git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[RFC PATCH 14/15] config.mak.dev: add a DEVOPTS=analyzer mode to use GCC's -fanalyzer

From
Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date
Jun 3, 2022, 18:37 UTC
Message-ID
<RFC-patch-14.15-9cf550688d4-20220603T183608Z-avarab@gmail.com>
In-Reply-To
<RFC-cover-00.15-00000000000-20220603T183608Z-avarab@gmail.com>

Add an option to compile with GCC's -fanalyzer, which as noted in [1] and [2] has become much more useful in the recently released GCC v12 series.

Here we're quieting a few outstanding -fanalyzer issues that require us to use -Wno-error=* on an entire file:

 * range-diff.c, because it involves loop variables and would be
   painfully verbose to instrument with the ASSERT_FOR_FANALYZER() macro
   introduced in the subsequent commit.
 * http-fetch.c and fsmonitor-settings.c, because those aren't issues
   where we're referencing NULL, and therefore we can't quiet it with an
   assert().
For non-GCC compilers I considered wrapping the DEVOPTS logic in:
	ifneq ($(filter gcc10,$(COMPILER_FEATURES)),)
	endif

Which would make it OK to specify DEVOPTS=analyzer under other compilers, or on older GCC. But then we'd silently ignore the option on those. Let's instead trust the compiler to error out if it doesn't support -fanalyzer.

There are various limitations and bugs in the analyzer engine, e.g. I filed [3] for a false positive in builtin/merge-file.c before GCC v12 was released, which was subsequently fixed in GCC 12 trunk in [4], but many other issues remain.

1. https://developers.redhat.com/articles/2022/04/12/state-static-analysis-gcc-12-compiler
2. https://gcc.gnu.org/gcc-12/changes.html
3. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105264
4. https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=a358e4b60815b41e27f3508014ceb592f86b9b45
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
---
 Makefile       | 14 ++++++++++++
 config.mak.dev | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 75 insertions(+)
diff --git a/Makefile b/Makefile
index 18ca6744a50..129d55f5992 100644
--- a/Makefile
+++ b/Makefile
@@ -507,6 +507,20 @@ include shared.mak
 #    no-pedantic:
 #
 #        Disable -pedantic compilation.
+#
+#    analyzer:
+#
+#        Compile with GCC's -fanalyzer, this analysis is much more
+#        expensive than other GCC warnings.
+#
+#        The set of analysis flags is curated based on known issues
+#        and compiler version. Known issues are made into non-fatal
+#        warnings (even "no-error" isn't set).
+#
+#    no-suppress-analyzer:
+#
+#        When using "analyzer" disable the suppression of known
+#        -fanalyzer issues.
 
 GIT-VERSION-FILE: FORCE
 	@$(SHELL_PATH) ./GIT-VERSION-GEN
diff --git a/config.mak.dev b/config.mak.dev
index c3104f400b2..d6f5be92297 100644
--- a/config.mak.dev
+++ b/config.mak.dev
@@ -72,3 +72,64 @@ DEVELOPER_CFLAGS += -Wno-error=dangling-pointer
 endif
 
