mirror of
https://github.com/git/git
synced 2024-08-27 19:49:21 +00:00
![]() 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@coredump.intra.peff.net/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Taylor Blau <me@ttaylorr.com> |
||
---|---|---|
.. | ||
tests | ||
.gitignore | ||
array.cocci | ||
commit.cocci | ||
equals-null.cocci | ||
flex_alloc.cocci | ||
free.cocci | ||
hashmap.cocci | ||
object_id.cocci | ||
preincr.cocci | ||
qsort.cocci | ||
README | ||
spatchcache | ||
strbuf.cocci | ||
swap.cocci | ||
the_repository.pending.cocci | ||
unused.cocci | ||
xcalloc.cocci | ||
xopen.cocci | ||
xstrdup_or_null.cocci |
This directory provides examples of Coccinelle (http://coccinelle.lip6.fr/) semantic patches that might be useful to developers. There are two types of semantic patches: * Using the semantic transformation to check for bad patterns in the code; The target 'make coccicheck' is designed to check for these patterns and it is expected that any resulting patch indicates a regression. The patches resulting from 'make coccicheck' are small and infrequent, so once they are found, they can be sent to the mailing list as per usual. Example for introducing new patterns: 67947c34ae (convert "hashcmp() != 0" to "!hasheq()", 2018-08-28) b84c783882 (fsck: s/++i > 1/i++/, 2018-10-24) Example of fixes using this approach: 248f66ed8e (run-command: use strbuf_addstr() for adding a string to a strbuf, 2018-03-25) f919ffebed (Use MOVE_ARRAY, 2018-01-22) These types of semantic patches are usually part of testing, c.f. 0860a7641b (travis-ci: fail if Coccinelle static analysis found something to transform, 2018-07-23) * Using semantic transformations in large scale refactorings throughout the code base. When applying the semantic patch into a real patch, sending it to the mailing list in the usual way, such a patch would be expected to have a lot of textual and semantic conflicts as such large scale refactorings change function signatures that are used widely in the code base. A textual conflict would arise if surrounding code near any call of such function changes. A semantic conflict arises when other patch series in flight introduce calls to such functions. So to aid these large scale refactorings, semantic patches can be used. However we do not want to store them in the same place as the checks for bad patterns, as then automated builds would fail. That is why semantic patches 'contrib/coccinelle/*.pending.cocci' are ignored for checks, and can be applied using 'make coccicheck-pending'. This allows to expose plans of pending large scale refactorings without impacting the bad pattern checks. Git-specific tips & things to know about how we run "spatch": * The "make coccicheck" will piggy-back on "COMPUTE_HEADER_DEPENDENCIES". If you've built a given object file the "coccicheck" target will consider its depednency to decide if it needs to re-run on the corresponding source file. This means that a "make coccicheck" will re-compile object files before running. This might be unexpected, but speeds up the run in the common case, as e.g. a change to "column.h" won't require all coccinelle rules to be re-run against "grep.c" (or another file that happens not to use "column.h"). To disable this behavior use the "SPATCH_USE_O_DEPENDENCIES=NoThanks" flag. * To speed up our rules the "make coccicheck" target will by default concatenate all of the *.cocci files here into an "ALL.cocci", and apply it to each source file. This makes the run faster, as we don't need to run each rule against each source file. See the Makefile for further discussion, this behavior can be disabled with "SPATCH_CONCAT_COCCI=". But since they're concatenated any <id> in the <rulname> (e.g. "@ my_name", v.s. anonymous "@@") needs to be unique across all our *.cocci files. You should only need to name rules if other rules depend on them (currently only one rule is named). * To speed up incremental runs even more use the "spatchcache" tool in this directory as your "SPATCH". It aimns to be a "ccache" for coccinelle, and piggy-backs on "COMPUTE_HEADER_DEPENDENCIES". It caches in Redis by default, see it source for a how-to. In one setup with a primed cache "make coccicheck" followed by a "make clean && make" takes around 10s to run, but 2m30s with the default of "SPATCH_CONCAT_COCCI=Y". With "SPATCH_CONCAT_COCCI=" the total runtime is around ~6m, sped up to ~1m with "spatchcache". Most of the 10s (or ~1m) being spent on re-running "spatch" on files we couldn't cache, as we didn't compile them (in contrib/* and compat/* mostly). The absolute times will differ for you, but the relative speedup from caching should be on that order.