1Quick Summary 2------------- 3 4Install ksymoops from 5ftp://ftp.<country>.kernel.org/pub/linux/utils/kernel/ksymoops 6Read the ksymoops man page. 7ksymoops < the_oops.txt 8 9and send the output the maintainer of the kernel area that seems to be 10involved with the problem, not to the ksymoops maintainer. Don't worry 11too much about getting the wrong person. If you are unsure send it to 12the person responsible for the code relevant to what you were doing. 13If it occurs repeatably try and describe how to recreate it. Thats 14worth even more than the oops 15 16If you are totally stumped as to whom to send the report, send it to 17linux-kernel@vger.kernel.org. Thanks for your help in making Linux as 18stable as humanly possible. 19 20Where is the_oops.txt? 21---------------------- 22 23Normally the Oops text is read from the kernel buffers by klogd and 24handed to syslogd which writes it to a syslog file, typically 25/var/log/messages (depends on /etc/syslog.conf). Sometimes klogd dies, 26in which case you can run dmesg > file to read the data from the kernel 27buffers and save it. Or you can cat /proc/kmsg > file, however you 28have to break in to stop the transfer, kmsg is a "never ending file". 29If the machine has crashed so badly that you cannot enter commands or 30the disk is not available then you have three options :- 31 32(1) Hand copy the text from the screen and type it in after the machine 33 has restarted. Messy but it is the only option if you have not 34 planned for a crash. 35 36(2) Boot with a serial console (see Documentation/serial-console.txt), 37 run a null modem to a second machine and capture the output there 38 using your favourite communication program. Minicom works well. 39 40(3) Patch the kernel with one of the crash dump patches. These save 41 data to a floppy disk or video rom or a swap partition. None of 42 these are standard kernel patches so you have to find and apply 43 them yourself. Search kernel archives for kmsgdump, lkcd and 44 oops+smram. 45 46No matter how you capture the log output, feed the resulting file to 47ksymoops along with /proc/ksyms and /proc/modules that applied at the 48time of the crash. /var/log/ksymoops can be useful to capture the 49latter, man ksymoops for details. 50 51 52Full Information 53---------------- 54 55From: Linus Torvalds <torvalds@transmeta.com> 56 57How to track down an Oops.. [originally a mail to linux-kernel] 58 59The main trick is having 5 years of experience with those pesky oops 60messages ;-) 61 62Actually, there are things you can do that make this easier. I have two 63separate approaches: 64 65 gdb /usr/src/linux/vmlinux 66 gdb> disassemble <offending_function> 67 68That's the easy way to find the problem, at least if the bug-report is 69well made (like this one was - run through ksymoops to get the 70information of which function and the offset in the function that it 71happened in). 72 73Oh, it helps if the report happens on a kernel that is compiled with the 74same compiler and similar setups. 75 76The other thing to do is disassemble the "Code:" part of the bug report: 77ksymoops will do this too with the correct tools, but if you don't have 78the tools you can just do a silly program: 79 80 char str[] = "\xXX\xXX\xXX..."; 81 main(){} 82 83and compile it with gcc -g and then do "disassemble str" (where the "XX" 84stuff are the values reported by the Oops - you can just cut-and-paste 85and do a replace of spaces to "\x" - that's what I do, as I'm too lazy 86to write a program to automate this all). 87 88Finally, if you want to see where the code comes from, you can do 89 90 cd /usr/src/linux 91 make fs/buffer.s # or whatever file the bug happened in 92 93and then you get a better idea of what happens than with the gdb 94disassembly. 95 96Now, the trick is just then to combine all the data you have: the C 97sources (and general knowledge of what it _should_ do), the assembly 98listing and the code disassembly (and additionally the register dump you 99also get from the "oops" message - that can be useful to see _what_ the 100corrupted pointers were, and when you have the assembler listing you can 101also match the other registers to whatever C expressions they were used 102for). 103 104Essentially, you just look at what doesn't match (in this case it was the 105"Code" disassembly that didn't match with what the compiler generated). 106Then you need to find out _why_ they don't match. Often it's simple - you 107see that the code uses a NULL pointer and then you look at the code and 108wonder how the NULL pointer got there, and if it's a valid thing to do 109you just check against it.. 110 111Now, if somebody gets the idea that this is time-consuming and requires 112some small amount of concentration, you're right. Which is why I will 113mostly just ignore any panic reports that don't have the symbol table 114info etc looked up: it simply gets too hard to look it up (I have some 115programs to search for specific patterns in the kernel code segment, and 116sometimes I have been able to look up those kinds of panics too, but 117that really requires pretty good knowledge of the kernel just to be able 118to pick out the right sequences etc..) 119 120_Sometimes_ it happens that I just see the disassembled code sequence 121from the panic, and I know immediately where it's coming from. That's when 122I get worried that I've been doing this for too long ;-) 123 124 Linus 125 126 127--------------------------------------------------------------------------- 128Notes on Oops tracing with klogd: 129 130In order to help Linus and the other kernel developers there has been 131substantial support incorporated into klogd for processing protection 132faults. In order to have full support for address resolution at least 133version 1.3-pl3 of the sysklogd package should be used. 134 135When a protection fault occurs the klogd daemon automatically 136translates important addresses in the kernel log messages to their 137symbolic equivalents. This translated kernel message is then 138forwarded through whatever reporting mechanism klogd is using. The 139protection fault message can be simply cut out of the message files 140and forwarded to the kernel developers. 