ia64/linux-2.6.18-xen.hg

annotate Documentation/oops-tracing.txt @ 897:329ea0ccb344

balloon: try harder to balloon up under memory pressure.

Currently if the balloon driver is unable to increase the guest's
reservation it assumes the failure was due to reaching its full
allocation, gives up on the ballooning operation and records the limit
it reached as the "hard limit". The driver will not try again until
the target is set again (even to the same value).

However it is possible that ballooning has in fact failed due to
memory pressure in the host and therefore it is desirable to keep
attempting to reach the target in case memory becomes available. The
most likely scenario is that some guests are ballooning down while
others are ballooning up and therefore there is temporary memory
pressure while things stabilise. You would not expect a well behaved
toolstack to ask a domain to balloon to more than its allocation nor
would you expect it to deliberately over-commit memory by setting
balloon targets which exceed the total host memory.

This patch drops the concept of a hard limit and causes the balloon
driver to retry increasing the reservation on a timer in the same
manner as when decreasing the reservation.

Also if we partially succeed in increasing the reservation
(i.e. receive less pages than we asked for) then we may as well keep
those pages rather than returning them to Xen.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
author Keir Fraser <keir.fraser@citrix.com>
date Fri Jun 05 14:01:20 2009 +0100 (2009-06-05)
parents 831230e53067
children
rev   line source
ian@0 1 NOTE: ksymoops is useless on 2.6. Please use the Oops in its original format
ian@0 2 (from dmesg, etc). Ignore any references in this or other docs to "decoding
ian@0 3 the Oops" or "running it through ksymoops". If you post an Oops from 2.6 that
ian@0 4 has been run through ksymoops, people will just tell you to repost it.
ian@0 5
ian@0 6 Quick Summary
ian@0 7 -------------
ian@0 8
ian@0 9 Find the Oops and send it to the maintainer of the kernel area that seems to be
ian@0 10 involved with the problem. Don't worry too much about getting the wrong person.
ian@0 11 If you are unsure send it to the person responsible for the code relevant to
ian@0 12 what you were doing. If it occurs repeatably try and describe how to recreate
ian@0 13 it. That's worth even more than the oops.
ian@0 14
ian@0 15 If you are totally stumped as to whom to send the report, send it to
ian@0 16 linux-kernel@vger.kernel.org. Thanks for your help in making Linux as
ian@0 17 stable as humanly possible.
ian@0 18
ian@0 19 Where is the Oops?
ian@0 20 ----------------------
ian@0 21
ian@0 22 Normally the Oops text is read from the kernel buffers by klogd and
ian@0 23 handed to syslogd which writes it to a syslog file, typically
ian@0 24 /var/log/messages (depends on /etc/syslog.conf). Sometimes klogd dies,
ian@0 25 in which case you can run dmesg > file to read the data from the kernel
ian@0 26 buffers and save it. Or you can cat /proc/kmsg > file, however you
ian@0 27 have to break in to stop the transfer, kmsg is a "never ending file".
ian@0 28 If the machine has crashed so badly that you cannot enter commands or
ian@0 29 the disk is not available then you have three options :-
ian@0 30
ian@0 31 (1) Hand copy the text from the screen and type it in after the machine
ian@0 32 has restarted. Messy but it is the only option if you have not
ian@0 33 planned for a crash. Alternatively, you can take a picture of
ian@0 34 the screen with a digital camera - not nice, but better than
ian@0 35 nothing. If the messages scroll off the top of the console, you
ian@0 36 may find that booting with a higher resolution (eg, vga=791)
ian@0 37 will allow you to read more of the text. (Caveat: This needs vesafb,
ian@0 38 so won't help for 'early' oopses)
ian@0 39
ian@0 40 (2) Boot with a serial console (see Documentation/serial-console.txt),
ian@0 41 run a null modem to a second machine and capture the output there
ian@0 42 using your favourite communication program. Minicom works well.
