git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH 04/19] index-helper: new daemon for caching index and related stuff

From
Junio C Hamano <gitster@pobox.com>
Date
Mar 9, 2016, 23:09 UTC
Message-ID
<xmqqtwkffdus.fsf@gitster.mtv.corp.google.com>
In-Reply-To
<1457548582-28302-5-git-send-email-dturner@twopensource.com>
David Turner <dturner@twopensource.com> writes:
Show 24 quoted lines
> From: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
>
> Instead of reading the index from disk and worrying about disk
> corruption, the index is cached in memory (memory bit-flips happen
> too, but hopefully less often). The result is faster read. Read time
> is reduced by 70%.
>
> The biggest gain is not having to verify the trailing SHA-1, which
> takes lots of time especially on large index files. But this also
> opens doors for further optimiztions:
>
>  - we could create an in-memory format that's essentially the memory
>    dump of the index to eliminate most of parsing/allocation
>    overhead. The mmap'd memory can be used straight away. Experiment
>    [1] shows we could reduce read time by 88%.
>
>  - we could cache non-index info such as name hash
>
> The shared memory's name folows the template "git-<object>-<SHA1>"
> where <SHA1> is the trailing SHA-1 of the index file. <object> is
> "index" for cached index files (and may be "name-hash" for name-hash
> cache). If such shared memory exists, it contains the same index
> content as on disk. The content is already validated by the daemon and
> git won't validate it again (except comparing the trailing SHA-1s).

This indeed is an interesting approach; what is not explained but must be is when the on-disk index is updated to reflect the reality (if I am reading the explanation and the code right, while the daemon is running, its in-core cache becomes the source of truth by forcing everybody's read-index-from() to go to the daemon). The explanation could be "this is only for read side, and updating the index happens via the traditional 'write a new file and rename it to the final place' codepath, at which time the daemon must be told to re-read it."

Previous: David TurnerNext: Junio C Hamano
Message 11 of 60 in “index-helper, watchman”
  1. 00/19 index-helper, watchmanDavid Turner, Mar 9, 2016
  2. 01/19 trace.c: add GIT_TRACE_PACK_STATS for pack usage statisticsDavid Turner, Mar 9, 2016
  3. Junio C HamanoMar 9, 2016
  4. David TurnerMar 10, 2016
  5. Duy NguyenMar 10, 2016
  6. 02/19 read-cache.c: fix constness of verify_hdr()David Turner, Mar 9, 2016
  7. 03/19 read-cache: allow to keep mmap'd memory after readingDavid Turner, Mar 9, 2016
  8. Junio C HamanoMar 9, 2016
  9. David TurnerMar 10, 2016
  10. 04/19 index-helper: new daemon for caching index and related stuffDavid Turner, Mar 9, 2016
  11. Junio C HamanoMar 9, 2016
  12. Junio C HamanoMar 9, 2016
  13. David TurnerMar 10, 2016
  14. Duy NguyenMar 10, 2016
  15. David TurnerMar 10, 2016
  16. Duy NguyenMar 11, 2016
  17. David TurnerMar 10, 2016
  18. Duy NguyenMar 15, 2016
  19. Junio C HamanoMar 15, 2016
  20. Duy NguyenMar 15, 2016
  21. 05/19 trace.c: add GIT_TRACE_INDEX_STATS for index statisticsDavid Turner, Mar 9, 2016
  22. 06/19 index-helper: add --strictDavid Turner, Mar 9, 2016
  23. 07/19 daemonize(): set a flag before exiting the main processDavid Turner, Mar 9, 2016
  24. 08/19 index-helper: add --detachDavid Turner, Mar 9, 2016
  25. 09/19 index-helper: add Windows supportDavid Turner, Mar 9, 2016
  26. Duy NguyenMar 16, 2016
  27. Johannes SchindelinMar 17, 2016
  28. Duy NguyenMar 17, 2016
  29. 10/19 read-cache: add watchman 'WAMA' extensionDavid Turner, Mar 9, 2016
  30. 11/19 Add watchman support to reduce index refresh costDavid Turner, Mar 9, 2016
  31. 12/19 read-cache: allow index-helper to prepare shm before git reads itDavid Turner, Mar 9, 2016
  32. 13/19 index-helper: use watchman to avoid refreshing index with lstat()David Turner, Mar 9, 2016
  33. 14/19 update-index: enable/disable watchman supportDavid Turner, Mar 9, 2016
  34. 15/19 unpack-trees: preserve index extensionsDavid Turner, Mar 9, 2016
  35. 16/19 index-helper: rewrite pidfile after daemonizingDavid Turner, Mar 9, 2016
  36. 17/19 index-helper: process managementDavid Turner, Mar 9, 2016
  37. 18/19 index-helper: autorunDavid Turner, Mar 9, 2016
  38. Duy NguyenMar 15, 2016
  39. Johannes SchindelinMar 15, 2016
  40. Duy NguyenMar 16, 2016
  41. David TurnerMar 16, 2016
  42. Johannes SchindelinMar 16, 2016
  43. Duy NguyenMar 17, 2016
  44. Johannes SchindelinMar 17, 2016
  45. David TurnerMar 17, 2016
  46. Duy NguyenMar 18, 2016
  47. Johannes SchindelinMar 18, 2016
  48. Duy NguyenMar 18, 2016
  49. David TurnerMar 18, 2016
  50. Duy NguyenMar 18, 2016
  51. Johannes SchindelinMar 18, 2016
  52. Duy NguyenMar 18, 2016
  53. Johannes SchindelinMar 18, 2016
  54. 19/19 hack: watchman/untracked cache mashupDavid Turner, Mar 9, 2016
  55. Duy NguyenMar 15, 2016
  56. David TurnerMar 17, 2016
  57. Duy NguyenMar 17, 2016
  58. David TurnerMar 17, 2016
  59. Torsten BögershausenMar 29, 2016
  60. David TurnerMar 29, 2016

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.