[PATCH] kcov: don't instrument lib/find_bit.c

From: andrey . konovalov
Date: Mon Jul 22 2024 - 18:37:43 EST


From: Andrey Konovalov <andreyknvl@xxxxxxxxx>

This file produces large amounts of flaky coverage not useful for the
KCOV's intended use case (guiding the fuzzing process).

Signed-off-by: Andrey Konovalov <andreyknvl@xxxxxxxxx>

---

I noticed this while running one of the syzkaller's programs.

In one run of the program, the number of KCOV entries amounts to ~300k,
with the top ones:

117285 /home/user/src/lib/find_bit.c:137 (discriminator 10)
116752 /home/user/src/lib/find_bit.c:137 (discriminator 3)
2455 /home/user/src/lib/vsprintf.c:2559
2033 /home/user/src/fs/kernfs/dir.c:317
1662 /home/user/src/fs/kernfs/kernfs-internal.h:72
...

In another run (that triggers exactly the same behavior in the kernel),
the amount of entries drops to ~110k:

7141 /home/user/src/lib/find_bit.c:137 (discriminator 10)
7110 /home/user/src/lib/find_bit.c:137 (discriminator 3)
2455 /home/user/src/lib/vsprintf.c:2559
2033 /home/user/src/fs/kernfs/dir.c:317
1662 /home/user/src/fs/kernfs/kernfs-internal.h:72
...

With this patch applied, the amount of KCOV entries for the same program
remains somewhat stable at ~100k.
---
lib/Makefile | 1 +
1 file changed, 1 insertion(+)

diff --git a/lib/Makefile b/lib/Makefile
index 322bb127b4dc..0fde1c360f32 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -14,6 +14,7 @@ KCOV_INSTRUMENT_list_debug.o := n
KCOV_INSTRUMENT_debugobjects.o := n
KCOV_INSTRUMENT_dynamic_debug.o := n
KCOV_INSTRUMENT_fault-inject.o := n
+KCOV_INSTRUMENT_find_bit.o := n

# string.o implements standard library functions like memset/memcpy etc.
# Use -ffreestanding to ensure that the compiler does not try to "optimize"
--
2.25.1