Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
kbuild: generate offset range data for builtin modules
Create file module.builtin.ranges that can be used to find where
built-in modules are located by their addresses. This will be useful for
tracing tools to find what functions are for various built-in modules.
The offset range data for builtin modules is generated using:
- modules.builtin: associates object files with module names
- vmlinux.map: provides load order of sections and offset of first member
per section
- vmlinux.o.map: provides offset of object file content per section
- .*.cmd: build cmd file with KBUILD_MODFILE
The generated data will look like:
.text 00000000-00000000 = _text
.text 0000baf0-0000cb10 amd_uncore
.text 0009bd10-0009c8e0 iosf_mbi
...
.text 00b9f080-00ba011a intel_skl_int3472_discrete
.text 00ba0120-00ba03c0 intel_skl_int3472_discrete intel_skl_int3472_tps68470
.text 00ba03c0-00ba08d6 intel_skl_int3472_tps68470
...
.data 00000000-00000000 = _sdata
.data 0000f020-0000f680 amd_uncore
For each ELF section, it lists the offset of the first symbol. This can
be used to determine the base address of the section at runtime.
Next, it lists (in strict ascending order) offset ranges in that section
that cover the symbols of one or more builtin modules. Multiple ranges
can apply to a single module, and ranges can be shared between modules.
The CONFIG_BUILTIN_MODULE_RANGES option controls whether offset range data
is generated for kernel modules that are built into the kernel image.
How it works:
1. The modules.builtin file is parsed to obtain a list of built-in
module names and their associated object names (the .ko file that
the module would be in if it were a loadable module, hereafter
referred to as <kmodfile>). This object name can be used to
identify objects in the kernel compile because any C or assembler
code that ends up into a built-in module will have the option
-DKBUILD_MODFILE=<kmodfile> present in its build command, and those
can be found in the .<obj>.cmd file in the kernel build tree.
If an object is part of multiple modules, they will all be listed
in the KBUILD_MODFILE option argument.
This allows us to conclusively determine whether an object in the
kernel build belong to any modules, and which.
2. The vmlinux.map is parsed next to determine the base address of each
top level section so that all addresses into the section can be
turned into offsets. This makes it possible to handle sections
getting loaded at different addresses at system boot.
We also determine an 'anchor' symbol at the beginning of each
section to make it possible to calculate the true base address of
a section at runtime (i.e. symbol address - symbol offset).
We collect start addresses of sections that are included in the top
level section. This is used when vmlinux is linked using vmlinux.o,
because in that case, we need to look at the vmlinux.o linker map to
know what object a symbol is found in.
And finally, we process each symbol that is listed in vmlinux.map
(or vmlinux.o.map) based on the following structure:
vmlinux linked from vmlinux.a:
vmlinux.map:
<top level section>
<included section> -- might be same as top level section)
<object> -- built-in association known
<symbol> -- belongs to module(s) object belongs to
...
vmlinux linked from vmlinux.o:
vmlinux.map:
<top level section>
<included section> -- might be same as top level section)
vmlinux.o -- need to use vmlinux.o.map
<symbol> -- ignored
...
vmlinux.o.map:
<section>
<object> -- built-in association known
<symbol> -- belongs to module(s) object belongs to
...
3. As sections, objects, and symbols are processed, offset ranges are
constructed in a straight-forward way:
- If the symbol belongs to one or more built-in modules:
- If we were working on the same module(s), extend the range
to include this object
- If we were working on another module(s), close that range,
and start the new one
- If the symbol does not belong to any built-in modules:
- If we were working on a module(s) range, close that range
Signed-off-by: Kris Van Hees <kris.van.hees@oracle.com>
Reviewed-by: Nick Alcock <nick.alcock@oracle.com>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Tested-by: Sam James <sam@gentoo.org>
Reviewed-by: Sami Tolvanen <samitolvanen@google.com>
Tested-by: Sami Tolvanen <samitolvanen@google.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
authored by
Kris Van Hees
and committed by
Masahiro Yamada
5f5e7344
23d93aa4
+559
+1
.gitignore
+1
.gitignore
+1
Documentation/dontdiff
+1
Documentation/dontdiff
+5
Documentation/kbuild/kbuild.rst
+5
Documentation/kbuild/kbuild.rst
···
22
22
This file contains modinfo from all modules that are built into the kernel.
23
23
Unlike modinfo of a separate module, all fields are prefixed with module name.
