| blob | bdd344aa18d91d94db4ed3537320dfd42bfe7523 |
2 Firmware-Assisted Dump
3 ------------------------
4 July 2011
6 The goal of firmware-assisted dump is to enable the dump of
7 a crashed system, and to do so from a fully-reset system, and
8 to minimize the total elapsed time until the system is back
9 in production use.
11 - Firmware assisted dump (fadump) infrastructure is intended to replace
12 the existing phyp assisted dump.
13 - Fadump uses the same firmware interfaces and memory reservation model
14 as phyp assisted dump.
15 - Unlike phyp dump, fadump exports the memory dump through /proc/vmcore
16 in the ELF format in the same way as kdump. This helps us reuse the
17 kdump infrastructure for dump capture and filtering.
18 - Unlike phyp dump, userspace tool does not need to refer any sysfs
19 interface while reading /proc/vmcore.
20 - Unlike phyp dump, fadump allows user to release all the memory reserved
21 for dump, with a single operation of echo 1 > /sys/kernel/fadump_release_mem.
22 - Once enabled through kernel boot parameter, fadump can be
23 started/stopped through /sys/kernel/fadump_registered interface (see
24 sysfs files section below) and can be easily integrated with kdump
25 service start/stop init scripts.
27 Comparing with kdump or other strategies, firmware-assisted
28 dump offers several strong, practical advantages:
30 -- Unlike kdump, the system has been reset, and loaded
31 with a fresh copy of the kernel. In particular,
32 PCI and I/O devices have been reinitialized and are
33 in a clean, consistent state.
34 -- Once the dump is copied out, the memory that held the dump
35 is immediately available to the running kernel. And therefore,
36 unlike kdump, fadump doesn't need a 2nd reboot to get back
37 the system to the production configuration.
39 The above can only be accomplished by coordination with,
40 and assistance from the Power firmware. The procedure is
41 as follows:
43 -- The first kernel registers the sections of memory with the
44 Power firmware for dump preservation during OS initialization.
45 These registered sections of memory are reserved by the first
46 kernel during early boot.
48 -- When a system crashes, the Power firmware will save
49 the low memory (boot memory of size larger of 5% of system RAM
50 or 256MB) of RAM to the previous registered region. It will
51 also save system registers, and hardware PTE's.
53 NOTE: The term 'boot memory' means size of the low memory chunk
54 that is required for a kernel to boot successfully when
55 booted with restricted memory. By default, the boot memory
56 size will be the larger of 5% of system RAM or 256MB.
57 Alternatively, user can also specify boot memory size
58 through boot parameter 'crashkernel=' which will override
59 the default calculated size. Use this option if default
60 boot memory size is not sufficient for second kernel to
61 boot successfully. For syntax of crashkernel= parameter,
62 refer to Documentation/kdump/kdump.txt. If any offset is
63 provided in crashkernel= parameter, it will be ignored
64 as fadump uses a predefined offset to reserve memory
65 for boot memory dump preservation in case of a crash.
67 -- After the low memory (boot memory) area has been saved, the
68 firmware will reset PCI and other hardware state. It will
69 *not* clear the RAM. It will then launch the bootloader, as
70 normal.
72 -- The freshly booted kernel will notice that there is a new
73 node (ibm,dump-kernel) in the device tree, indicating that
74 there is crash data available from a previous boot. During
75 the early boot OS will reserve rest of the memory above
76 boot memory size effectively booting with restricted memory
77 size. This will make sure that the second kernel will not
78 touch any of the dump memory area.
80 -- User-space tools will read /proc/vmcore to obtain the contents
81 of memory, which holds the previous crashed kernel dump in ELF
82 format. The userspace tools may copy this info to disk, or
83 network, nas, san, iscsi, etc. as desired.
85 -- Once the userspace tool is done saving dump, it will echo
86 '1' to /sys/kernel/fadump_release_mem to release the reserved
87 memory back to general use, except the memory required for
88 next firmware-assisted dump registration.
90 e.g.
91 # echo 1 > /sys/kernel/fadump_release_mem
93 Please note that the firmware-assisted dump feature
94 is only available on Power6 and above systems with recent
95 firmware versions.
97 Implementation details:
98 ----------------------
100 During boot, a check is made to see if firmware supports
101 this feature on that particular machine. If it does, then
102 we check to see if an active dump is waiting for us. If yes
103 then everything but boot memory size of RAM is reserved during
104 early boot (See Fig. 2). This area is released once we finish
105 collecting the dump from user land scripts (e.g. kdump scripts)
106 that are run. If there is dump data, then the
107 /sys/kernel/fadump_release_mem file is created, and the reserved
108 memory is held.
110 If there is no waiting dump data, then only the memory required
111 to hold CPU state, HPTE region, boot memory dump and elfcore
112 header, is usually reserved at an offset greater than boot memory
113 size (see Fig. 1). This area is *not* released: this region will
114 be kept permanently reserved, so that it can act as a receptacle
115 for a copy of the boot memory content in addition to CPU state
116 and HPTE region, in the case a crash does occur.
118 o Memory Reservation during first kernel
120 Low memory Top of memory
121 0 boot memory size |
122 | | |<--Reserved dump area -->| |
123 V V | Permanent Reservation | V
124 +-----------+----------/ /---+---+----+-----------+----+------+
125 | | |CPU|HPTE| DUMP |ELF | |
126 +-----------+----------/ /---+---+----+-----------+----+------+
127 | ^
128 | |
129 \ /
130 -------------------------------------------
131 Boot memory content gets transferred to
132 reserved area by firmware at the time of
133 crash
134 Fig. 1
136 o Memory Reservation during second kernel after crash
138 Low memory Top of memory
139 0 boot memory size |
140 | |<------------- Reserved dump area ----------- -->|
141 V V V
142 +-----------+----------/ /---+---+----+-----------+----+------+
143 | | |CPU|HPTE| DUMP |ELF | |
144 +-----------+----------/ /---+---+----+-----------+----+------+
145 | |
146 V V
147 Used by second /proc/vmcore
148 kernel to boot
149 Fig. 2
151 Currently the dump will be copied from /proc/vmcore to a
152 a new file upon user intervention. The dump data available through
153 /proc/vmcore will be in ELF format. Hence the existing kdump
154 infrastructure (kdump scripts) to save the dump works fine with
155 minor modifications.
