[TSan][NFC] fixup for comment of Shadow

[Enna1]

Summary

There should be 1-bit unused field between tid field and is_atomic field of Shadow.

[Enna1]

When we access is_atomic field, we use kAtomicShift, which is 48. And when we access tid field, we use kTidShift, which is 50.
If I Understand It Correctly, there should be 1-bit unused field between tid field and is_atomic field of Shadow.

[dvyukov]

I don't mind this change, but rtl-old should be deleted now.

[Enna1]

but rtl-old should be deleted now.

Hi @dvyukov, Is there any plan when rtl-old will be deleted, right now ? If so, I won't commit this patch.

BTW, AFAIK tsan v2 uses pure happen-befores similar to FastTrack(PLDI'09), Is there any change in v3, are we still use this algorithm ?
Thanks!

[dvyukov]

In D119417#3310365, @Enna1 wrote:

but rtl-old should be deleted now.

Hi @dvyukov, Is there any plan when rtl-old will be deleted, right now ? If so, I won't commit this patch.

@thakis does Chrome still need rtl-old?
It already misses some fixes that rtl/ has and may be build broken.

[dvyukov]

In D119417#3310365, @Enna1 wrote:

BTW, AFAIK tsan v2 uses pure happen-befores similar to FastTrack(PLDI'09), Is there any change in v3, are we still use this algorithm ?

I would say tsan v2 used algorithm similar to Relacy'08, but ok :)
v3 uses the same algorithm on high-level.

[Enna1]

In D119417#3313664, @dvyukov wrote:

In D119417#3310365, @Enna1 wrote:

BTW, AFAIK tsan v2 uses pure happen-befores similar to FastTrack(PLDI'09), Is there any change in v3, are we still use this algorithm ?

I would say tsan v2 used algorithm similar to Relacy'08, but ok :)
v3 uses the same algorithm on high-level.

Haha, I read tsan v2 use algorithm similar to FastTrack(PLDI'09) from "AddressSanitizer, ThreadSanitizer and MemorySanitizer -- Dynamic Testing Tools for C++ @ GTAC2013".
Is there any paper about the Relacy'08 algorithm? I only found your relacy github repo
I want to dive into the algorithm behind tsan.
Thanks!

[dvyukov]

In D119417#3313678, @Enna1 wrote:

Is there any paper about the Relacy'08 algorithm? I only found your relacy github repo

No, only the sources.

compiler-rt/lib/tsan/rtl-old/tsan_shadow.h

private:
   static const u64 kHistoryShift = kClkBits;
   static const u64 kHistoryMask = 7;
   u64 x_;
 };
 
 // Shadow (from most significant bit):
 //   freed           : 1
 //   tid             : kTidBits
+//   unused          : 1
 //   is_atomic       : 1
 //   is_read         : 1
 //   size_log        : 2
 //   addr0           : 3
 //   epoch           : kClkBits
 class Shadow : public FastState {
  public:
   explicit Shadow(u64 x) : FastState(x) {}