24
24
25
+
modules.builtin.ranges
26
+
----------------------
27
+
This file contains address offset ranges (per ELF section) for all modules
28
+
that are built into the kernel. Together with System.map, it can be used
29
+
to associate module names with symbols.
25
30
26
31
Environment variables
27
32
=====================
+7
Documentation/process/changes.rst
+7
Documentation/process/changes.rst
···
64
64
gtags (optional) 6.6.5 gtags --version
65
65
mkimage (optional) 2017.01 mkimage --version
66
66
Python (optional) 3.5.x python3 --version
67
+
GNU AWK (optional) 5.1.0 gawk --version
67
68
====================== =============== ========================================
68
69
69
70
.. [#f1] Sphinx is needed only to build the Kernel documentation
···
192
191
platforms. The tool is available via the ``u-boot-tools`` package or can be
193
192
built from the U-Boot source code. See the instructions at
194
193
https://docs.u-boot.org/en/latest/build/tools.html#building-tools-for-linux
194
+
195
+
GNU AWK
196
+
-------
197
+
198
+
GNU AWK is needed if you want kernel builds to generate address range data for
199
+
builtin modules (CONFIG_BUILTIN_MODULE_RANGES).
195
200
196
201
System utilities
197
202
****************
+1
Makefile
+1
Makefile
···
1482
1482
# Directories & files removed with 'make clean'
1483
1483
CLEAN_FILES += vmlinux.symvers modules-only.symvers \
1484
1484
modules.builtin modules.builtin.modinfo modules.nsdeps \
1485
+
modules.builtin.ranges vmlinux.o.map \
1485
1486
compile_commands.json rust/test \
1486
1487
rust-project.json .vmlinux.objs .vmlinux.export.c
1487
1488
+15
lib/Kconfig.debug
+15
lib/Kconfig.debug
···
571
571
pieces of code get eliminated with
572
572
CONFIG_LD_DEAD_CODE_DATA_ELIMINATION.
573
573
574
+
config BUILTIN_MODULE_RANGES
575
+
bool "Generate address range information for builtin modules"
576
+
depends on !LTO
577
+
depends on VMLINUX_MAP
578
+
help
579
+
When modules are built into the kernel, there will be no module name
580
+
associated with its symbols in /proc/kallsyms. Tracers may want to
581
+
identify symbols by module name and symbol name regardless of whether
582
+
the module is configured as loadable or not.
583
+
584
+
This option generates modules.builtin.ranges in the build tree with
585
+
offset ranges (per ELF section) for the module(s) they belong to.
586
+
It also records an anchor symbol to determine the load address of the
587
+
section.
588
+
574
589
config DEBUG_FORCE_WEAK_PER_CPU
575
590
bool "Force weak per-cpu definitions"
576
591
depends on DEBUG_KERNEL
+18
scripts/Makefile.vmlinux
+18
scripts/Makefile.vmlinux
···
33
33
vmlinux: scripts/link-vmlinux.sh vmlinux.o $(KBUILD_LDS) FORCE
34
34
+$(call if_changed_dep,link_vmlinux)
35
35
36
+
# module.builtin.ranges
37
+
# ---------------------------------------------------------------------------
38
+
ifdef CONFIG_BUILTIN_MODULE_RANGES
39
+
__default: modules.builtin.ranges
40
+
41
+
quiet_cmd_modules_builtin_ranges = GEN $@
42
+
cmd_modules_builtin_ranges = gawk -f $(real-prereqs) > $@
43
+
44
+
targets += modules.builtin.ranges
45
+
modules.builtin.ranges: $(srctree)/scripts/generate_builtin_ranges.awk \
46
+
modules.builtin vmlinux.map vmlinux.o.map FORCE
47
+
$(call if_changed,modules_builtin_ranges)
48
+
49
+
vmlinux.map: vmlinux
50
+
@:
51
+
52
+
endif
53
+
36
54
# Add FORCE to the prerequisites of a target to force it to be always rebuilt.
37
55
# ---------------------------------------------------------------------------
38
56
+3
scripts/Makefile.vmlinux_o
+3
scripts/Makefile.vmlinux_o
···
45
45
# Link of vmlinux.o used for section mismatch analysis
46
46
# ---------------------------------------------------------------------------
47
47
48
+
vmlinux-o-ld-args-$(CONFIG_BUILTIN_MODULE_RANGES) += -Map=$@.map
49
+
48
50
quiet_cmd_ld_vmlinux.o = LD $@
49
51
cmd_ld_vmlinux.o = \
50
52
$(LD) ${KBUILD_LDFLAGS} -r -o $@ \
53
+
$(vmlinux-o-ld-args-y) \
51
54
$(addprefix -T , $(initcalls-lds)) \
52
55
--whole-archive vmlinux.a --no-whole-archive \
53
56
--start-group $(KBUILD_VMLINUX_LIBS) --end-group \
+508
scripts/generate_builtin_ranges.awk
+508
scripts/generate_builtin_ranges.awk
···
1
+
#!/usr/bin/gawk -f
2
+
# SPDX-License-Identifier: GPL-2.0
3
+
# generate_builtin_ranges.awk: Generate address range data for builtin modules
4
+
# Written by Kris Van Hees <kris.van.hees@oracle.com>
5
+
#
6
+
# Usage: generate_builtin_ranges.awk modules.builtin vmlinux.map \
7
+
# vmlinux.o.map > modules.builtin.ranges
8
+
#
9
+
10
+
# Return the module name(s) (if any) associated with the given object.
11
+
#
12
+
# If we have seen this object before, return information from the cache.
13
+
# Otherwise, retrieve it from the corresponding .cmd file.
14
+
#
15
+
function get_module_info(fn, mod, obj, s) {
16
+
if (fn in omod)
17
+
return omod[fn];
18
+
19
+
if (match(fn, /\/[^/]+$/) == 0)
20
+
return "";
21
+
22
+
obj = fn;
23
+
mod = "";
24
+
fn = substr(fn, 1, RSTART) "." substr(fn, RSTART + 1) ".cmd";
25
+
if (getline s <fn == 1) {
26
+
if (match(s, /DKBUILD_MODFILE=['"]+[^'"]+/) > 0) {
27
+
mod = substr(s, RSTART + 16, RLENGTH - 16);
28
+
gsub(/['"]/, "", mod);
29
+
} else if (match(s, /RUST_MODFILE=[^ ]+/) > 0)
30
+
mod = substr(s, RSTART + 13, RLENGTH - 13);
31
+
}
32
+
close(fn);
33
+
34
+
# A single module (common case) also reflects objects that are not part
35
+
# of a module. Some of those objects have names that are also a module
36
+
# name (e.g. core). We check the associated module file name, and if
37
+
# they do not match, the object is not part of a module.
38
+
if (mod !~ / /) {
39
+
if (!(mod in mods))
40
+
mod = "";
41
+
}
42
+
43
+
gsub(/([^/ ]*\/)+/, "", mod);
44
+
gsub(/-/, "_", mod);
45
+
46
+
# At this point, mod is a single (valid) module name, or a list of
47
+
# module names (that do not need validation).
48
+
omod[obj] = mod;
49
+
50
+
return mod;
51
+
}
52
+
53
+
# Update the ranges entry for the given module 'mod' in section 'osect'.
54
+
#
55
+
# We use a modified absolute start address (soff + base) as index because we
56
+
# may need to insert an anchor record later that must be at the start of the
57
+
# section data, and the first module may very well start at the same address.
58
+
# So, we use (addr << 1) + 1 to allow a possible anchor record to be placed at
59
+
# (addr << 1). This is safe because the index is only used to sort the entries
60
+
# before writing them out.
61
+
#
62
+
function update_entry(osect, mod, soff, eoff, sect, idx) {
63
+
sect = sect_in[osect];
64
+
idx = sprintf("%016x", (soff + sect_base[osect]) * 2 + 1);
65
+
entries[idx] = sprintf("%s %08x-%08x %s", sect, soff, eoff, mod);
66
+
count[sect]++;
67
+
}
68
+
69
+
# (1) Build a lookup map of built-in module names.
70
+
#
71
+
# The first file argument is used as input (modules.builtin).
72
+
#
73
+
# Lines will be like:
74
+
# kernel/crypto/lzo-rle.ko
75
+
# and we record the object name "crypto/lzo-rle".
76
+
#
77
+
ARGIND == 1 {
78
+
sub(/kernel\//, ""); # strip off "kernel/" prefix
79
+
sub(/\.ko$/, ""); # strip off .ko suffix
80
+
81
+
mods[$1] = 1;
82
+
next;