ian@0 43
ian@0 44 (3) Use Kdump (see Documentation/kdump/kdump.txt),
ian@0 45 extract the kernel ring buffer from old memory with using dmesg
ian@0 46 gdbmacro in Documentation/kdump/gdbmacros.txt.
ian@0 47
ian@0 48
ian@0 49 Full Information
ian@0 50 ----------------
ian@0 51
ian@0 52 NOTE: the message from Linus below applies to 2.4 kernel. I have preserved it
ian@0 53 for historical reasons, and because some of the information in it still
ian@0 54 applies. Especially, please ignore any references to ksymoops.
ian@0 55
ian@0 56 From: Linus Torvalds <torvalds@osdl.org>
ian@0 57
ian@0 58 How to track down an Oops.. [originally a mail to linux-kernel]
ian@0 59
ian@0 60 The main trick is having 5 years of experience with those pesky oops
ian@0 61 messages ;-)
ian@0 62
ian@0 63 Actually, there are things you can do that make this easier. I have two
ian@0 64 separate approaches:
ian@0 65
ian@0 66 gdb /usr/src/linux/vmlinux
ian@0 67 gdb> disassemble <offending_function>
ian@0 68
ian@0 69 That's the easy way to find the problem, at least if the bug-report is
ian@0 70 well made (like this one was - run through ksymoops to get the
ian@0 71 information of which function and the offset in the function that it
ian@0 72 happened in).
ian@0 73
ian@0 74 Oh, it helps if the report happens on a kernel that is compiled with the
ian@0 75 same compiler and similar setups.
ian@0 76
ian@0 77 The other thing to do is disassemble the "Code:" part of the bug report:
ian@0 78 ksymoops will do this too with the correct tools, but if you don't have
ian@0 79 the tools you can just do a silly program:
ian@0 80
ian@0 81 char str[] = "\xXX\xXX\xXX...";
ian@0 82 main(){}
ian@0 83
ian@0 84 and compile it with gcc -g and then do "disassemble str" (where the "XX"
ian@0 85 stuff are the values reported by the Oops - you can just cut-and-paste
ian@0 86 and do a replace of spaces to "\x" - that's what I do, as I'm too lazy
ian@0 87 to write a program to automate this all).
ian@0 88
ian@0 89 Finally, if you want to see where the code comes from, you can do
ian@0 90
ian@0 91 cd /usr/src/linux
ian@0 92 make fs/buffer.s # or whatever file the bug happened in
ian@0 93
ian@0 94 and then you get a better idea of what happens than with the gdb
ian@0 95 disassembly.
ian@0 96
ian@0 97 Now, the trick is just then to combine all the data you have: the C
ian@0 98 sources (and general knowledge of what it _should_ do), the assembly
ian@0 99 listing and the code disassembly (and additionally the register dump you
ian@0 100 also get from the "oops" message - that can be useful to see _what_ the
ian@0 101 corrupted pointers were, and when you have the assembler listing you can
ian@0 102 also match the other registers to whatever C expressions they were used
ian@0 103 for).
ian@0 104
ian@0 105 Essentially, you just look at what doesn't match (in this case it was the
ian@0 106 "Code" disassembly that didn't match with what the compiler generated).
ian@0 107 Then you need to find out _why_ they don't match. Often it's simple - you
ian@0 108 see that the code uses a NULL pointer and then you look at the code and
ian@0 109 wonder how the NULL pointer got there, and if it's a valid thing to do
ian@0 110 you just check against it..
ian@0 111
ian@0 112 Now, if somebody gets the idea that this is time-consuming and requires
ian@0 113 some small amount of concentration, you're right. Which is why I will
ian@0 114 mostly just ignore any panic reports that don't have the symbol table
ian@0 115 info etc looked up: it simply gets too hard to look it up (I have some
ian@0 116 programs to search for specific patterns in the kernel code segment, and
ian@0 117 sometimes I have been able to look up those kinds of panics too, but
ian@0 118 that really requires pretty good knowledge of the kernel just to be able
ian@0 119 to pick out the right sequences etc..)
