-
Notifications
You must be signed in to change notification settings - Fork 3
/
IMPLEMENTATION
327 lines (278 loc) · 13.7 KB
/
IMPLEMENTATION
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
* The kdump-compressed format
makedumpfile provides two DUMPFILE formats (the ELF format and the
kdump-compressed format). By default, makedumpfile makes a DUMPFILE
in the kdump-compressed format. The kdump-compressed format is readable
only with the crash utility, and it can be smaller than the ELF format
because of the compression support.
- The file structure
File offset
+------------------------------------------+ 0x0
| main header (struct disk_dump_header) |
|------------------------------------------+ block 1
| sub header (struct kdump_sub_header) |
|------------------------------------------+ block 2
| 1st-bitmap |
|------------------------------------------+ block 2 + X blocks
| 2nd-bitmap | (aligned by block)
|------------------------------------------+ block 2 + 2 * X blocks
| page header for pfn 0 (struct page_desc) | (aligned by block)
| page header for pfn 1 (struct page_desc) |
| : |
| page header for pfn Z (struct page_desc) |
|------------------------------------------| (not aligned by block)
| page data (pfn 0) |
| page data (pfn 1) |
| : |
| page data (pfn Z) |
+------------------------------------------+ offset_eraseinfo
| erase mystruct2.mystruct1.var size 4 |
| erase mystruct2.mystruct1.ptr nullify |
| erase mystruct2.mystruct.array size 100 |
+------------------------------------------+
- main header
The main header of the kdump compressed format is the almost same as the
one of diskdump. This header has the following members, and the member
signature and header_version are different from diskdump.
struct disk_dump_header {
char signature[SIG_LEN]; /* = "KDUMP " */
int header_version; /* Dump header version */
struct new_utsname utsname; /* copy of system_utsname */
struct timeval timestamp; /* Time stamp */
unsigned int status; /* Above flags */
int block_size; /* Size of a block in byte */
int sub_hdr_size; /* Size of arch dependent
header in blocks */
unsigned int bitmap_blocks; /* Size of Memory bitmap in
block */
unsigned int max_mapnr; /* = max_mapnr, OBSOLETE!
32bit only, full 64bit
in sub header. */
unsigned int total_ram_blocks;/* Number of blocks should be
written */
unsigned int device_blocks; /* Number of total blocks in
* the dump device */
unsigned int written_blocks; /* Number of written blocks */
unsigned int current_cpu; /* CPU# which handles dump */
int nr_cpus; /* Number of CPUs */
struct task_struct *tasks[0];
};
- sub header
The sub header of the kdump compressed format is original. This header
has the member phys_base and dump_level. The member phys_base is for
an x86_64 relocatable kernel, and the member dump_level has '-d' option's
value of makedumpfile command.
struct kdump_sub_header {
unsigned long phys_base;
int dump_level; /* header_version 1 and later */
int split; /* header_version 2 and later */
unsigned long start_pfn; /* header_version 2 and later,
OBSOLETE! 32bit only, full
64bit in start_pfn_64. */
unsigned long end_pfn; /* header_version 2 and later,
OBSOLETE! 32bit only, full
64bit in end_pfn_64. */
off_t offset_vmcoreinfo;/* header_version 3 and later */
unsigned long size_vmcoreinfo; /* header_version 3 and later */
off_t offset_note; /* header_version 4 and later */
unsigned long size_note; /* header_version 4 and later */
off_t offset_eraseinfo; /* header_version 5 and later */
unsigned long size_eraseinfo; /* header_version 5 and later */
unsigned long long start_pfn_64; /* header_version 6 and later */
unsigned long long end_pfn_64; /* header_version 6 and later */
unsigned long long max_mapnr_64; /* header_version 6 and later */
};
- 1st-bitmap
The bit of 1st-bitmap presents either a page on memory hole, or not.
If a page is on memory hole, the corresponding bit is off. Otherwise,
it is on.
- 2nd-bitmap
The bit of 2nd-bitmap presents either a dumpable page, or not.
If a page is on memory hole or excluded by makedumpfile command, the
corresponding bit is off. Otherwise, it is on.
- page header
There are page headers corresponding to dumpable pages.
This header presents the corresponding page information (compressed, or not.
etc.)
typedef struct page_desc {
off_t offset; /* the offset of the page data*/
unsigned int size; /* the size of this dump page */
unsigned int flags; /* flags */
unsigned long long page_flags; /* page flags */
} page_desc_t;
* The ELF format
There are two different ELF format(ELF32, ELF64) for K-bit architectures (K=32,64).
Since they almost have the same behaviour in this situation, the following will use
ELF32 as a example.
- The file structure
+---------------------------------+
| elf_header (struct elf32_hdr) |
|---------------------------------+
| PT_NOTE (struct elf32_phdr) |
|---------------------------------+
| PT_LOAD(1) (struct elf32_phdr) |
| PT_LOAD(2) (struct elf32_phdr) |
| : |
| PT_LOAD(Z) (struct elf32_phdr) |
|---------------------------------+
| NOTE |
|---------------------------------+
| segment(1) |
| segment(2) |
| : |
| segment(Z) |
+---------------------------------+
- elf_header
This header is almost the same as a normal elf_header. The difference is that the
e_flags is used for indicating whether the dump file is complete or not.
0x0 : complete,
0x1 : incomplete
typedef struct elf32_hdr{
unsigned char e_ident[EI_NIDENT]; /* ELF "magic number" */
Elf32_Half e_type; /* Object file type (CORE) */
Elf32_Half e_machine;
Elf32_Word e_version;
Elf32_Addr e_entry;
Elf32_Off e_phoff; /* Program header table file offset */
Elf32_Off e_shoff;
Elf32_Word e_flags;
Elf32_Half e_ehsize; /* Size of this header */
Elf32_Half e_phentsize; /* Size of program headers */
Elf32_Half e_phnum; /* Number of program headers */
Elf32_Half e_shentsize;
Elf32_Half e_shnum;
Elf32_Half e_shstrndx;
} Elf32_Ehdr;
- PT_NOTE and PT_LOAD
PT_NOTE corresponds to NOTE and PT_LOAD to segment.
