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

Re: [PATCH 25/27] attr: store attribute stacks in hashmap

From
Junio C Hamano <gitster@pobox.com>
Date
Jan 13, 2017, 21:20 UTC
Message-ID
<xmqqshomejwt.fsf@gitster.mtv.corp.google.com>
In-Reply-To
<20170112235354.153403-26-bmwill@google.com>
Brandon Williams <bmwill@google.com> writes:
Show 8 quoted lines
> The last big hurdle towards a thread-safe API for the attribute system
> is the reliance on a global attribute stack that is modified during each
> call into the attribute system.
>
> This patch removes this global stack and instead a stack is retrieved or
> constructed locally.  Since each of these stacks is only used as a
> read-only structure once constructed, they can be stored in a hashmap
> and shared between threads.
Very good.

The reason why the original code used a stack was because it wanted to keep only the info read from releavant files in-core, discarding ones from files no-longer relevant (because the traversal switched to another subdirectory of the same parent directory), to avoid the memory consumption grow unbounded. It probably was a premature "optimization" that we can do without, so keeping everything we have read so far in a hashmap (which is my understanding of what is going on in this patch) is probably OK.

I suspect that this hashmap may eventually need to become per attr_check if we want to follow through the optimization envisioned by patch 15/27.

Inside fill(), path_matches() is called for the number of match_attr in the entire attribute stack but it is wasteful to check if the path matches with the a.u.pat if none of the a.state[] entries talk about attributes and macros that are eventually get used by the caller of check_attr(). By introducing a wrapping structure, 15/27 wanted to make sure that we have a place to store a "reduced" attribute stack that is kept per attr_check that has only entries from the files that talk about the attributes the particular attr_check wants to learn about.

I need to think about this a bit more, but I do not offhand think that it makes future such enhancement to make it per-check harder to move from a global stack to a global hashmap, i.e. the above is not an objection to this step.