83
+
}
84
+
85
+
# (2) Collect address information for each section.
86
+
#
87
+
# The second file argument is used as input (vmlinux.map).
88
+
#
89
+
# We collect the base address of the section in order to convert all addresses
90
+
# in the section into offset values.
91
+
#
92
+
# We collect the address of the anchor (or first symbol in the section if there
93
+
# is no explicit anchor) to allow users of the range data to calculate address
94
+
# ranges based on the actual load address of the section in the running kernel.
95
+
#
96
+
# We collect the start address of any sub-section (section included in the top
97
+
# level section being processed). This is needed when the final linking was
98
+
# done using vmlinux.a because then the list of objects contained in each
99
+
# section is to be obtained from vmlinux.o.map. The offset of the sub-section
100
+
# is recorded here, to be used as an addend when processing vmlinux.o.map
101
+
# later.
102
+
#
103
+
104
+
# Both GNU ld and LLVM lld linker map format are supported by converting LLVM
105
+
# lld linker map records into equivalent GNU ld linker map records.
106
+
#
107
+
# The first record of the vmlinux.map file provides enough information to know
108
+
# which format we are dealing with.
109
+
#
110
+
ARGIND == 2 && FNR == 1 && NF == 7 && $1 == "VMA" && $7 == "Symbol" {
111
+
map_is_lld = 1;
112
+
if (dbg)
113
+
printf "NOTE: %s uses LLVM lld linker map format\n", FILENAME >"/dev/stderr";
114
+
next;
115
+
}
116
+
117
+
# (LLD) Convert a section record fronm lld format to ld format.
118
+
#
119
+
# lld: ffffffff82c00000 2c00000 2493c0 8192 .data
120
+
# ->
121
+
# ld: .data 0xffffffff82c00000 0x2493c0 load address 0x0000000002c00000
122
+
#
123
+
ARGIND == 2 && map_is_lld && NF == 5 && /[0-9] [^ ]+$/ {
124
+
$0 = $5 " 0x"$1 " 0x"$3 " load address 0x"$2;
125
+
}
126
+
127
+
# (LLD) Convert an anchor record from lld format to ld format.
128
+
#
129
+
# lld: ffffffff81000000 1000000 0 1 _text = .
130
+
# ->
131
+
# ld: 0xffffffff81000000 _text = .
132
+
#
133
+
ARGIND == 2 && map_is_lld && !anchor && NF == 7 && raw_addr == "0x"$1 && $6 == "=" && $7 == "." {
134
+
$0 = " 0x"$1 " " $5 " = .";
135
+
}
136
+
137
+
# (LLD) Convert an object record from lld format to ld format.
138
+
#
139
+
# lld: 11480 11480 1f07 16 vmlinux.a(arch/x86/events/amd/uncore.o):(.text)
140
+
# ->
141
+
# ld: .text 0x0000000000011480 0x1f07 arch/x86/events/amd/uncore.o
142
+
#
143
+
ARGIND == 2 && map_is_lld && NF == 5 && $5 ~ /:\(/ {
144
+
gsub(/\)/, "");
145
+
sub(/ vmlinux\.a\(/, " ");
146
+
sub(/:\(/, " ");
147
+
$0 = " "$6 " 0x"$1 " 0x"$3 " " $5;
148
+
}
149
+
150
+
# (LLD) Convert a symbol record from lld format to ld format.
151
+
#
152
+
# We only care about these while processing a section for which no anchor has
153
+
# been determined yet.
154
+
#
155
+
# lld: ffffffff82a859a4 2a859a4 0 1 btf_ksym_iter_id
156
+
# ->
157
+
# ld: 0xffffffff82a859a4 btf_ksym_iter_id
158
+
#
159
+
ARGIND == 2 && map_is_lld && sect && !anchor && NF == 5 && $5 ~ /^[_A-Za-z][_A-Za-z0-9]*$/ {
160
+
$0 = " 0x"$1 " " $5;
161
+
}
162
+
163
+
# (LLD) We do not need any other ldd linker map records.
164
+
#
165
+
ARGIND == 2 && map_is_lld && /^[0-9a-f]{16} / {
166
+
next;