141 142Two types of address resolution are performed by klogd. The first is 143static translation and the second is dynamic translation. Static 144translation uses the System.map file in much the same manner that 145ksymoops does. In order to do static translation the klogd daemon 146must be able to find a system map file at daemon initialization time. 147See the klogd man page for information on how klogd searches for map 148files. 149 150Dynamic address translation is important when kernel loadable modules 151are being used. Since memory for kernel modules is allocated from the 152kernel's dynamic memory pools there are no fixed locations for either 153the start of the module or for functions and symbols in the module. 154 155The kernel supports system calls which allow a program to determine 156which modules are loaded and their location in memory. Using these 157system calls the klogd daemon builds a symbol table which can be used 158to debug a protection fault which occurs in a loadable kernel module. 159 160At the very minimum klogd will provide the name of the module which 161generated the protection fault. There may be additional symbolic 162information available if the developer of the loadable module chose to 163export symbol information from the module. 164 165Since the kernel module environment can be dynamic there must be a 166mechanism for notifying the klogd daemon when a change in module 167environment occurs. There are command line options available which 168allow klogd to signal the currently executing daemon that symbol 169information should be refreshed. See the klogd manual page for more 170information. 171 172A patch is included with the sysklogd distribution which modifies the 173modules-2.0.0 package to automatically signal klogd whenever a module 174is loaded or unloaded. Applying this patch provides essentially 175seamless support for debugging protection faults which occur with 176kernel loadable modules. 177 178The following is an example of a protection fault in a loadable module 179processed by klogd: 180--------------------------------------------------------------------------- 181Aug 29 09:51:01 blizard kernel: Unable to handle kernel paging request at virtual address f15e97cc 182Aug 29 09:51:01 blizard kernel: current->tss.cr3 = 0062d000, %cr3 = 0062d000 183Aug 29 09:51:01 blizard kernel: *pde = 00000000 184Aug 29 09:51:01 blizard kernel: Oops: 0002 185Aug 29 09:51:01 blizard kernel: CPU: 0 186Aug 29 09:51:01 blizard kernel: EIP: 0010:[oops:_oops+16/3868] 187Aug 29 09:51:01 blizard kernel: EFLAGS: 00010212 188Aug 29 09:51:01 blizard kernel: eax: 315e97cc ebx: 003a6f80 ecx: 001be77b edx: 00237c0c 189Aug 29 09:51:01 blizard kernel: esi: 00000000 edi: bffffdb3 ebp: 00589f90 esp: 00589f8c 190Aug 29 09:51:01 blizard kernel: ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018 191Aug 29 09:51:01 blizard kernel: Process oops_test (pid: 3374, process nr: 21, stackpage=00589000) 192Aug 29 09:51:01 blizard kernel: Stack: 315e97cc 00589f98 0100b0b4 bffffed4 0012e38e 00240c64 003a6f80 00000001 193Aug 29 09:51:01 blizard kernel: 00000000 00237810 bfffff00 0010a7fa 00000003 00000001 00000000 bfffff00 194Aug 29 09:51:01 blizard kernel: bffffdb3 bffffed4 ffffffda 0000002b 0007002b 0000002b 0000002b 00000036 195Aug 29 09:51:01 blizard kernel: Call Trace: [oops:_oops_ioctl+48/80] [_sys_ioctl+254/272] [_system_call+82/128] 196Aug 29 09:51:01 blizard kernel: Code: c7 00 05 00 00 00 eb 08 90 90 90 90 90 90 90 90 89 ec 5d c3 197--------------------------------------------------------------------------- 198 199Dr. G.W. Wettstein Oncology Research Div. Computing Facility 200Roger Maris Cancer Center INTERNET: greg@wind.rmcc.com 201820 4th St. N. 202Fargo, ND 58122 203Phone: 701-234-7556 204 205 206--------------------------------------------------------------------------- 207Tainted kernels: 208 209Some oops reports contain the string 'Tainted: ' after the program 210counter, this indicates that the kernel has been tainted by some 211mechanism. The string is followed by a series of position sensitive 212characters, each representing a particular tainted value. 213 214 1: 'G' if all modules loaded have a GPL or compatible license, 'P' if 215 any proprietary module has been loaded. Modules without a 216 MODULE_LICENSE or with a MODULE_LICENSE that is not recognised by 217 insmod as GPL compatible are assumed to be proprietary. 218 219 2: 'F' if any module was force loaded by insmod -f, ' ' if all 220 modules were loaded normally. 221 222The primary reason for the 'Tainted: ' string is to tell kernel 223debuggers if this is a clean kernel or if anything unusual has 224occurred. Tainting is permanent, even if an offending module is 225unloading the tainted value remains to indicate that the kernel is not 226trustworthy. 227