Show 9 quoted lines
> One caveat with storing and sharing the stack frames like this is that
> the info stack needs to be treated separately from the rest of the
> attribute stack.  This is because each stack frame holds a pointer to
> the stack that comes before it and if it was placed on top of the rest
> of the attribute stack then this pointer would be different for each
> attribute stack and wouldn't be able to be shared between threads.  In
> order to allow for sharing the info stack frame it needs to be its own
> isolated frame and can simply be processed first to have the same affect
> of being at the top of the stack.
Good.
Thanks.
Previous: Brandon WilliamsNext: Brandon Williams
Message 14 of 111 in “Revamp the attribute system; another round”
  1. 00/27 Revamp the attribute system; another roundBrandon Williams, Jan 12, 2017
  2. 20/27 attr: change validity check for attribute names to use positive logicBrandon Williams, Jan 12, 2017
  3. 19/27 attr: pass struct attr_check to collect_some_attrsBrandon Williams, Jan 12, 2017
  4. 18/27 attr: retire git_check_attrs() APIBrandon Williams, Jan 12, 2017
  5. 17/27 attr: convert git_check_attrs() callers to use the new APIBrandon Williams, Jan 12, 2017
  6. 15/27 attr: (re)introduce git_check_attr() and struct attr_checkBrandon Williams, Jan 12, 2017
  7. 16/27 attr: convert git_all_attrs() to use "struct attr_check"Brandon Williams, Jan 12, 2017
  8. 23/27 attr: remove maybe-real, maybe-macro from git_attrBrandon Williams, Jan 12, 2017
  9. 24/27 attr: tighten const correctness with git_attr and match_attrBrandon Williams, Jan 12, 2017
  10. 21/27 attr: use hashmap for attribute dictionaryBrandon Williams, Jan 12, 2017
  11. Stefan BellerJan 18, 2017
  12. Brandon WilliamsJan 18, 2017
  13. 25/27 attr: store attribute stacks in hashmapBrandon Williams, Jan 12, 2017
  14. Junio C HamanoJan 13, 2017
  15. Brandon WilliamsJan 18, 2017
  16. Brandon WilliamsJan 23, 2017
  17. Stefan BellerJan 18, 2017
  18. Brandon WilliamsJan 18, 2017
  19. Stefan BellerJan 18, 2017
  20. 22/27 attr: eliminate global check_all_attr arrayBrandon Williams, Jan 12, 2017
  21. 27/27 attr: reformat git_attr_set_direction() functionBrandon Williams, Jan 12, 2017
  22. 26/27 attr: push the bare repo check into read_attr()Brandon Williams, Jan 12, 2017
  23. 13/27 attr.c: outline the future plans by heavily commentingBrandon Williams, Jan 12, 2017
  24. 12/27 Documentation: fix a typoBrandon Williams, Jan 12, 2017
  25. 01/27 commit.c: use strchrnul() to scan for one lineBrandon Williams, Jan 12, 2017
  26. 02/27 attr.c: use strchrnul() to scan for one lineBrandon Williams, Jan 12, 2017
  27. 11/27 attr.c: add push_stack() helperBrandon Williams, Jan 12, 2017
  28. 08/27 attr.c: tighten constness around "git_attr" structureBrandon Williams, Jan 12, 2017
  29. 14/27 attr: rename function and struct related to checking attributesBrandon Williams, Jan 12, 2017
  30. 06/27 attr.c: mark where #if DEBUG ends more clearlyBrandon Williams, Jan 12, 2017
  31. 07/27 attr.c: simplify macroexpand_one()Brandon Williams, Jan 12, 2017
  32. 10/27 attr: support quoting pathname patterns in C styleBrandon Williams, Jan 12, 2017
  33. 09/27 attr.c: plug small leak in parse_attr_line()Brandon Williams, Jan 12, 2017
  34. 04/27 attr.c: explain the lack of attr-name syntax check in parse_attr()Brandon Williams, Jan 12, 2017
  35. 05/27 attr.c: complete a sentence in a commentBrandon Williams, Jan 12, 2017
  36. 03/27 attr.c: update a stale comment on "struct match_attr"Brandon Williams, Jan 12, 2017
  37. Junio C HamanoJan 15, 2017
  38. Jeff KingJan 16, 2017
  39. 00/27 Revamp the attribute system; another roundBrandon Williams, Jan 23, 2017
  40. 01/27 commit.c: use strchrnul() to scan for one lineBrandon Williams, Jan 23, 2017
  41. 02/27 attr.c: use strchrnul() to scan for one lineBrandon Williams, Jan 23, 2017
  42. 03/27 attr.c: update a stale comment on "struct match_attr"Brandon Williams, Jan 23, 2017
  43. 04/27 attr.c: explain the lack of attr-name syntax check in parse_attr()Brandon Williams, Jan 23, 2017
  44. 05/27 attr.c: complete a sentence in a commentBrandon Williams, Jan 23, 2017
  45. 07/27 attr.c: simplify macroexpand_one()Brandon Williams, Jan 23, 2017
  46. 10/27 attr: support quoting pathname patterns in C styleBrandon Williams, Jan 23, 2017
  47. 12/27 Documentation: fix a typoBrandon Williams, Jan 23, 2017
  48. 13/27 attr.c: outline the future plans by heavily commentingBrandon Williams, Jan 23, 2017
  49. 09/27 attr.c: plug small leak in parse_attr_line()Brandon Williams, Jan 23, 2017
  50. 11/27 attr.c: add push_stack() helperBrandon Williams, Jan 23, 2017
  51. 14/27 attr: rename function and struct related to checking attributesBrandon Williams, Jan 23, 2017
