Add a rather trivial "spatchcache", with this running e.g.: make cocciclean make contrib/coccinelle/free.cocci.patch \ SPATCH=contrib/coccicheck/spatchcache \ SPATCH_FLAGS=--very-quiet Is cut down from ~20s to ~5s on my system. Much of that is either fixable shell overhead, or the around 40 files we "CANTCACHE" (see the implementation). This uses "redis" as a cache by default, but it's configurable. See the embedded documentation. This is *not* like ccache in that we won't cache failed spatch invocations, or those where spatch suggests changes for us. Those cases are so rare that I didn't think it was worth the bother, by far the most common case is that it has no suggested changes. We'll also refuse to cache any "spatch" invocation that has output on stderr, which means that "--very-quiet" must be added to "SPATCH_FLAGS". Because we narrow the cache to that we don't need to save away stdout, stderr & the exit code. We simply cache the cases where we had no suggested changes. Another benchmark is to compare this with the previous SPATCH_BATCH_SIZE=N, as noted in [1]. Before this (on my 8 core system) running: make clean; time make contrib/coccinelle/array.cocci.patch SPATCH_BATCH_SIZE=0 Would take 33s, but with the preceding changes running without this "spatchcache" is slightly slower, or around 35s: make clean; time make contrib/coccinelle/array.cocci.patch Now doing the same with SPATCH=contrib/coccinelle/spatchcache will take around 6s, but we'll need to compile the *.o files first to take full advantage of it (which can be fast with "ccache"): make clean; make; time make contrib/coccinelle/array.cocci.patch SPATCH=contrib/coccinelle/spatchcache 1. https://lore.kernel.org/git/YwdRqP1CyUAzCEn2@xxxxxxxxxxxxxxxxxxxxxxx/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> --- contrib/coccinelle/spatchcache | 234 +++++++++++++++++++++++++++++++++ 1 file changed, 234 insertions(+) create mode 100755 contrib/coccinelle/spatchcache diff --git a/contrib/coccinelle/spatchcache b/contrib/coccinelle/spatchcache new file mode 100755 index 00000000000..f0b7b24e0ae --- /dev/null +++ b/contrib/coccinelle/spatchcache @@ -0,0 +1,234 @@ +#!/bin/sh +# +# spatchcache: a poor-man's "ccache"-alike for "spatch" in git.git +# +# This caching command relies on the peculiarities of the Makefile +# driving "spatch" in git.git, in particular if we invoke: +# +# make +# make coccicheck SPATCH_FLAGS=--very-quiet +# +# We can with COMPUTE_HEADER_DEPENDENCIES (auto-detected as true with +# "gcc" and "clang") write e.g. a .depend/grep.o.d for grep.c, when we +# compile grep.o. +# +# The .depend/grep.o.d will have the full header dependency tree of +# grep.c, and we can thus cache the output of "spatch" by: +# +# 1. Hashing all of those files +# 2. Hashing our source file, and the *.cocci rule we're +# applying +# 3. Running spatch, if suggests no changes (by far the common +# case) we invoke "spatchCache.getCmd" and +# "spatchCache.setCmd" with a hash SHA-256 to ask "does this +# ID have no changes" or "say that ID had no changes> +# 4. If no "spatchCache.{set,get}Cmd" is specified we'll use +# "redis-cli" and maintain a SET called "spatch-cache". Set +# appropriate redis memory policies to keep it from growing +# out of control. +# +# This along with the general incremental "make" support for +# "contrib/coccinelle" makes it viable to (re-)run coccicheck +# e.g. when merging integration branches. +# +# Note that the "--very-quiet" flag is currently critical. The cache +# will refuse to cache anything that has output on STDERR (which might +# be errors from spatch). The STDERR (and exit code) could in +# principle be cached (as with ccache), but then the simple structure +# in the Redis cache would need