ian@0 120
ian@0 121 _Sometimes_ it happens that I just see the disassembled code sequence
ian@0 122 from the panic, and I know immediately where it's coming from. That's when
ian@0 123 I get worried that I've been doing this for too long ;-)
ian@0 124
ian@0 125 Linus
ian@0 126
ian@0 127
ian@0 128 ---------------------------------------------------------------------------
ian@0 129 Notes on Oops tracing with klogd:
ian@0 130
ian@0 131 In order to help Linus and the other kernel developers there has been
ian@0 132 substantial support incorporated into klogd for processing protection
ian@0 133 faults. In order to have full support for address resolution at least
ian@0 134 version 1.3-pl3 of the sysklogd package should be used.
ian@0 135
ian@0 136 When a protection fault occurs the klogd daemon automatically
ian@0 137 translates important addresses in the kernel log messages to their
ian@0 138 symbolic equivalents. This translated kernel message is then
ian@0 139 forwarded through whatever reporting mechanism klogd is using. The
ian@0 140 protection fault message can be simply cut out of the message files
ian@0 141 and forwarded to the kernel developers.
ian@0 142
ian@0 143 Two types of address resolution are performed by klogd. The first is
ian@0 144 static translation and the second is dynamic translation. Static
ian@0 145 translation uses the System.map file in much the same manner that
ian@0 146 ksymoops does. In order to do static translation the klogd daemon
ian@0 147 must be able to find a system map file at daemon initialization time.
ian@0 148 See the klogd man page for information on how klogd searches for map
ian@0 149 files.
ian@0 150
ian@0 151 Dynamic address translation is important when kernel loadable modules
ian@0 152 are being used. Since memory for kernel modules is allocated from the
ian@0 153 kernel's dynamic memory pools there are no fixed locations for either
ian@0 154 the start of the module or for functions and symbols in the module.
ian@0 155
ian@0 156 The kernel supports system calls which allow a program to determine
ian@0 157 which modules are loaded and their location in memory. Using these
ian@0 158 system calls the klogd daemon builds a symbol table which can be used
ian@0 159 to debug a protection fault which occurs in a loadable kernel module.
ian@0 160
ian@0 161 At the very minimum klogd will provide the name of the module which
ian@0 162 generated the protection fault. There may be additional symbolic
ian@0 163 information available if the developer of the loadable module chose to
ian@0 164 export symbol information from the module.
ian@0 165
ian@0 166 Since the kernel module environment can be dynamic there must be a
ian@0 167 mechanism for notifying the klogd daemon when a change in module
ian@0 168 environment occurs. There are command line options available which
ian@0 169 allow klogd to signal the currently executing daemon that symbol
ian@0 170 information should be refreshed. See the klogd manual page for more
ian@0 171 information.
ian@0 172
ian@0 173 A patch is included with the sysklogd distribution which modifies the
ian@0 174 modules-2.0.0 package to automatically signal klogd whenever a module
ian@0 175 is loaded or unloaded. Applying this patch provides essentially
ian@0 176 seamless support for debugging protection faults which occur with
ian@0 177 kernel loadable modules.
