From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <3AB7E349.8AC5A1B0@enst.fr> Date: Wed, 21 Mar 2001 00:10:01 +0100 From: Stefan Nunninger Reply-To: nunninger@web.de MIME-Version: 1.0 To: linuxppc-embedded@lists.linuxppc.org Subject: strange problem while booting Content-Type: text/plain; charset=iso-8859-1 Sender: owner-linuxppc-embedded@lists.linuxppc.org List-Id: Hello everybody, Once more I've got a problem while trying to boot a montavista linux 2.2.14 on a custom board with a MPC860. I'm using BDM4GDB for debugging and running linux. Usually when booting linux it boots until it tries to load the ramdisk. There it stops for a reason I did not manage to figure out yet. The output of the kernel looks like this: -------------------------------------------------------- decomp : ZIMAGE_OFFSET: 0001C1F0 INITRD_OFFSET: 00077DF3 INITRD_SIZE: 0014D729 loaded at: 00100000 00100000 board data at: 001001E4 00100210 relocated to: 000F0100 000F012C ZIMAGE_OFFSET: 0001C1F0 zimage_start : 0010C1F0 zimage_size : 00032BA2 initrd_start : 00167DF3 initrd_end : 002B551C zimage at: 0010C1F0 0013ED92 initrd at: 00167DF3 002B551C avail ram: 002B6000 01000000 Linux/PPC load: Uncompressing Linux... Gunzip ... Dest address : 00000000 Dest length : 00100000 Source address : 0010C1F0 Now booting the kernel Linux version 2.2.14 (stefan@paris.enst.fr) (gcc version 2.95.2 19991030 (2.95.3 prerelease/franzo)) #617 Mon Mar 19 18:15:35 CET 2001 Boot arguments: root=/dev/ram time_init: decrementer frequency = 120000000/60 parse_options finished Calibrating delay loop... 31.85 BogoMIPS Memory: 14232k available (412k kernel code, 376k data, 24k init) [c0000000,c1000000] Dentry hash table entries: 2048 (order 2, 16k) Buffer cache hash table entries: 16384 (order 4, 64k) Page cache hash table entries: 4096 (order 2, 16k) POSIX conformance testing by UNIFIX do_basic_setup Linux NET4.0 for Linux 2.2 Based upon Swansea University Computer Society NET3.039 Starting kswapd v 1.5 CPM UART driver version 0.03 ttyS00 at 0x0280 is a SMC ttyS01 at 0x0380 is a SMC ttyS02 at 0x0100 is a SCC ttyS03 at 0x0200 is a SCC pty: 256 Unix98 ptys configured RAM disk driver initialized: 16 RAM disks of 4096K size loop: registered device at major 7 RAMDISK: Couldn't find valid RAM disk image starting at 0. function: /fs/super.c::mount_root says Kernel panic: VFS: Unable to mount root fs on 01:00 Rebooting in 180 seconds.. While trying to debug this it appears that my board behaves in a strange way I can not really understand. After resetting the board (usually I turn of power for a second) and rebooting the kernel, there is not the same kernel output instead I see one of the following outputs: -------------------------------------------------------- decomp : ZIMAGE_OFFSET: 0001C1F0 INITRD_OFFSET: 00077DE3 INITRD_SIZE: 0014D729 loaded at: 00100000 0010002C board data at: 001001E4 00100210 relocated to: 000F0100 000F012C ZIMAGE_OFFSET: 0001C1F0 zimage_start : 0010C1F0 zimage_size : 00032B94 initrd_start : 00167DE3 initrd_end : 002B550C zimage at: 0010C1F0 0013ED84 initrd at: 00167DE3 002B550C avail ram: 002B6000 01000000 Linux/PPC load: Uncompressing Linux... Gunzip ... Dest address : 00000000 Dest length : 00100000 Source address : 0010C1F0 Now booting the kernel - and nothing more. When debugging this I found that the kernel switches on virtual adresses and goes on until identify_machine. From there it branches to a c-function. However this is not identify_machine but any other function. As this makes no sense the kernel stops somewhere after. -------------------------------------------------------- decomp : ZIMAGE_OFFSET: 0001C1F0 INITRD_OFFSET: 00077DF7 INITRD_SIZE: 0014D729 loaded at: 00100000 00110004 board data at: 001001E4 00100210 relocated to: 000F0100 000F012C ZIMAGE_OFFSET: 0001C1F0 zimage_start : 0010C1F0 zimage_size : 00032BA5 initrd_start : 00167DF7 initrd_end : 002B5520 zimage at: 0010C1F0 0013ED95 initrd at: 00167DF7 002B5520 avail ram: 002B6000 01000000 Linux/PPC load: Uncompressing Linux... Gunzip ... Dest address : 00000000 Dest length : 00100000 Source address : 0010C1F0 Now booting the kernel Linux version 2.2.14 (stefan@paris.enst.fr) (gcc