Horton

Re: [PoC] Improve dead tuple storage for lazy vacuum

John Naylor <johncnaylorls@gmail.com>

View thread View on postgr.es
From: John Naylor <johncnaylorls@gmail.com>
To: Masahiko Sawada <sawada.mshk@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>, Nathan Bossart <nathandbossart@gmail.com>, Andres Freund <andres@anarazel.de>, Matthias van de Meent <boekewurm+postgres@gmail.com>, Yura Sokolov <y.sokolov@postgrespro.ru>
Date: 2024-01-09T11:19:46Z
Lists: pgsql-hackers 
On Tue, Jan 9, 2024 at 9:40 AM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> In addition, I've made some changes and cleanups:

These look good to me, although I have not tried dumping a node in a while.

> 0011 - simplify the radix tree iteration code. I hope it makes the
> code clear and readable. Also I removed RT_UPDATE_ITER_STACK().

I'm very pleased with how much simpler it is now!

> 0013 - In RT_SHMEM case, we use SIZEOF_VOID_P for
> RT_VALUE_IS_EMBEDDABLE check, but I think it's not correct. Because
> DSA has its own pointer size, SIZEOF_DSA_POINTER, it could be 4 bytes
> even if SIZEOF_VOID_P is 8 bytes, for example in a case where
> !defined(PG_HAVE_ATOMIC_U64_SUPPORT). Please refer to dsa.h for
> details.

Thanks for the pointer. ;-)

> BTW, now that the inner and leaf nodes use the same structure, do we
> still need RT_NODE_BASE_XXX types? Most places where we use
> RT_NODE_BASE_XXX types can be replaced with RT_NODE_XXX types.

That's been in the back of my mind as well. Maybe the common header
should be the new "base" member? At least, something other than "n".

> Exceptions are RT_FANOUT_XX calculations:
>
> #if SIZEOF_VOID_P < 8
> #define RT_FANOUT_16_LO ((96 - sizeof(RT_NODE_BASE_16)) / sizeof(RT_PTR_ALLOC))
> #define RT_FANOUT_48    ((512 - sizeof(RT_NODE_BASE_48)) / sizeof(RT_PTR_ALLOC))
> #else
> #define RT_FANOUT_16_LO ((160 - sizeof(RT_NODE_BASE_16)) / sizeof(RT_PTR_ALLOC))
> #define RT_FANOUT_48    ((768 - sizeof(RT_NODE_BASE_48)) / sizeof(RT_PTR_ALLOC))
> #endif                          /* SIZEOF_VOID_P < 8 */
>
> But I think we can replace them with offsetof(RT_NODE_16, children) etc.

That makes sense. Do you want to have a go at it, or shall I?

I think after that, the only big cleanup needed is putting things in a
more readable order. I can do that at a later date, and other
opportunities for beautification are pretty minor and localized.

Rationalizing locking is the only thing left that requires a bit of thought.

Commits

  1. radixtree: Fix SIGSEGV at update of embeddable value to non-embeddable.

  2. Get rid of anonymous struct

  3. Teach radix tree to embed values at runtime

  4. Teach TID store to skip bitmap for small numbers of offsets

  5. Use bump context for TID bitmaps stored by vacuum

  6. Fix alignment of stack variable

  7. Use TidStore for dead tuple TIDs storage during lazy vacuum.

  8. Rethink create and attach APIs of shared TidStore.

  9. Fix inconsistent function prototypes with function definitions.

  10. Fix a calculation in TidStoreCreate().

  11. Fix potential integer handling issue in radixtree.h.

  12. Add TIDStore, to store sets of TIDs (ItemPointerData) efficiently.

  13. Fix link error for test_radixtree module on Windows

  14. Blind attempt to fix ODR violations

  15. Fix incorrect format specifier for int64

  16. Fix redefinition of typedefs

  17. Add template for adaptive radix tree

  18. Fix signedness error in 9f225e992 for gcc

  19. Introduce helper SIMD functions for small byte arrays

  20. Optimize vacuuming of relations with no indexes.

  21. Add bound check before bsearch() for performance

  22. Allocate consecutive blocks during parallel seqscans