  52. 20/27 attr: change validity check for attribute names to use positive logicBrandon Williams, Jan 23, 2017
  53. 21/27 attr: use hashmap for attribute dictionaryBrandon Williams, Jan 23, 2017
  54. 22/27 attr: eliminate global check_all_attr arrayBrandon Williams, Jan 23, 2017
  55. Junio C HamanoJan 23, 2017
  56. 16/27 attr: convert git_all_attrs() to use "struct attr_check"Brandon Williams, Jan 23, 2017
  57. 23/27 attr: remove maybe-real, maybe-macro from git_attrBrandon Williams, Jan 23, 2017
  58. 24/27 attr: tighten const correctness with git_attr and match_attrBrandon Williams, Jan 23, 2017
  59. 17/27 attr: convert git_check_attrs() callers to use the new APIBrandon Williams, Jan 23, 2017
  60. 18/27 attr: retire git_check_attrs() APIBrandon Williams, Jan 23, 2017
  61. 15/27 attr: (re)introduce git_check_attr() and struct attr_checkBrandon Williams, Jan 23, 2017
  62. 06/27 attr.c: mark where #if DEBUG ends more clearlyBrandon Williams, Jan 23, 2017
  63. 08/27 attr.c: tighten constness around "git_attr" structureBrandon Williams, Jan 23, 2017
  64. 19/27 attr: pass struct attr_check to collect_some_attrsBrandon Williams, Jan 23, 2017
  65. 26/27 attr: push the bare repo check into read_attr()Brandon Williams, Jan 23, 2017
  66. 27/27 attr: reformat git_attr_set_direction() functionBrandon Williams, Jan 23, 2017
  67. 25/27 attr: store attribute stack in attr_check structureBrandon Williams, Jan 23, 2017
  68. Junio C HamanoJan 23, 2017
  69. Brandon WilliamsJan 23, 2017
  70. Brandon WilliamsJan 24, 2017
  71. Junio C HamanoJan 24, 2017
  72. Brandon WilliamsJan 25, 2017
  73. Stefan BellerJan 25, 2017
  74. Junio C HamanoJan 25, 2017
  75. Brandon WilliamsJan 25, 2017
  76. Brandon WilliamsJan 25, 2017
  77. 00/27 Revamp the attribute system; another roundBrandon Williams, Jan 28, 2017
  78. 01/27 commit.c: use strchrnul() to scan for one lineBrandon Williams, Jan 28, 2017
  79. 02/27 attr.c: use strchrnul() to scan for one lineBrandon Williams, Jan 28, 2017
  80. 05/27 attr.c: complete a sentence in a commentBrandon Williams, Jan 28, 2017
  81. 03/27 attr.c: update a stale comment on "struct match_attr"Brandon Williams, Jan 28, 2017
  82. 07/27 attr.c: simplify macroexpand_one()Brandon Williams, Jan 28, 2017
  83. 06/27 attr.c: mark where #if DEBUG ends more clearlyBrandon Williams, Jan 28, 2017
  84. 08/27 attr.c: tighten constness around "git_attr" structureBrandon Williams, Jan 28, 2017
  85. 04/27 attr.c: explain the lack of attr-name syntax check in parse_attr()Brandon Williams, Jan 28, 2017
  86. 09/27 attr.c: plug small leak in parse_attr_line()Brandon Williams, Jan 28, 2017
  87. 10/27 attr: support quoting pathname patterns in C styleBrandon Williams, Jan 28, 2017
  88. 11/27 attr.c: add push_stack() helperBrandon Williams, Jan 28, 2017
  89. 12/27 Documentation: fix a typoBrandon Williams, Jan 28, 2017
  90. 15/27 attr: (re)introduce git_check_attr() and struct attr_checkBrandon Williams, Jan 28, 2017
  91. 15/27 attr: (re)introduce git_check_attr() and struct attr_checkBrandon Williams, Jan 30, 2017
  92. 13/27 attr.c: outline the future plans by heavily commentingBrandon Williams, Jan 28, 2017
  93. 16/27 attr: convert git_all_attrs() to use "struct attr_check"Brandon Williams, Jan 28, 2017
  94. Stefan BellerJan 28, 2017
  95. Brandon WilliamsJan 29, 2017
  96. 16/27 attr: convert git_all_attrs() to use "struct attr_check"Brandon Williams, Jan 30, 2017
  97. 14/27 attr: rename function and struct related to checking attributesBrandon Williams, Jan 28, 2017
  98. 18/27 attr: retire git_check_attrs() APIBrandon Williams, Jan 28, 2017
  99. 19/27 attr: pass struct attr_check to collect_some_attrsBrandon Williams, Jan 28, 2017
  100. 17/27 attr: convert git_check_attrs() callers to use the new APIBrandon Williams, Jan 28, 2017
  101. 22/27 attr: eliminate global check_all_attr arrayBrandon Williams, Jan 28, 2017
  102. 21/27 attr: use hashmap for attribute dictionaryBrandon Williams, Jan 28, 2017
  103. 26/27 attr: push the bare repo check into read_attr()Brandon Williams, Jan 28, 2017
  104. 25/27 attr: store attribute stack in attr_check structureBrandon Williams, Jan 28, 2017
  105. 20/27 attr: change validity check for attribute names to use positive logicBrandon Williams, Jan 28, 2017
  106. 24/27 attr: tighten const correctness with git_attr and match_attrBrandon Williams, Jan 28, 2017
  107. 27/27 attr: reformat git_attr_set_direction() functionBrandon Williams, Jan 28, 2017
  108. 23/27 attr: remove maybe-real, maybe-macro from git_attrBrandon Williams, Jan 28, 2017
  109. Junio C HamanoFeb 2, 2017
  110. Brandon WilliamsFeb 9, 2017
  111. Junio C HamanoFeb 9, 2017

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.