to change, so just supply +# "--very-quiet" for now. +# +# To use this, simply set SPATCH to +# contrib/coccinelle/spatchcache. Then optionally set: +# +# [spatchCache] +# # Optional: path to a custom spatch +# spatch = ~/g/coccicheck/spatch.opt +# +# As well as this trace config (debug implies trace): +# +# trace = false +# debug = false +# +# Setting "trace" to "true" allows for seeing when we have a cache HIT +# or MISS. To debug whether the cache is working do that, and run e.g.: +# +# redis-cli FLUSHALL +# <make && make coccicheck, as above> +# grep -hore HIT -e MISS -e SET -e NOCACHE -e CANTCACHE .build/contrib/coccinelle | sort | uniq -c +# 600 CANTCACHE +# 7365 MISS +# 7365 SET +# +# A subsequent "make cocciclean && make coccicheck" should then have +# all "HIT"'s and "CANTCACHE"'s. + +set -e + +## Our own configuration & options +debug=$(git config --bool "spatchCache.debug") +if test "$debug" != "true" +then + debug= +fi +if test -n "$debug" +then + set -x +fi + +trace=$(git config --bool "spatchCache.trace") +if test "$trace" != "true" +then + trace= +fi +if test -n "$debug" +then + # debug implies trace + trace=true +fi + +trace_it () { + if test -z "$trace" + then + return + fi + echo "$@" >&2 +} + +spatch=$(git config --path "spatchCache.spatch" || :) +if test -n "$spatch" +then + if test -n "$debug" + then + trace_it "custom spatchCache.spatch='$spatch'" + fi +else + spatch=spatch +fi + +set=$(git config spatchCache.setCmd || :) +get=$(git config spatchCache.getCmd || :) + +## Parse spatch()-like command-line for caching info +arg_sp= +arg_file= +args="$@" +spatch_opts() { + while test $# != 0 + do + arg_file="$1" + case "$1" in + --sp-file) + arg_sp="$2" + ;; + esac + shift + done +} +spatch_opts "$@" +if ! test -f "$arg_file" +then + arg_file= +fi + +hash_for_cache() { + # Parameters that should affect the cache + echo "spatch=$spatch" + echo "args=$args" + echo + + # Our target file and its dependencies + git hash-object "$1" "$2" $(grep -E -o '^[^:]+:$' "$3" | tr -d ':') +} + +# Sanity checks +if ! test -f "$arg_sp" && ! test -f "$arg_file" +then + echo $0: no idea how to cache "$@" >&2 + exit 128 +fi + +# Main logic +d=$(dirname "$arg_file") +b=$(basename "$arg_file") +bwoext="${b%.c}" +dep="$d/.depend/$bwoext.o.d" + +if ! test -f "$dep" +then + trace_it "$0: CANTCACHE have no '$dep' for '$arg_file'!" + exec "$spatch" "$@" +fi + +if test -n "$debug" +then + trace_it "$0: The full cache input for '$arg_sp' '$arg_file' '$dep'" + hash_for_cache "$arg_sp" "$arg_file" "$dep" >&2 +fi +sum=$(hash_for_cache "$arg_sp" "$arg_file" "$dep" | git hash-object --stdin) + +trace_it "$0: processing '$arg_file' with '$arg_sp' rule, and got hash '$sum' for it + '$dep'" + +getret= +if test -z "$get" +then + if test $(redis-cli SISMEMBER spatch-cache "$sum") = 1 + then + getret=0 + else + getret=1 + fi +else + $set "$sum" + getret=$? +fi + +if test "$getret" = 0 +then + trace_it "$0: HIT for '$arg_file' with '$arg_sp'" + exit 0 +else + trace_it "$0: MISS: for '$arg_file' with '$arg_sp'" +fi + +out="$(mktemp)" +err="$(mktemp)" + +set +e +"$spatch" "$@" >"$out" 2>"$err" +ret=$? +set -e + +if test $ret = 0 && ! test -s "$out" && ! test -s "$err" +then + rm -f "$out" "$err" + + trace_it "$0: SET: for '$arg_file' with '$arg_sp'" + + setret= + if test -z "$set" + then + if test $(redis-cli SADD spatch-cache "$sum") = 1 + then + setret=0 + else + setret=1 + fi + else + "$set" "$sum" + setret=$? + fi + + if test "$setret" != 0 + then + echo "FAILED to set '$sum' in cache!" >&2 + exit 128 + fi +else + trace_it "$0: NOCACHE (have changes): for '$arg_file' with '$arg_sp'" + cat "$out" + cat "$err" >&2 + rm -f "$out" "$err" + exit "$ret" +fi +rm -f "$out" "$err" -- 2.37.3.1420.g76f8a3d556c