ian@0 178
ian@0 179 The following is an example of a protection fault in a loadable module
ian@0 180 processed by klogd:
ian@0 181 ---------------------------------------------------------------------------
ian@0 182 Aug 29 09:51:01 blizard kernel: Unable to handle kernel paging request at virtual address f15e97cc
ian@0 183 Aug 29 09:51:01 blizard kernel: current->tss.cr3 = 0062d000, %cr3 = 0062d000
ian@0 184 Aug 29 09:51:01 blizard kernel: *pde = 00000000
ian@0 185 Aug 29 09:51:01 blizard kernel: Oops: 0002
ian@0 186 Aug 29 09:51:01 blizard kernel: CPU: 0
ian@0 187 Aug 29 09:51:01 blizard kernel: EIP: 0010:[oops:_oops+16/3868]
ian@0 188 Aug 29 09:51:01 blizard kernel: EFLAGS: 00010212
ian@0 189 Aug 29 09:51:01 blizard kernel: eax: 315e97cc ebx: 003a6f80 ecx: 001be77b edx: 00237c0c
ian@0 190 Aug 29 09:51:01 blizard kernel: esi: 00000000 edi: bffffdb3 ebp: 00589f90 esp: 00589f8c
ian@0 191 Aug 29 09:51:01 blizard kernel: ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018
ian@0 192 Aug 29 09:51:01 blizard kernel: Process oops_test (pid: 3374, process nr: 21, stackpage=00589000)
ian@0 193 Aug 29 09:51:01 blizard kernel: Stack: 315e97cc 00589f98 0100b0b4 bffffed4 0012e38e 00240c64 003a6f80 00000001
ian@0 194 Aug 29 09:51:01 blizard kernel: 00000000 00237810 bfffff00 0010a7fa 00000003 00000001 00000000 bfffff00
ian@0 195 Aug 29 09:51:01 blizard kernel: bffffdb3 bffffed4 ffffffda 0000002b 0007002b 0000002b 0000002b 00000036
ian@0 196 Aug 29 09:51:01 blizard kernel: Call Trace: [oops:_oops_ioctl+48/80] [_sys_ioctl+254/272] [_system_call+82/128]
ian@0 197 Aug 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
ian@0 198 ---------------------------------------------------------------------------
ian@0 199
ian@0 200 Dr. G.W. Wettstein Oncology Research Div. Computing Facility
ian@0 201 Roger Maris Cancer Center INTERNET: greg@wind.rmcc.com
ian@0 202 820 4th St. N.
ian@0 203 Fargo, ND 58122
ian@0 204 Phone: 701-234-7556
ian@0 205
ian@0 206
ian@0 207 ---------------------------------------------------------------------------
ian@0 208 Tainted kernels:
ian@0 209
ian@0 210 Some oops reports contain the string 'Tainted: ' after the program
ian@0 211 counter. This indicates that the kernel has been tainted by some
ian@0 212 mechanism. The string is followed by a series of position-sensitive
ian@0 213 characters, each representing a particular tainted value.
ian@0 214
ian@0 215 1: 'G' if all modules loaded have a GPL or compatible license, 'P' if
ian@0 216 any proprietary module has been loaded. Modules without a
ian@0 217 MODULE_LICENSE or with a MODULE_LICENSE that is not recognised by
ian@0 218 insmod as GPL compatible are assumed to be proprietary.
ian@0 219
ian@0 220 2: 'F' if any module was force loaded by "insmod -f", ' ' if all
ian@0 221 modules were loaded normally.
ian@0 222
ian@0 223 3: 'S' if the oops occurred on an SMP kernel running on hardware that
ian@0 224 hasn't been certified as safe to run multiprocessor.
ian@0 225 Currently this occurs only on various Athlons that are not
ian@0 226 SMP capable.
ian@0 227
ian@0 228 4: 'R' if a module was force unloaded by "rmmod -f", ' ' if all
ian@0 229 modules were unloaded normally.
ian@0 230
ian@0 231 5: 'M' if any processor has reported a Machine Check Exception,
ian@0 232 ' ' if no Machine Check Exceptions have occurred.
ian@0 233
ian@0 234 6: 'B' if a page-release function has found a bad page reference or
ian@0 235 some unexpected page flags.
ian@0 236
ian@0 237 The primary reason for the 'Tainted: ' string is to tell kernel
ian@0 238 debuggers if this is a clean kernel or if anything unusual has
ian@0 239 occurred. Tainting is permanent: even if an offending module is
ian@0 240 unloaded, the tainted value remains to indicate that the kernel is not
ian@0 241 trustworthy.