xref: /dragonfly/contrib/bmake/mk/meta2deps.sh (revision e95199c5) 
1#!/bin/sh
2
3# NAME:
4#	meta2deps.sh - extract useful info from .meta files
5#
6# SYNOPSIS:
7#	meta2deps.sh SB="SB" "meta" ...
8#
9# DESCRIPTION:
10#	This script looks each "meta" file and extracts the
11#	information needed to deduce build and src dependencies.
12#
13#	To do this, we extract the 'CWD' record as well as all the
14#	syscall traces which describe 'R'ead, 'C'hdir and 'E'xec
15#	syscalls.
16#
17#	The typical meta file looks like::
18#.nf
19#
20#	# Meta data file "path"
21#	CMD "command-line"
22#	CWD "cwd"
23#	TARGET "target"
24#	-- command output --
25#	-- filemon acquired metadata --
26#	# buildmon version 2
27#	V 2
28#	E "pid" "path"
29#	R "pid" "path"
30#	C "pid" "cwd"
31#	R "pid" "path"
32#	X "pid" "status"
33#.fi
34#
35#	The fact that all the syscall entry lines start with a single
36#	character make these files quite easy to process using sed(1).
37#
38#	To simplify the logic the 'CWD' line is made to look like a
39#	normal 'C'hdir entry, and "cwd" is remembered so that it can
40#	be prefixed to any "path" which is not absolute.
41#
42#	If the "path" being read ends in '.srcrel' it is the content
43#	of (actually the first line of) that file that we are
44#	interested in.
45#
46#	Any "path" which lies outside of the sandbox "SB" is generally
47#	not of interest and is ignored.
48#
49#	The output, is a set of absolute paths with "SB" like:
50#.nf
51#
52#	$SB/obj-i386/bsd/gnu/lib/csu
53#	$SB/obj-i386/bsd/gnu/lib/libgcc
54#	$SB/obj-i386/bsd/include
55#	$SB/obj-i386/bsd/lib/csu/i386-elf
56#	$SB/obj-i386/bsd/lib/libc
57#	$SB/src/bsd/include
58#	$SB/src/bsd/sys/i386/include
59#	$SB/src/bsd/sys/sys
60#	$SB/src/pan-release/rtsock
61#	$SB/src/pfe-shared/include/jnx
62#.fi
63#
64#	Which can then be further processed by 'gendirdeps.mk'
65#
66#	If we are passed 'DPDEPS='"dpdeps", then for each src file
67#	outside of "CURDIR" we read, we output a line like:
68#.nf
69#
70#	DPDEPS_$path += $RELDIR
71#.fi
72#
73#	with "$path" geting turned into reldir's, so that we can end
74#	up with a list of all the directories which depend on each src
75#	file in another directory.  This can allow for efficient yet
76#	complete testing of changes.
77
78
79# RCSid:
80#	$Id: meta2deps.sh,v 1.15 2020/11/08 06:31:08 sjg Exp $
81
82# Copyright (c) 2010-2013, Juniper Networks, Inc.
83# All rights reserved.
84#
85# Redistribution and use in source and binary forms, with or without
86# modification, are permitted provided that the following conditions
87# are met:
88# 1. Redistributions of source code must retain the above copyright
89#    notice, this list of conditions and the following disclaimer.
90# 2. Redistributions in binary form must reproduce the above copyright
91#    notice, this list of conditions and the following disclaimer in the
92#    documentation and/or other materials provided with the distribution.