157 The tools to examine the dump will be same as the ones
158 used for kdump.
160 How to enable firmware-assisted dump (fadump):
161 -------------------------------------
163 1. Set config option CONFIG_FA_DUMP=y and build kernel.
164 2. Boot into linux kernel with 'fadump=on' kernel cmdline option.
165 3. Optionally, user can also set 'crashkernel=' kernel cmdline
166 to specify size of the memory to reserve for boot memory dump
167 preservation.
169 NOTE: 1. 'fadump_reserve_mem=' parameter has been deprecated. Instead
170 use 'crashkernel=' to specify size of the memory to reserve
171 for boot memory dump preservation.
172 2. If firmware-assisted dump fails to reserve memory then it
173 will fallback to existing kdump mechanism if 'crashkernel='
174 option is set at kernel cmdline.
176 Sysfs/debugfs files:
177 ------------
179 Firmware-assisted dump feature uses sysfs file system to hold
180 the control files and debugfs file to display memory reserved region.
182 Here is the list of files under kernel sysfs:
184 /sys/kernel/fadump_enabled
186 This is used to display the fadump status.
187 0 = fadump is disabled
188 1 = fadump is enabled
190 This interface can be used by kdump init scripts to identify if
191 fadump is enabled in the kernel and act accordingly.
193 /sys/kernel/fadump_registered
195 This is used to display the fadump registration status as well
196 as to control (start/stop) the fadump registration.
197 0 = fadump is not registered.
198 1 = fadump is registered and ready to handle system crash.
200 To register fadump echo 1 > /sys/kernel/fadump_registered and
201 echo 0 > /sys/kernel/fadump_registered for un-register and stop the
202 fadump. Once the fadump is un-registered, the system crash will not
203 be handled and vmcore will not be captured. This interface can be
204 easily integrated with kdump service start/stop.
206 /sys/kernel/fadump_release_mem
208 This file is available only when fadump is active during
209 second kernel. This is used to release the reserved memory
210 region that are held for saving crash dump. To release the
211 reserved memory echo 1 to it:
213 echo 1 > /sys/kernel/fadump_release_mem
215 After echo 1, the content of the /sys/kernel/debug/powerpc/fadump_region
216 file will change to reflect the new memory reservations.
218 The existing userspace tools (kdump infrastructure) can be easily
219 enhanced to use this interface to release the memory reserved for
220 dump and continue without 2nd reboot.
222 Here is the list of files under powerpc debugfs:
223 (Assuming debugfs is mounted on /sys/kernel/debug directory.)
225 /sys/kernel/debug/powerpc/fadump_region
227 This file shows the reserved memory regions if fadump is
228 enabled otherwise this file is empty. The output format
229 is:
230 <region>: [<start>-<end>] <reserved-size> bytes, Dumped: <dump-size>
232 e.g.
233 Contents when fadump is registered during first kernel
235 # cat /sys/kernel/debug/powerpc/fadump_region
236 CPU : [0x0000006ffb0000-0x0000006fff001f] 0x40020 bytes, Dumped: 0x0
237 HPTE: [0x0000006fff0020-0x0000006fff101f] 0x1000 bytes, Dumped: 0x0
238 DUMP: [0x0000006fff1020-0x0000007fff101f] 0x10000000 bytes, Dumped: 0x0
240 Contents when fadump is active during second kernel
242 # cat /sys/kernel/debug/powerpc/fadump_region
243 CPU : [0x0000006ffb0000-0x0000006fff001f] 0x40020 bytes, Dumped: 0x40020
244 HPTE: [0x0000006fff0020-0x0000006fff101f] 0x1000 bytes, Dumped: 0x1000
245 DUMP: [0x0000006fff1020-0x0000007fff101f] 0x10000000 bytes, Dumped: 0x10000000
246 : [0x00000010000000-0x0000006ffaffff] 0x5ffb0000 bytes, Dumped: 0x5ffb0000
248 NOTE: Please refer to Documentation/filesystems/debugfs.txt on
249 how to mount the debugfs filesystem.
252 TODO:
253 -----
254 o Need to come up with the better approach to find out more
255 accurate boot memory size that is required for a kernel to
256 boot successfully when booted with restricted memory.
257 o The fadump implementation introduces a fadump crash info structure
258 in the scratch area before the ELF core header. The idea of introducing
259 this structure is to pass some important crash info data to the second
260 kernel which will help second kernel to populate ELF core header with
261 correct data before it gets exported through /proc/vmcore. The current
262 design implementation does not address a possibility of introducing
263 additional fields (in future) to this structure without affecting
264 compatibility. Need to come up with the better approach to address this.
265 The possible approaches are:
266 1. Introduce version field for version tracking, bump up the version
267 whenever a new field is added to the structure in future. The version
268 field can be used to find out what fields are valid for the current
269 version of the structure.
270 2. Reserve the area of predefined size (say PAGE_SIZE) for this
271 structure and have unused area as reserved (initialized to zero)
272 for future field additions.
273 The advantage of approach 1 over 2 is we don't need to reserve extra space.
274 ---
275 Author: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
276 This document is based on the original documentation written for phyp
277 assisted dump by Linas Vepstas and Manish Ahuja.