version 2.95.2 19991030 (21Boot arguments: root=/dev/ram time_init: decrementer frequency = 120000000/60 parse_options finished Calibrating delay loop... 31.74 BogoMIPS Memory: 14232k available (412k kernel code, 376k data, 24k init) [c0000000,]Dentry hash table entries: 2048 (order 2, 16k) Buffer cache hash table entries: 16384 (order 4, 64k) Page cache hash table entries: 4096 (order 2, 16k) POSIX conformance testing by UNIFIX do_basic_setup Linux NET4.0 for Linux 2.2 Based upon Swansea University Computer Society NET3.039 Starting kswapd v 1.5 CPM UART driver version 0.03 ttyS00 at 0x0280 is a SMC ttyS01 at 0x0380 is a SMC ttyS02 at 0x0100 is a SCC ttyS03 at 0x0200 is a SCC pty: 256 Unix98 ptys configured RAM disk driver initialized: 16 RAM disks of 4096K size loop: registered device at major 7 Machine check in kernel mode. Caused by (from msr): regs c0165ca0 Unknown values in msr NIP: 00000200 XER: 8000FF0A LR: C0019394 REGS: c0165ca0 TRAP: 0200 MSR: 00009002 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 00 TASK = c0164000[1] 'swapper' mm->pgd c0070000 Last syscall: 120 last math 00000000 GPR00: C0019394 C0165D50 C0164000 0006F885 C0165D68 C015D000 00000000 00000 GPR08: C0076028 C015D000 00000000 000003F2 35000004 1111000D 1111000E 11110 GPR16: 11110010 11110011 11110012 11110013 00009032 00165E10 00000000 C0002 GPR24: C0009814 11110019 00000000 C015D000 C0164000 C00715E0 0006F885 C0071 Call backtrace: 00000023 30303020 this time I get a machine check exception. The backtrace does not really help as there is nothing at this addresses. yet another output looks like this: -------------------------------------------------------- decomp : ZIMAGE_OFFSET: 0001C1F0 °²@..+#± _+#_+ : 0010C1F0 >+#± _> : 00032BA6 Å+_ _+#_+ : 00167DF7 Å+_ Å : 002B5520 >+#± #+: 0010C1F0 0013ED96 Å+_ #+: 00167DF7 002B5520 #+#+ _#+: 002B6000 01000000 LÅ+|/PPC +Ä#: UÅÄ+Ä___ű LÅ+|... G+Å>Ä ... D_+ #___ : 00000000 D_+ +ű+ : 00100000 SÄ+_ #___ : 0010C1F0 NÄ+ ÄÄ+ű + +_Å+ LÅ+| +__ÄÅ 2.2.14 (_+°#Å@Ä#__.Å_+.°_) (± +__ÄÅ 2.95.2 19991030 (2.95.3 Ä__+1BÄÄ+ #_±++Å+_: _ÄÄ+=/+/_#+ ++ Å+: _+Å+_ °_Ä+Å< = 120000000/60 Ä#__ ÄÄ+ÄÅ_ °Å_ C#+_#+ű +#< +ÄÄÄ... 31.74 BıÄMIPS M+Ä_<: 14232+ #+#+#+ (412+ +_Å+ Ä, 376+ #+#, 24+ Å+) [0000000,1000000] DÅ+_< #_ +#+ Å+__: 2048 (Ä__ 2, 16+) B+°°_ # + Å++ Å+__: 16384 (Ä__ 4, 64+) B+°°_ # #_ +#+ Å+__: 16384 (Ä__ 4, 64+) P#± # #_ +#+ Å+__: 4096 (Ä__ 2, 16+) POSIX ÄÅ°Ä_+#Å +_+ű < UNIFIX Ä #_ _++Ä LÅ+| NET4.0 °Ä_ LÅ+| 2.2 B#_ +ÄÄÅ S+#Å_# UÅ+__+< CÄ+Ä++_ SÄ+< NET3.039 S+#_+ű +_+#Ä + 1.5 CPM UART _+_ +__ÄÅ 0.03 ++: 16 RAM _+_ Ä° 4096K _> +ÄÄÄ: _±_+_ + #+ +#+Ä_ 7 It appears that all upper case letters and numbers are correct. All other characters are changed. These changes are not random. When running the kernel several time the output is exactly the same. By chance I connected this board to another debugger (VisionClick) to see what happens there. Interestingly everything worked fine. It gets until trying to boot the ramdisk. Thus I conclude there is something wrong with my initialisation code in the bootloader. Some values which are set by VisionClick might not be set properly by my bootcode. Thus I tried to find any differences. All I could find was this REG VisionClick BDM4GDB SIPEND 00000000 02A80000 MCR 0000003F 4080003F PLPRCR 00700000 00704000 RSR 00000000 C0000000 PBDAT 00001EC4 00001DC4 I used the values of VisionClick to set up these register in BDM4GDB. It did not help in any way. Next I've copied all register values from VisionClick and set the values excactly like this in BDB4GDB. But still no improvement. The values which I normally initialize in my bootloader are as follows: DER = 0x2002000F DEC = 0x7FFFFFFF SYPCR = 0xffffff83 ICTRL = 0x00000007 ORx, BRx UPMAx MSTAT = 0x00000200 MAMR = 0x07821330 ICR = 0x0 MSR = 0x1042 IMMR = 0xFA200000 SIUMCR = 0x00292900 TBL = 0x0 TBU = 0x0 plprcr = 0x00700000 Is there something missing here? Could anybody give me a hint what might be the reason for all this strange behaviour. What could solve the problem? I hope somebody has an idea as I spend now nearly two weeks on that without any result. Many thanks Stefan ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/