93#
94# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
95# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
96# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
97# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
98# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
99# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
100# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
101# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
102# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
103# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
104# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
105
106meta2src() {
107    cat /dev/null "$@" |
108    sed -n '/^R .*\.[chyl]$/s,^..[0-9]* ,,p' |
109    sort -u
110}
111
112meta2dirs() {
113    cat /dev/null "$@" |
114    sed -n '/^R .*\/.*\.[a-z0-9][^\/]*$/s,^..[0-9]* \(.*\)/[^/]*$,\1,p' |
115    sort -u
116}
117
118add_list() {
119    sep=' '
120    suffix=
121    while :
122    do
123	case "$1" in
124	"|") sep="$1"; shift;;
125	-s) suffix="$2"; shift 2;;
126	*) break;;
127	esac
128    done
129    name=$1
130    shift
131    eval list="\$$name"
132    for top in "$@"
133    do
134	case "$sep$list$sep" in
135	*"$sep$top$suffix$sep"*) continue;;
136	esac
137	list="${list:+$list$sep}$top$suffix"
138    done
139    eval "$name=\"$list\""
140}
141
142_excludes_f() {
143    egrep -v "$EXCLUDES"
144}
145
146error() {
147    echo "ERROR: $@" >&2
148    exit 1
149}
150
151meta2deps() {
152    DPDEPS=
153    SRCTOPS=$SRCTOP
154    OBJROOTS=
155    EXCLUDES=
156    while :
157    do
158	case "$1" in
159	*=*) eval export "$1"; shift;;
160	-a) MACHINE_ARCH=$2; shift 2;;
161	-m) MACHINE=$2; shift 2;;
162	-C) CURDIR=$2; shift 2;;
163	-H) HOST_TARGET=$2; shift 2;;
164	-S) add_list SRCTOPS $2; shift 2;;
165	-O) add_list OBJROOTS $2; shift 2;;
166	-X) add_list EXCLUDES '|' $2; shift 2;;
167	-R) RELDIR=$2; shift 2;;
168	-T) TARGET_SPEC=$2; shift 2;;
169	*) break;;
170	esac
171    done
172
173    _th= _o=
174    case "$MACHINE" in
175    host) _ht=$HOST_TARGET;;
176    esac
177
178    for o in $OBJROOTS
179    do
180	case "$MACHINE,/$o/" in
181	host,*$HOST_TARGET*) ;;
182	*$MACHINE*|*${TARGET_SPEC:-$MACHINE}*) ;;
183	*) add_list _o $o; continue;;
184	esac
185	for x in $_ht $TARGET_SPEC $MACHINE
186	do
187	    case "$o" in
188	    "") continue;;
189	    */$x/) add_list _o ${o%$x/}; o=;;
190	    */$x) add_list _o ${o%$x}; o=;;
191	    *$x/) add_list _o ${o%$x/}; o=;;
192	    *$x) add_list _o ${o%$x}; o=;;
193	    esac
194	done
195    done
196    OBJROOTS="$_o"
197
198    case "$OBJTOP" in
199    "")
200	for o in $OBJROOTS
201	do
202	    OBJTOP=$o${TARGET_SPEC:-$MACHINE}
203	    break
204	done
205	;;
206    esac
207    src_re=
208    obj_re=
209    add_list '|' -s '/*' src_re $SRCTOPS
210    add_list '|' -s '*' obj_re $OBJROOTS
211
212    [ -z "$RELDIR" ] && unset DPDEPS
213    tf=/tmp/m2d$$-$USER
214    rm -f $tf.*
215    trap 'rm -f $tf.*; trap 0' 0
216
217    > $tf.dirdep
218    > $tf.qual
219    > $tf.srcdep
220    > $tf.srcrel
221    > $tf.dpdeps
222
223    seenit=
224    seensrc=
225    lpid=
226    case "$EXCLUDES" in
227    "") _excludes=cat;;
228    *) _excludes=_excludes_f;;
229    esac
230    # handle @list files
231    case "$@" in
232    *@[!.]*)
233	for f in "$@"
234	do
235	    case "$f" in
236	    *.meta) cat $f;;
237	    @*) xargs cat < ${f#@};;
238	    *) cat $f;;
239	    esac
240	done
241	;;
242    *) cat /dev/null "$@";;
243    esac 2> /dev/null |
244    sed -e 's,^CWD,C C,;/^[CREFLMV] /!d' -e "s,',,g" |
245    $_excludes | ( version=no
246    while read op pid path junk
247    do
248	: op=$op pid=$pid path=$path
249	# we track cwd and ldir (of interest) per pid
250	# CWD is bmake's cwd
251	case "$lpid,$pid" in
252	,C) CWD=$path cwd=$path ldir=$path
253	    if [ -z "$SB" ]; then
254		SB=`echo $CWD | sed 's,/obj.*,,'`
255	    fi
256	    SRCTOP=${SRCTOP:-$SB/src}
257	    case "$verion" in
258	    no) ;;		# ignore
259	    0) error "no filemon data";;
260	    *) ;;
261	    esac
262	    version=0
263	    continue
264	    ;;
265	$pid,$pid) ;;
266	*)
267	    case "$lpid" in
268	    "") ;;
269	    *) eval ldir_$lpid=$ldir;;
270	    esac
271	    eval ldir=\${ldir_$pid:-$CWD} cwd=\${cwd_$pid:-$CWD}
272	    lpid=$pid
273	    ;;
274	esac
275
276	case "$op,$path" in
277	V,*) version=$path; continue;;
278	W,*srcrel|*.dirdep) continue;;
279	C,*)
280	    case "$path" in
281	    /*) cwd=$path;;
282	    *) cwd=`cd $cwd/$path 2> /dev/null && /bin/pwd`;;
283	    esac
284	    # watch out for temp dirs that no longer exist
285	    test -d ${cwd:-/dev/null/no/such} || cwd=$CWD
286	    eval cwd_$pid=$cwd
287	    continue
288	    ;;
289	F,*) # $path is new pid
290	    eval cwd_$path=$cwd ldir_$path=$ldir
291	    continue
292	    ;;
293	*)  dir=${path%/*}
294	    case "$path" in
295	    $src_re|$obj_re) ;;
296	    /*/stage/*) ;;
297	    /*) continue;;
298	    *)	for path in $ldir/$path $cwd/$path
299		do
300			test -e $path && break
301		done
302		dir=${path%/*}
303		;;
304	    esac
305	    ;;
306	esac
307	# avoid repeating ourselves...