167
+
}
168
+
169
+
# (LD) Section records with just the section name at the start of the line
170
+
# need to have the next line pulled in to determine whether it is a
171
+
# loadable section. If it is, the next line will contains a hex value
172
+
# as first and second items.
173
+
#
174
+
ARGIND == 2 && !map_is_lld && NF == 1 && /^[^ ]/ {
175
+
s = $0;
176
+
getline;
177
+
if ($1 !~ /^0x/ || $2 !~ /^0x/)
178
+
next;
179
+
180
+
$0 = s " " $0;
181
+
}
182
+
183
+
# (LD) Object records with just the section name denote records with a long
184
+
# section name for which the remainder of the record can be found on the
185
+
# next line.
186
+
#
187
+
# (This is also needed for vmlinux.o.map, when used.)
188
+
#
189
+
ARGIND >= 2 && !map_is_lld && NF == 1 && /^ [^ \*]/ {
190
+
s = $0;
191
+
getline;
192
+
$0 = s " " $0;
193
+
}
194
+
195
+
# Beginning a new section - done with the previous one (if any).
196
+
#
197
+
ARGIND == 2 && /^[^ ]/ {
198
+
sect = 0;
199
+
}
200
+
201
+
# Process a loadable section (we only care about .-sections).
202
+
#
203
+
# Record the section name and its base address.
204
+
# We also record the raw (non-stripped) address of the section because it can
205
+
# be used to identify an anchor record.
206
+
#
207
+
# Note:
208
+
# Since some AWK implementations cannot handle large integers, we strip off the
209
+
# first 4 hex digits from the address. This is safe because the kernel space
210
+
# is not large enough for addresses to extend into those digits. The portion
211
+
# to strip off is stored in addr_prefix as a regexp, so further clauses can
212
+
# perform a simple substitution to do the address stripping.
213
+
#
214
+
ARGIND == 2 && /^\./ {
215
+
# Explicitly ignore a few sections that are not relevant here.
216
+
if ($1 ~ /^\.orc_/ || $1 ~ /_sites$/ || $1 ~ /\.percpu/)
217
+
next;
218
+
219
+
# Sections with a 0-address can be ignored as well.
220
+
if ($2 ~ /^0x0+$/)
221
+
next;
222
+
223
+
raw_addr = $2;
224
+
addr_prefix = "^" substr($2, 1, 6);
225
+
base = $2;
226
+
sub(addr_prefix, "0x", base);
227
+
base = strtonum(base);
228
+
sect = $1;
229
+
anchor = 0;
230
+
sect_base[sect] = base;
231
+
sect_size[sect] = strtonum($3);
232
+
233
+
if (dbg)
234
+
printf "[%s] BASE %016x\n", sect, base >"/dev/stderr";
235
+
236
+
next;
237
+
}
238
+
239
+
# If we are not in a section we care about, we ignore the record.
240
+
#
241
+
ARGIND == 2 && !sect {
242
+
next;
243
+
}
244
+
245
+
# Record the first anchor symbol for the current section.
246
+
#
247
+
# An anchor record for the section bears the same raw address as the section
248
+
# record.
249
+
#
250
+
ARGIND == 2 && !anchor && NF == 4 && raw_addr == $1 && $3 == "=" && $4 == "." {
251
+
anchor = sprintf("%s %08x-%08x = %s", sect, 0, 0, $2);
252
+
sect_anchor[sect] = anchor;
253
+
254
+
if (dbg)
255
+
printf "[%s] ANCHOR %016x = %s (.)\n", sect, 0, $2 >"/dev/stderr";
256
+
257
+
next;
258
+
}
259
+
260
+
# If no anchor record was found for the current section, use the first symbol
261
+
# in the section as anchor.
262
+
#
263
+
ARGIND == 2 && !anchor && NF == 2 && $1 ~ /^0x/ && $2 !~ /^0x/ {
264
+
addr = $1;
265
+
sub(addr_prefix, "0x", addr);
266
+
addr = strtonum(addr) - base;
267
+
anchor = sprintf("%s %08x-%08x = %s", sect, addr, addr, $2);
268
+
sect_anchor[sect] = anchor;
269
+
270
+
if (dbg)
271
+
printf "[%s] ANCHOR %016x = %s\n", sect, addr, $2 >"/dev/stderr";
272
+
273
+
next;
