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

Re: [PATCH 12/26] split-index: fix memory leak in `move_cache_to_base_index()`

From
Rubén Justo <rjusto@gmail.com>
Date
Nov 10, 2024, 21:45 UTC
Message-ID
<df95ded8-2068-416c-a4cc-2dd564f94e6c@gmail.com>
In-Reply-To
<88e727407242248173f1e0f39d7b85b681234f4a.1730901926.git.ps@pks.im>
On Wed, Nov 06, 2024 at 04:10:52PM +0100, Patrick Steinhardt wrote:
Show 28 quoted lines
> In `move_cache_to_base_index()` we move the index cache of the main
> index into the split index, which is used when writing a shared index.
> But we don't release the old split index base in case we already had a
> split index before this operation, which can thus leak memory.
> 
> Plug the leak by releasing the previous base.
> 
> Signed-off-by: Patrick Steinhardt <ps@pks.im>
> ---
>  split-index.c          | 6 +++++-
>  t/t1700-split-index.sh | 1 +
>  2 files changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/split-index.c b/split-index.c
> index 120c8190b18..cfbc773e6cd 100644
> --- a/split-index.c
> +++ b/split-index.c
> @@ -97,7 +97,11 @@ void move_cache_to_base_index(struct index_state *istate)
>  		mem_pool_combine(istate->ce_mem_pool, istate->split_index->base->ce_mem_pool);
>  	}
>  
> -	ALLOC_ARRAY(si->base, 1);
> +	if (si->base)
> +		release_index(si->base);
> +	else
> +		ALLOC_ARRAY(si->base, 1);
> +
>  	index_state_init(si->base, istate->repo);
This change sounded strange to me.  It's like:
   if (si-base) {
   	discard_index(si->base);
   } else {
   	ALLOC_ARRAY(si->base, 1);
   	index_state_init(si->base, istate->repo);
   }
OK.
Show 17 quoted lines
>  	si->base->version = istate->version;
>  	/* zero timestamp disables racy test in ce_write_index() */
> diff --git a/t/t1700-split-index.sh b/t/t1700-split-index.sh
> index ac4a5b2734c..2625ac44889 100755
> --- a/t/t1700-split-index.sh
> +++ b/t/t1700-split-index.sh
> @@ -5,6 +5,7 @@ test_description='split index mode tests'
>  GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME=main
>  export GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME
>  
> +TEST_PASSES_SANITIZE_LEAK=true
>  . ./test-lib.sh
>  
>  # We need total control of index splitting here
> -- 
> 2.47.0.229.g8f8d6eee53.dirty
> 
Previous: Patrick SteinhardtNext: Patrick Steinhardt
Message 17 of 39 in “Memory leak fixes (pt.10, final)”
  1. 00/26 Memory leak fixes (pt.10, final)Patrick Steinhardt, Nov 6, 2024
  2. 01/26 builtin/blame: fix leaking blame entries with `--incremental`Patrick Steinhardt, Nov 6, 2024
  3. 02/26 bisect: fix leaking good/bad terms when reading multipe timesPatrick Steinhardt, Nov 6, 2024
  4. 03/26 bisect: fix leaking string in `handle_bad_merge_base()`Patrick Steinhardt, Nov 6, 2024
  5. 04/26 bisect: fix leaking `current_bad_oid`Patrick Steinhardt, Nov 6, 2024
  6. 05/26 bisect: fix multiple leaks in `bisect_next_all()`Patrick Steinhardt, Nov 6, 2024
  7. 06/26 bisect: fix leaking commit list items in `check_merge_base()`Patrick Steinhardt, Nov 6, 2024
  8. 07/26 bisect: fix various cases where we leak commit list itemsPatrick Steinhardt, Nov 6, 2024
  9. 08/26 line-log: fix leak when rewriting commit parentsPatrick Steinhardt, Nov 6, 2024
  10. 09/26 strvec: introduce new `strvec_splice()` functionPatrick Steinhardt, Nov 6, 2024
  11. Rubén JustoNov 10, 2024
  12. Patrick SteinhardtNov 11, 2024
  13. 10/26 git: refactor alias handling to use a `struct strvec`Patrick Steinhardt, Nov 6, 2024
  14. Rubén JustoNov 10, 2024
  15. 11/26 git: refactor builtin handling to use a `struct strvec`Patrick Steinhardt, Nov 6, 2024
  16. 12/26 split-index: fix memory leak in `move_cache_to_base_index()`Patrick Steinhardt, Nov 6, 2024
  17. Rubén JustoNov 10, 2024
  18. 13/26 builtin/sparse-checkout: fix leaking sanitized patternsPatrick Steinhardt, Nov 6, 2024
  19. 14/26 help: refactor to not use globals for reading configPatrick Steinhardt, Nov 6, 2024
  20. 15/26 help: fix leaking `struct cmdnames`Patrick Steinhardt, Nov 6, 2024
  21. Rubén JustoNov 10, 2024
  22. Patrick SteinhardtNov 11, 2024
  23. 16/26 help: fix leaking return value from `help_unknown_cmd()`Patrick Steinhardt, Nov 6, 2024
  24. 17/26 builtin/help: fix leaks in `check_git_cmd()`Patrick Steinhardt, Nov 6, 2024
  25. 18/26 builtin/init-db: fix leaking directory pathsPatrick Steinhardt, Nov 6, 2024
  26. Rubén JustoNov 10, 2024
  27. 19/26 builtin/branch: fix leaking sorting optionsPatrick Steinhardt, Nov 6, 2024
  28. Rubén JustoNov 10, 2024
  29. 20/26 t/helper: fix leaking commit graph in "read-graph" subcommandPatrick Steinhardt, Nov 6, 2024
  30. 21/26 git-compat-util: drop `UNLEAK()` annotationPatrick Steinhardt, Nov 6, 2024
  31. Rubén JustoNov 10, 2024
  32. Patrick SteinhardtNov 11, 2024
  33. 22/26 t5601: work around leak sanitizer issuePatrick Steinhardt, Nov 6, 2024
  34. 23/26 t: mark some tests as leak freePatrick Steinhardt, Nov 6, 2024
  35. 24/26 t: remove unneeded !SANITIZE_LEAK prerequisitesPatrick Steinhardt, Nov 6, 2024
  36. 25/26 test-lib: unconditionally enable leak checkingPatrick Steinhardt, Nov 6, 2024
  37. 26/26 t: remove TEST_PASSES_SANITIZE_LEAK annotationsPatrick Steinhardt, Nov 6, 2024
  38. Rubén JustoNov 10, 2024
  39. Patrick SteinhardtNov 11, 2024

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.