308	case "$DPDEPS,$seensrc," in
309	,*)
310	    case ",$seenit," in
311	    *,$dir,*) continue;;
312	    esac
313	    ;;
314	*,$path,*) continue;;
315	esac
316	# canonicalize if needed
317	case "/$dir/" in
318	*/../*|*/./*)
319	    rdir=$dir
320	    dir=`cd $dir 2> /dev/null && /bin/pwd`
321	    seen="$rdir,$dir"
322	    ;;
323	*)  seen=$dir;;
324	esac
325	case "$dir" in
326	${CURDIR:-.}|"") continue;;
327	$src_re)
328	    # avoid repeating ourselves...
329	    case "$DPDEPS,$seensrc," in
330	    ,*)
331		case ",$seenit," in
332		*,$dir,*) continue;;
333		esac
334		;;
335	    esac
336	    ;;
337	*)
338	    case ",$seenit," in
339	    *,$dir,*) continue;;
340	    esac
341	    ;;
342	esac
343	if [ -d $path ]; then
344	    case "$path" in
345	    */..) ldir=${dir%/*};;
346	    *) ldir=$path;;
347	    esac
348	    continue
349	fi
350	[ -f $path ] || continue
351	case "$dir" in
352	$CWD) continue;;		# ignore
353	$src_re)
354	    seenit="$seenit,$seen"
355	    echo $dir >> $tf.srcdep
356	    case "$DPDEPS,$reldir,$seensrc," in
357	    ,*) ;;
358	    *)	seensrc="$seensrc,$path"
359		echo "DPDEPS_$dir/${path##*/} += $RELDIR" >> $tf.dpdeps
360		;;
361	    esac
362	    continue
363	    ;;
364	esac
365	# if there is a .dirdep we cannot skip
366	# just because we've seen the dir before.
367	if [ -s $path.dirdep ]; then
368	    # this file contains:
369	    # '# ${RELDIR}.<machine>'
370	    echo $path.dirdep >> $tf.qual
371	    continue
372	elif [ -s $dir.dirdep ]; then
373	    echo $dir.dirdep >> $tf.qual
374	    seenit="$seenit,$seen"
375	    continue
376	fi
377	seenit="$seenit,$seen"
378	case "$dir" in
379	$obj_re)
380	    echo $dir;;
381	esac
382    done > $tf.dirdep
383    case "$version" in
384    0) error "no filemon data";;
385    esac ) || exit 1
386    _nl=echo
387    for f in $tf.dirdep $tf.qual $tf.srcdep
388    do
389	[ -s $f ] || continue
390	case $f in
391	*qual) # a list of .dirdep files
392	    # we can prefix everything with $OBJTOP to
393	    # tell gendirdeps.mk that these are
394	    # DIRDEP entries, since they are already
395	    # qualified with .<machine> as needed.
396	    # We strip .$MACHINE though
397	    xargs cat < $f | sort -u |
398	    sed "s,^# ,,;s,^,$OBJTOP/,;s,\.${TARGET_SPEC:-$MACHINE}\$,,;s,\.$MACHINE\$,,"
399	    ;;
400	*)  sort -u $f;;
401	esac
402	_nl=:
403    done
404    if [ -s $tf.dpdeps ]; then
405	case "$DPDEPS" in
406	*/*) ;;
407	*) echo > $DPDEPS;;		# the echo is needed!
408	esac
409	sort -u $tf.dpdeps |
410	sed "s,${SRCTOP}/,,;s,${SB_BACKING_SB:-$SB}/src/,," >> $DPDEPS
411    fi
412    # ensure we produce _something_ else egrep -v gets upset
413    $_nl
414}
415
416case /$0 in
417*/meta2dep*) meta2deps "$@";;
418*/meta2dirs*) meta2dirs "$@";;
419*/meta2src*) meta2src "$@";;
420esac
421