274
+
}
275
+
276
+
# The first occurrence of a section name in an object record establishes the
277
+
# addend (often 0) for that section. This information is needed to handle
278
+
# sections that get combined in the final linking of vmlinux (e.g. .head.text
279
+
# getting included at the start of .text).
280
+
#
281
+
# If the section does not have a base yet, use the base of the encapsulating
282
+
# section.
283
+
#
284
+
ARGIND == 2 && sect && NF == 4 && /^ [^ \*]/ && !($1 in sect_addend) {
285
+
if (!($1 in sect_base)) {
286
+
sect_base[$1] = base;
287
+
288
+
if (dbg)
289
+
printf "[%s] BASE %016x\n", $1, base >"/dev/stderr";
290
+
}
291
+
292
+
addr = $2;
293
+
sub(addr_prefix, "0x", addr);
294
+
addr = strtonum(addr);
295
+
sect_addend[$1] = addr - sect_base[$1];
296
+
sect_in[$1] = sect;
297
+
298
+
if (dbg)
299
+
printf "[%s] ADDEND %016x - %016x = %016x\n", $1, addr, base, sect_addend[$1] >"/dev/stderr";
300
+
301
+
# If the object is vmlinux.o then we will need vmlinux.o.map to get the
302
+
# actual offsets of objects.
303
+
if ($4 == "vmlinux.o")
304
+
need_o_map = 1;
305
+
}
306
+
307
+
# (3) Collect offset ranges (relative to the section base address) for built-in
308
+
# modules.
309
+
#
310
+
# If the final link was done using the actual objects, vmlinux.map contains all
311
+
# the information we need (see section (3a)).
312
+
# If linking was done using vmlinux.a as intermediary, we will need to process
313
+
# vmlinux.o.map (see section (3b)).
314
+
315
+
# (3a) Determine offset range info using vmlinux.map.
316
+
#
317
+
# Since we are already processing vmlinux.map, the top level section that is
318
+
# being processed is already known. If we do not have a base address for it,
319
+
# we do not need to process records for it.
320
+
#
321
+
# Given the object name, we determine the module(s) (if any) that the current
322
+
# object is associated with.
323
+
#
324
+
# If we were already processing objects for a (list of) module(s):
325
+
# - If the current object belongs to the same module(s), update the range data
326
+
# to include the current object.
327
+
# - Otherwise, ensure that the end offset of the range is valid.
328
+
#
329
+
# If the current object does not belong to a built-in module, ignore it.
330
+
#
331
+
# If it does, we add a new built-in module offset range record.
332
+
#
333
+
ARGIND == 2 && !need_o_map && /^ [^ ]/ && NF == 4 && $3 != "0x0" {
334
+
if (!(sect in sect_base))
335
+
next;
336
+
337
+
# Turn the address into an offset from the section base.
338
+
soff = $2;
339
+
sub(addr_prefix, "0x", soff);
340
+
soff = strtonum(soff) - sect_base[sect];
341
+
eoff = soff + strtonum($3);
342
+
343
+
# Determine which (if any) built-in modules the object belongs to.
344
+
mod = get_module_info($4);
345
+
346
+
# If we are processing a built-in module:
347
+
# - If the current object is within the same module, we update its
348
+
# entry by extending the range and move on
349
+
# - Otherwise:
350
+
# + If we are still processing within the same main section, we
351
+
# validate the end offset against the start offset of the
352
+
# current object (e.g. .rodata.str1.[18] objects are often
353
+
# listed with an incorrect size in the linker map)
354
+
# + Otherwise, we validate the end offset against the section
355
+
# size
356
+
if (mod_name) {
357
+
if (mod == mod_name) {
358
+
mod_eoff = eoff;
359
+
update_entry(mod_sect, mod_name, mod_soff, eoff);
360
+
361
+
next;
362
+
} else if (sect == sect_in[mod_sect]) {
363
+
if (mod_eoff > soff)
364
+
update_entry(mod_sect, mod_name, mod_soff, soff);
365
+
} else {
366
+
v = sect_size[sect_in[mod_sect]];
367
+
if (mod_eoff > v)
368
+
update_entry(mod_sect, mod_name, mod_soff, v);
369
+
}
370
+
}
371
+
372
+
mod_name = mod;
373
+
374
+
# If we encountered an object that is not part of a built-in module, we
375
+
# do not need to record any data.
376
+
if (!mod)
377
+
next;
378
+
379
+
# At this point, we encountered the start of a new built-in module.
380
+
mod_name = mod;
381
+
mod_soff = soff;
382
+
mod_eoff = eoff;
383
+
mod_sect = $1;
384
+
update_entry($1, mod, soff, mod_eoff);
385
+
386
+
next;
