrees about version of symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.942807] cfbimgblt: disagrees about version of symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.945085] cfbfillrect: disagrees about version of >> > > > > > > > symbol >> > > > > > > > struct_module >> > > > > > > > <4>[ 3.947296] >> >> ... >> >> read more » > > > --~--~---------~--~----~------------~-------~--~----~ unsubscribe: [email protected] website: http://groups.google.com/group/android-porting -~----------~----~----~----~------~----~------~--~---
