[PATCH] Makefile: suppress annotated leaks with certain ASan options

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



When building with `SANITIZE=leak`, we define `SUPPRESS_ANNOTATED_LEAKS`
in order to make the `UNLEAK()` macro work (without the aforementioned
define, `UNLEAK()` is a noop). This is from `UNLEAK()`'s introduction in
0e5bba53af (add UNLEAK annotation for reducing leak false positives,
2017-09-08), where `UNLEAK()` is a noop for performance reasons unless
we are using the leak sanitizer.

However, it is possible to use the leak sanitizer without
`SANITIZE=leak`. This happens when building with `SANITIZE=address` and
enabling the leak sanitizer via the `ASAN_OPTIONS` variable (by
including the string "detect_leaks=1").

This renders `UNLEAK()` useless when doing `SANITIZE=address` builds
which also use the leak checker.

Update our Makefile to pretend as if `SANITIZE=leak` was given when
`SANITIZE=address` is given and the leak checker is enabled via
`ASAN_OPTIONS`.

Playing around with all five options (two spelling "enabled", two
spelling "disabled", and the empty set of options) yields the correct
behavior:

    for opt in '' detect_leaks=1 detect_leaks=true detect_leaks=0 detect_leaks=false
    do
      echo "==> ${opt:-(nothing)}"
      make -B builtin/add.o V=1 SANITIZE=address ASAN_OPTIONS="$opt" 2>&1 |
        grep -o -- '-DSUPPRESS_ANNOTATED_LEAKS'
    done

gives us:

    ==> (nothing)
    -DSUPPRESS_ANNOTATED_LEAKS
    ==> detect_leaks=1
    -DSUPPRESS_ANNOTATED_LEAKS
    ==> detect_leaks=true
    -DSUPPRESS_ANNOTATED_LEAKS
    ==> detect_leaks=0
    ==> detect_leaks=false

Making it possible to rely on `UNLEAK()` when implicitly using the leak
checker via SANITIZE=address builds.

Signed-off-by: Taylor Blau <me@xxxxxxxxxxxx>
---
I found this while playing around with GitHub's ASan-enabled CI builds
for our internal fork following a merge with v2.38.3.

The check-chainlint recipe in t/Makefile started using "git diff" via
d00113ec34 (t/Makefile: apply chainlint.pl to existing self-tests,
2022-09-01), which triggered a leak in some of GitHub's custom code. I
was surprised when marking the variable with UNLEAK() didn't do the
trick, and ended up down this rabbit hole ;-).

 Makefile | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index db447d0738..b00bb8bd1e 100644
--- a/Makefile
+++ b/Makefile
@@ -1445,13 +1445,18 @@ ifneq ($(filter undefined,$(SANITIZERS)),)
 BASIC_CFLAGS += -DSHA1DC_FORCE_ALIGNED_ACCESS
 endif
 ifneq ($(filter leak,$(SANITIZERS)),)
-BASIC_CFLAGS += -DSUPPRESS_ANNOTATED_LEAKS
-BASIC_CFLAGS += -O0
 SANITIZE_LEAK = YesCompiledWithIt
 endif
 ifneq ($(filter address,$(SANITIZERS)),)
 NO_REGEX = NeededForASAN
 SANITIZE_ADDRESS = YesCompiledWithIt
+ifeq ($(filter $(patsubst detect_leaks=%,%,$(ASAN_OPTIONS)),0 false),)
+SANITIZE_LEAK = YesViaASanOptions
+endif
+endif
+ifneq ($(SANITIZE_LEAK),)
+BASIC_CFLAGS += -DSUPPRESS_ANNOTATED_LEAKS
+BASIC_CFLAGS += -O0
 endif
 endif

--
2.38.0.16.g393fd4c6db



[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]

  Powered by Linux