File built-in-where of Package kernel-source - j0ke.net Open Build Service

File built-in-where of Package kernel-source (Revision 33db295fb1d6ebd6ec1d1f7753f54299)

Currently displaying revision 33db295fb1d6ebd6ec1d1f7753f54299, show latest

#! /bin/bash

sourcedir=${0%/*}

# A lot of symbols are exported by the main kernel image. Find out
# more precisely which built-in.o file defines them, and fill in
# that information in Module.symvers. (The built-in.o files are
# linked together from one or more object files in a directory.)
#   We use this information to better group symbols by subsystems.
#
# Usage: built-in-where < Module.symvers

unset LANG ${!LC_*}

# Create a table of all symbol export in a built-in.o file, e.g.,
# 0xc87c1f84   ktime_get   kernel/built-in   EXPORT_SYMBOL_GPL
built_in_exports() {
    # a/b/c/built-in.o gets linked into a/b/built-in.o, so ensure
    # that we visit sub-directories first to split up symbols as
    # much as possible.
    for obj in $(find -name built-in.o -printf '%d %P\n' \
		| sort -r \
		| awk '{ print $2 }'); do
	$sourcedir/list-exported-symbols -n ${obj%.o} $obj
    done

# We could go through the libraries as well, but those functions
    # are so unlikely to change that this wouldn't help.
    # (All remaining symbols will end up in the vmlinux set.)
    #for archive in $(find -name '*.a'); do
    #	$sourcedir/list-exported-symbols -n ${archive%.a} $archive
    #done
}

# Filter out duplicates from a Module.symvers dump
unique_symbols() {
    awk '
      { if ($2 in seen)
	  next
	seen[$2] = 1
	print
      }
    '
}

# Join together the two tables, including all lines from the first
# file that don't have a match in the second.
# Finally, remove the duplicate columns.
join -t $'\t' -j 2 -a 1 \
    <(sort -k2) \
    <(built_in_exports | unique_symbols | sort -k2) \
| awk '
BEGIN	{ FS = "\t" ; OFS = "\t" }
NF == 7 { print $2, $1, $6, $4 }
NF == 4 { print $2, $1, $3, $4 }
'