Re: [v4.6-rc7-183-g1410b74e4061]

[Sedat Dilek]

On 5/16/16, Sedat Dilek <sedat.dilek@xxxxxxxxx> wrote:
> On 5/16/16, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>> On Mon, May 16, 2016 at 07:42:35PM +0200, Sedat Dilek wrote:
>>
>>> Unfortunately, I could not reproduce this again with none of my
>>> 183-kernels.
>>> When I first hit a "chain_key collision" issue, it was hard to
>>> redproduce,
>>> so.
>>> Any idea, how I can "force" this?
>>
>> Nope; I wish I knew, that'd be so much easier to work with :/
>>
>> I'm hoping someone will report a reproducer, even something that
>> triggers once every 5-10 runs would be awesome.
>>
>> In any case, like I've explained before, nothing regressed as such, we
>> only added this new warning under DEBUG_LOCKDEP because we want to
>> better understand the condition that triggers it.
>>
>> If it bothers you, just turn off DEBUG_LOCKDEP and know that your kernel
>> is as reliable as it was before. OTOH, if you do keep it on, please
>> let me know if you can (semi) reliably trigger this, as I'd really like
>> to have a better understanding.
>>
>
> OK, I keep checking my logs.
>
> I refreshed your patch Ingo pointed me to.
>
> But it fails like this (on top of Linux v4.6 final)...
> [...]
>   if [ "" = "-pg" ]; then if [ kernel/locking/mutex-debug.o !=
> "scripts/mod/empty.o" ]; then ./scripts/recordmcount
> "kernel/locking/mutex-debug.o"; fi; fi;
>   mycompiler -Wp,-MD,kernel/locking/.lockdep.o.d  -nostdinc -isystem
> /usr/lib/gcc/x86_64-linux-gnu/4.9/include -nostdinc -isystem
> /usr/lib/gcc/x86_64-linux-gnu/4.9/include -I./arch/x86/include
> -Iarch/x86/include/generated/uapi -Iarch/x86/include/generated
> -Iinclude -I./arch/x86/include/uapi -Iarch/x86/include/generated/uapi
> -I./include/uapi -Iinclude/generated/uapi -include
> ./include/linux/kconfig.h -D__KERNEL__ -Wall -Wundef
> -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common
> -Werror-implicit-function-declaration -Wno-format-security -std=gnu89
> -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -mno-avx -m64 -falign-jumps=1
> -falign-loops=1 -mno-80387 -mno-fp-ret-in-387
> -mpreferred-stack-boundary=3 -mtune=generic -mno-red-zone
> -mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args
> -DCONFIG_X86_X32_ABI -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1
> -DCONFIG_AS_CFI_SECTIONS=1 -DCONFIG_AS_FXSAVEQ=1 -DCONFIG_AS_SSSE3=1
> -DCONFIG_AS_CRC32=1 -DCONFIG_AS_AVX=1 -DCONFIG_AS_AVX2=1 -pipe
> -Wno-sign-compare -fno-asynchronous-unwind-tables
> -fno-delete-null-pointer-checks -O2 --param=allow-store-data-races=0
> -Wframe-larger-than=1024 -fno-stack-protector
> -Wno-unused-but-set-variable -fno-omit-frame-pointer
> -fno-optimize-sibling-calls -fno-var-tracking-assignments -mfentry
> -DCC_USING_FENTRY -Wdeclaration-after-statement -Wno-pointer-sign
> -fno-strict-overflow -fconserve-stack -Werror=implicit-int
> -Werror=strict-prototypes -Werror=date-time -DCC_HAVE_ASM_GOTO
> -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(lockdep)"
> -D"KBUILD_MODNAME=KBUILD_STR(lockdep)" -c -o
> kernel/locking/.tmp_lockdep.o kernel/locking/lockdep.c
> kernel/locking/lockdep.c: In function 'print_chain_keys_held_locks':
> kernel/locking/lockdep.c:2034:2: error: too few arguments to function
> 'print_chain_key_iteration'
>   print_chain_key_iteration(hlock_next->class_idx, chain_key);
>   ^
> kernel/locking/lockdep.c:2006:12: note: declared here
>  static u64 print_chain_key_iteration(int class_idx, u64 chain_key,
> u64 prev_key)
>             ^
> make[4]: *** [kernel/locking/lockdep.o] Error 1
> make[3]: *** [kernel/locking] Error 2
> make[2]: *** [kernel] Error 2
> [...]
>

Is the attached fix correct?

- Sedat -

P.S.: Attached is a refreshed version of your original proposal patch
which does not compile correctly.
diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index 5dc21eb101b0..b771a691b5e8 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -2031,7 +2031,8 @@ print_chain_keys_held_locks(struct task_struct *curr, struct held_lock *hlock_ne
 		print_lock(hlock);
 	}
 
-	print_chain_key_iteration(hlock_next->class_idx, chain_key);
+	print_chain_key_iteration(hlock_next->class_idx, chain_key,
+				  hlock->prev_chain_key);
 	print_lock(hlock_next);
 }

Attachment: 0001-locking-lockdep-Some-more-additional-chain_key-colli.patch
Description: Binary data