 GIT_TEST_PERL_FATAL_WARNINGS = YesPlease
+
+# GCC's -fanalyzer mode
+ifeq ($(filter analyzer,$(DEVOPTS)),analyzer)
+
+ifeq ($(filter gcc1,$(COMPILER_FEATURES)),)
+$(error you must be using a new-ish version of GCC for DEVOPTS=analyzer, your \
+$(CC) is not GCC at all!)
+endif
+
+DEVELOPER_CFLAGS += -fanalyzer
+
+## -fanalyzer exists exists as of gcc10, but versions older than gcc12
+## have a lot of false positives.
+ifeq ($(filter gcc12,$(COMPILER_FEATURES)),)
+DEVELOPER_CFLAGS += -Wno-analyzer-double-free
+DEVELOPER_CFLAGS += -Wno-analyzer-free-of-non-heap
+endif
+
+## Helper templates to whitelist existing issues
+define fn_disable_analyzer_tmpl
+$(1).sp $(1).s $(1).o: EXTRA_CPPFLAGS += $(2)
+
+endef
+
+define fn_disable_analyzer
+$(foreach f,$(2),$(call fn_disable_analyzer_tmpl,$(f),$(1)))
+endef
+
+## -Wno-error=analyzer-null-dereference
+$(eval $(call fn_disable_analyzer, \
+	-Wno-error=analyzer-null-dereference, \
+	range-diff \
+))
+## -Wno-error=analyzer-malloc-leak
+$(eval $(call fn_disable_analyzer, \
+	-Wno-error=analyzer-malloc-leak, \
+	fsmonitor-settings \
+))
+## per-GCC version annotations
+### -Wno-error=analyzer-use-of-uninitialized-value: gcc >= 12
+ifneq ($(filter gcc12,$(COMPILER_FEATURES)),)
+$(eval $(call fn_disable_analyzer, \
+	-Wno-error=analyzer-use-of-uninitialized-value, \
+	http-fetch \
+))
+else # gcc < 12
+### -Wno-error=analyzer-null-dereference: gcc == 11
+ifneq ($(filter gcc11,$(COMPILER_FEATURES)),)
+$(eval $(call fn_disable_analyzer, \
+	-Wno-error=analyzer-null-dereference, \
+	merge \
+	xdiff/xemit \
+	reftable/reader \
+))
+else
+$(error Your GCC version is too old for -fanalyze, or you are using \
+gcc10 which has it, but has too many false positives!)
+endif
+endif # gcc < 12
+
+endif
-- 
2.36.1.1124.g577fa9c2ebd
Previous: Ævar Arnfjörð BjarmasonNext: Ævar Arnfjörð Bjarmason
Message 38 of 51 in “Fix GCC -fanalyzer warnings & add -fanalyzer DEVOPTS mode”
  1. 00/15 Fix GCC -fanalyzer warnings & add -fanalyzer DEVOPTS modeÆvar Arnfjörð Bjarmason, Jun 3, 2022
  2. 01/15 remote.c: don't dereference NULL in freeing loopÆvar Arnfjörð Bjarmason, Jun 3, 2022
  3. René ScharfeJun 3, 2022
  4. Junio C HamanoJun 3, 2022
  5. Glen ChooJun 3, 2022
  6. Phillip WoodJun 4, 2022
  7. Ævar Arnfjörð BjarmasonJun 4, 2022
  8. 02/15 pull.c: don't feed NULL to strcmp() on get_rebase_fork_point() pathÆvar Arnfjörð Bjarmason, Jun 3, 2022
  9. René ScharfeJun 3, 2022
  10. 03/15 reftable: don't memset() a NULL from failed malloc()Ævar Arnfjörð Bjarmason, Jun 3, 2022
  11. René ScharfeJun 3, 2022
  12. Ævar Arnfjörð BjarmasonJun 4, 2022
  13. René ScharfeJun 4, 2022
  14. Ævar Arnfjörð BjarmasonJun 4, 2022
  15. René ScharfeJun 4, 2022
  16. Junio C HamanoJun 6, 2022
  17. Ævar Arnfjörð BjarmasonJun 6, 2022
  18. Junio C HamanoJun 6, 2022
  19. Ævar Arnfjörð BjarmasonJun 6, 2022
  20. 04/15 diff-lib.c: don't dereference NULL in oneway_diff()Ævar Arnfjörð Bjarmason, Jun 3, 2022
  21. René ScharfeJun 3, 2022
  22. 05/15 refs/packed-backend.c: add a BUG() if iter is NULLÆvar Arnfjörð Bjarmason, Jun 3, 2022
  23. René ScharfeJun 3, 2022
  24. 06/15 ref-filter.c: BUG() out on show_ref() with NULL refnameÆvar Arnfjörð Bjarmason, Jun 3, 2022
  25. René ScharfeJun 4, 2022
  26. 07/15 strbuf.c: placate -fanalyzer in strbuf_grow()Ævar Arnfjörð Bjarmason, Jun 3, 2022
  27. René ScharfeJun 4, 2022
  28. Phillip WoodJun 4, 2022
  29. Ævar Arnfjörð BjarmasonJun 4, 2022
  30. René ScharfeJun 4, 2022
  31. Phillip WoodJun 5, 2022
  32. 08/15 strbuf.c: use st_add3(), not unsigned_add_overflows()Ævar Arnfjörð Bjarmason, Jun 3, 2022
  33. René ScharfeJun 4, 2022
  34. 09/15 add-patch: assert parse_diff() expectations with BUG()Ævar Arnfjörð Bjarmason, Jun 3, 2022
  35. Phillip WoodJun 4, 2022
  36. 10/15 reftable: don't have reader_get_block() confuse -fanalyzerÆvar Arnfjörð Bjarmason, Jun 3, 2022
  37. 11/15 blame.c: clarify the state of "final_commit" for -fanalyzerÆvar Arnfjörð Bjarmason, Jun 3, 2022
  38. 14/15 config.mak.dev: add a DEVOPTS=analyzer mode to use GCC's -fanalyzerÆvar Arnfjörð Bjarmason, Jun 3, 2022
  39. 12/15 pack.h: wrap write_*file*() functionsÆvar Arnfjörð Bjarmason, Jun 3, 2022
  40. 13/15 pack-write API: pass down "verify" not arbitrary flagsÆvar Arnfjörð Bjarmason, Jun 3, 2022
  41. 15/15 config.mak.dev: add and use ASSERT_FOR_FANALYZER() macroÆvar Arnfjörð Bjarmason, Jun 3, 2022
  42. Phillip WoodJun 4, 2022
  43. 0/3 remote API: fix -fanalyzer-spotted freeing issueÆvar Arnfjörð Bjarmason, Jun 7, 2022
  44. 1/3 remote.c: remove braces from one-statement "for"-loopsÆvar Arnfjörð Bjarmason, Jun 7, 2022
  45. 2/3 remote.c: don't dereference NULL in freeing loopÆvar Arnfjörð Bjarmason, Jun 7, 2022
  46. Junio C HamanoJun 7, 2022
  47. 3/3 remote API: don't buggily FREE_AND_NULL(), free() insteadÆvar Arnfjörð Bjarmason, Jun 7, 2022
  48. Glen ChooJun 7, 2022
  49. Junio C HamanoJun 7, 2022
  50. Junio C HamanoJun 7, 2022
  51. Junio C HamanoJun 7, 2022

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.