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

Re: [PATCH] index-pack: avoid excessive re-reading of pack directory

From
Shawn Pearce <spearce@spearce.org>
Date
Jun 10, 2015, 03:46 UTC
Message-ID
<CAJo=hJv6O66yFC_O_4aeL2JxBOtk2f+k1wGt3VW7dk71Q1ov3A@mail.gmail.com>
In-Reply-To
<20150605122921.GA7586@peff.net>
On Fri, Jun 5, 2015 at 5:29 AM, Jeff King <peff@peff.net> wrote:
Show 14 quoted lines
>
> However, some code paths make a large number of
> has_sha1_file checks which are _not_ expected to return 1.
> The collision test in index-pack.c is such a case. On a
> local system, this can cause a performance slowdown of
> around 5%. But on a system with high-latency system calls
> (like NFS), it can be much worse.
>
> This patch introduces a "quick" flag to has_sha1_file which
> callers can use when they would prefer high performance at
> the cost of false negatives during repacks. There may be
> other code paths that can use this, but the index-pack one
> is the most obviously critical, so we'll start with
> switching that one.
Hilarious. We did this in JGit back in ... uhm before 2009. :)
But its Java. So of course we had to do optimizations.
Show 8 quoted lines
> @@ -3169,6 +3169,8 @@ int has_sha1_file(const unsigned char *sha1)
>                 return 1;
>         if (has_loose_object(sha1))
>                 return 1;
> +       if (flags & HAS_SHA1_QUICK)
> +               return 0;
>         reprepare_packed_git();
>         return find_pack_entry(sha1, &e);

Something else we do is readdir() over the loose objects and store them in a map in memory. That way we avoid stat() calls during that has_loose_object() path. This is apparently a win enough of the time that we always do that when receiving a pack over the wire (client or server).

Previous: Jeff KingNext: Jeff King
Message 27 of 32 in “Troubleshoot clone issue to NFS.”
  1. steve.norman@thomsonreuters.comMay 21, 2015
  2. Christian CouderMay 21, 2015
  3. Duy NguyenMay 21, 2015
  4. Duy NguyenMay 21, 2015
  5. steve.norman@thomsonreuters.comMay 21, 2015
  6. Duy NguyenMay 22, 2015
  7. Jeff KingMay 22, 2015
  8. steve.norman@thomsonreuters.comMay 22, 2015
  9. Duy NguyenMay 22, 2015
  10. Junio C HamanoMay 22, 2015
  11. steve.norman@thomsonreuters.comMay 22, 2015
  12. 0/3 using stat() to avoid re-scanning pack dirJeff King, May 22, 2015
  13. 1/3 stat_validity: handle non-regular filesJeff King, May 22, 2015
  14. Michael HaggertyMay 23, 2015
  15. Jeff KingMay 24, 2015
  16. 2/3 cache.h: move stat_validity definition upJeff King, May 22, 2015
  17. 3/3 prepare_packed_git: use stat_validity to avoid re-reading packsJeff King, May 22, 2015
  18. Duy NguyenMay 23, 2015
  19. Duy NguyenMay 23, 2015
  20. Jeff KingMay 24, 2015
  21. Duy NguyenMay 24, 2015
  22. steve.norman@thomsonreuters.comJun 5, 2015
  23. Jeff KingJun 5, 2015
  24. index-pack: avoid excessive re-reading of pack directoryJeff King, Jun 5, 2015
  25. Jeff KingJun 9, 2015
  26. Jeff KingJun 9, 2015
  27. Shawn PearceJun 10, 2015
  28. Jeff KingJun 10, 2015
  29. Duy NguyenJun 10, 2015
  30. Shawn PearceJun 10, 2015
  31. steve.norman@thomsonreuters.comJun 5, 2015
  32. Jeff KingJun 16, 2015

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.