They present the corresponding NOTE and segments information.
typedef struct elf32_phdr{
Elf32_Word p_type;
Elf32_Off p_offset; /* Segment file offset */
Elf32_Addr p_vaddr; /* Segment virtual address */
Elf32_Addr p_paddr; /* Segment physical address */
Elf32_Word p_filesz; /* Segment size in file */
Elf32_Word p_memsz; /* Segment size in memory */
Elf32_Word p_flags;
Elf32_Word p_align; /* Segment alignment, file & memory */
} Elf32_Phdr;
- note
The note structure
+------------------------------------+
| note header 1 (struct elf32_note) |
| note header 2 (struct elf32_note) |
| : |
| note header N (struct elf32_note) |
|------------------------------------+
| note data 1 |
| note data 2 |
| : |
| note data N |
+------------------------------------+
typedef struct elf32_note {
Elf32_Word n_namesz; /* Name size */
Elf32_Word n_descsz; /* Content size */
Elf32_Word n_type; /* Content type */
} Elf32_Nhdr;
- segments
The data dumped are all stored in segments and notes.
* The incomplete DUMPFILE
When generating DUMPFILE, if ENOSPACE error happens, the DUMPFILE will be
incomplete.
- The incomplete kdump-compressed DUMPFILE
the complete the incomplete
+-----------------------+ +-----------------------+
| main header | | main header | have incomplete flag
|-----------------------+ |-----------------------+
| sub header | | sub header |
|-----------------------+ |-----------------------+
| 1st-bitmap | | 1st-bitmap |
|-----------------------+ |-----------------------+
| 2nd-bitmap | | 2nd-bitmap |
|-----------------------+ |-----------------------+
| page header for pfn 0 | | page header for pfn 0 |
| page header for pfn 1 | | page header for pfn 1 |
| : | | : |
| page header for pfn N | | page header for pfn N | The page headers after
| : | | | N don't exist. The
| page header for pfn Z | | | value of it is zero,
|-----------------------| |-----------------------| when try to read it.
| page data (pfn 0) | | page data (pfn 0) |
| page data (pfn 1) | | page data (pfn 1) |
| : | | : |
| page data (pfn N) | | page data (pfn N) |
| : | +-----------------------+
| page data (pfn Z) |
+-----------------------+
The incomplete flag is set into status of disk_dump_header by
status |= DUMP_DH_COMPRESSED_INCOMPLETE
DUMP_DH_COMPRESSED_INCOMPLETE : 0x8
The page header and page data are written in pairs. When writing page data
(pfn N+1), if ENOSPACE error happens, the page headers after N won't be
written either.
Since the data lost is filled with zero when it is read, the page_desc->offset
will also be zero. And zero page has its own offset not equal 0. So when reading
page from incomplete core, only the page lost by ENOSPACE errors has 0 in its
corresponding page descriptor's member offset.
If there is no page data dumped into the DUMPFILE, the DUMPFILE can't be
analysed by crash.
- The incomplete elf DUMPFILE
the complete the incomplete
+-------------+ +-------------+
| elf header | | elf header | have incomplete flag
|-------------+ |-------------+
| PT_NOTE | | PT_NOTE |
| PT_LOAD(1) | | PT_LOAD(1) |
| : | | : |
| PT_LOAD(N) | | PT_LOAD(N) | The PT_LOAD after N don't exist.
| : | | | The value of it is zero, when try
| PT_LOAD(Z) | | | to read it.
|-------------+ |-------------+
| NOTE | | NOTE |
|-------------+ |-------------+
| segment(1) | | segment(1) |
| : | | : |
| segment(N) | | segment(N) | The segment(N) is incomplete.
| : | +-------------+ The segments after N don't exist.
| segment(Z) |
+-------------+
The incomplete flag is set into e_flags of elf_header by
e_flags |= DUMP_ELF_INCOMPLETE
DUMP_ELF_INCOMPLETE : 0x1
The PT_LOAD and segment are written in pairs. When writing segment(N)
, if ENOSPACE error happens, the PT_LOAD after N won't be written
either.
If there is no segment dumped into the DUMPFILE, the DUMPFILE can't be
analysed by crash.
* The flattened format
This format is mostly used to send dump data to a remote host via SSH.
The kdump-compressed format and ELF format require random access to create
a DUMPFILE, so it cannot be written directly to standard output.
The flattened format adds the original offset and size of each data block,
and it can be rearranged to the original format on the remote host.
- File structure
This is the case of the kdump-compressed format.
File offset
+------------------------------------------+ 0x0
| flat header (struct makedumpfile_header) |
| |
|------------------------------------------| 0x1000 (4096)
| flat data header |
| (struct makedumpfile_data_header) |
| main header (struct disk_dump_header) |
|------------------------------------------| (not aligned)
| flat data header |
| : |
Note: For the ELF format, ELF header is not always the first data.
- flat header
struct makedumpfile_header {
char signature[SIG_LEN_MDF]; /* = "makedumpfile\0\0\0\0" */
int64_t type; /* = TYPE_FLAT_HEADER (1) */
int64_t version; /* = VERSION_FLAT_HEADER (1) */
};
Note: the type and version are big endian.
- flat data header
struct makedumpfile_data_header {
int64_t offset;
int64_t buf_size;
};
Note: the offset and buf_size are big endian.