387
+
}
388
+
389
+
# If we do not need to parse the vmlinux.o.map file, we are done.
390
+
#
391
+
ARGIND == 3 && !need_o_map {
392
+
if (dbg)
393
+
printf "Note: %s is not needed.\n", FILENAME >"/dev/stderr";
394
+
exit;
395
+
}
396
+
397
+
# (3) Collect offset ranges (relative to the section base address) for built-in
398
+
# modules.
399
+
#
400
+
401
+
# (LLD) Convert an object record from lld format to ld format.
402
+
#
403
+
ARGIND == 3 && map_is_lld && NF == 5 && $5 ~ /:\(/ {
404
+
gsub(/\)/, "");
405
+
sub(/:\(/, " ");
406
+
407
+
sect = $6;
408
+
if (!(sect in sect_addend))
409
+
next;
410
+
411
+
sub(/ vmlinux\.a\(/, " ");
412
+
$0 = " "sect " 0x"$1 " 0x"$3 " " $5;
413
+
}
414
+
415
+
# (3b) Determine offset range info using vmlinux.o.map.
416
+
#
417
+
# If we do not know an addend for the object's section, we are interested in
418
+
# anything within that section.
419
+
#
420
+
# Determine the top-level section that the object's section was included in
421
+
# during the final link. This is the section name offset range data will be
422
+
# associated with for this object.
423
+
#
424
+
# The remainder of the processing of the current object record follows the
425
+
# procedure outlined in (3a).
426
+
#
427
+
ARGIND == 3 && /^ [^ ]/ && NF == 4 && $3 != "0x0" {
428
+
osect = $1;
429
+
if (!(osect in sect_addend))
430
+
next;
431
+
432
+
# We need to work with the main section.
433
+
sect = sect_in[osect];
434
+
435
+
# Turn the address into an offset from the section base.
436
+
soff = $2;
437
+
sub(addr_prefix, "0x", soff);
438
+
soff = strtonum(soff) + sect_addend[osect];
439
+
eoff = soff + strtonum($3);
440
+
441
+
# Determine which (if any) built-in modules the object belongs to.
442
+
mod = get_module_info($4);
443
+
444
+
# If we are processing a built-in module:
445
+
# - If the current object is within the same module, we update its
446
+
# entry by extending the range and move on
447
+
# - Otherwise:
448
+
# + If we are still processing within the same main section, we
449
+
# validate the end offset against the start offset of the
450
+
# current object (e.g. .rodata.str1.[18] objects are often
451
+
# listed with an incorrect size in the linker map)
452
+
# + Otherwise, we validate the end offset against the section
453
+
# size
454
+
if (mod_name) {
455
+
if (mod == mod_name) {
456
+
mod_eoff = eoff;
457
+
update_entry(mod_sect, mod_name, mod_soff, eoff);
458
+
459
+
next;
460
+
} else if (sect == sect_in[mod_sect]) {
461
+
if (mod_eoff > soff)
462
+
update_entry(mod_sect, mod_name, mod_soff, soff);
463
+
} else {
464
+
v = sect_size[sect_in[mod_sect]];
465
+
if (mod_eoff > v)
466
+
update_entry(mod_sect, mod_name, mod_soff, v);
467
+
}
468
+
}
469
+
470
+
mod_name = mod;
471
+
472
+
# If we encountered an object that is not part of a built-in module, we
473
+
# do not need to record any data.
474
+
if (!mod)
475
+
next;
476
+
477
+
# At this point, we encountered the start of a new built-in module.
478
+
mod_name = mod;
479
+
mod_soff = soff;
480
+
mod_eoff = eoff;
481
+
mod_sect = osect;
482
+
update_entry(osect, mod, soff, mod_eoff);
483
+
484
+
next;
485
+
}
486
+
487
+
# (4) Generate the output.
488
+
#
489
+
# Anchor records are added for each section that contains offset range data
490
+
# records. They are added at an adjusted section base address (base << 1) to
491
+
# ensure they come first in the second records (see update_entry() above for
492
+
# more information).
493
+
#
494
+
# All entries are sorted by (adjusted) address to ensure that the output can be
495
+
# parsed in strict ascending address order.
496
+
#
497
+
END {
498
+
for (sect in count) {
499
+
if (sect in sect_anchor) {
500
+
idx = sprintf("%016x", sect_base[sect] * 2);
501
+
entries[idx] = sect_anchor[sect];
502
+
}
503
+
}
504
+
505
+
n = asorti(entries, indices);
506
+
for (i = 1; i <= n; i++)
507
+
print entries[indices[i]];
508
+
}