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

Re: [PATCH] repository: pre-initialize hash algo pointer

From
Junio C Hamano <gitster@pobox.com>
Date
Jan 19, 2018, 19:24 UTC
Message-ID
<xmqqr2qlps7r.fsf@gitster.mtv.corp.google.com>
In-Reply-To
<CAPig+cTOw5NsSmLHYcBEidDzNyiidJ0Dw1dF227KWDL9JrASvw@mail.gmail.com>
Eric Sunshine <sunshine@sunshineco.com> writes:
Show 8 quoted lines
>> I'm still quite mystified as to why this is working on Linux and not
>> macOS, but I can only guess that compilers are just very advanced and
>> have somehow concluded that we would clearly never dereference a NULL
>> pointer, so they picked the only non-NULL value.
>
> Now that we know (due to Duy's excellent detective work[1]) that the
> trigger is files with names differing only in case on case-insensitive
> filesystems, the commit message can be updated appropriately.

Thanks. Let me apply the following and do a 2.16.1, hopefully by the end of day or mid tomorrow at the latest. Test to protect the fix can come as a separate follow-up patch.

-- >8 --
Subject: [PATCH] repository: pre-initialize hash algo pointer

There are various git subcommands (among them, clone) which don't set up the repository (that is, they lack RUN_SETUP or RUN_SETUP_GENTLY) but end up needing to have information about the hash algorithm in use. Because the hash algorithm is part of struct repository and it's only initialized in repository setup, we can end up dereferencing a NULL pointer in some cases if we call one of these subcommands and look up the empty blob or empty tree values.

A "git clone" of a project that has two paths that differ only in case suffers from this if it is run on a case insensitive platform. When the command attempts to check out one of these two paths after checking out the other one, the checkout codepath needs to see if the version that is already on the filesystem (which should not happen if the FS were case sensitive), and it needs to exercise the hashing code.

In the future, we can add a command line option for this or read it from the configuration, but until we're ready to expose that functionality to the user, simply initialize the repository structure to use the current hash algorithm, SHA-1.

Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 repository.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/repository.c b/repository.c
index 998413b8bb..f66fcb1342 100644
--- a/repository.c
+++ b/repository.c
@@ -5,7 +5,7 @@
 
 /* The main repository */
 static struct repository the_repo = {
-	NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &the_index, NULL, 0, 0
+	NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, &the_index, &hash_algos[GIT_HASH_SHA1], 0, 0
 };
 struct repository *the_repository = &the_repo;
 
-- 
2.16.0-204-gc262421c89
Previous: Eric SunshineNext: Eric Sunshine
Message 20 of 33 in “git 2.16.0 segfaults on clone of specific repo”
  1. Александр БулаевJan 18, 2018
  2. Randall S. BeckerJan 18, 2018
  3. Eric SunshineJan 19, 2018
  4. brian m. carlsonJan 19, 2018
  5. Eric SunshineJan 19, 2018
  6. Randall S. BeckerJan 19, 2018
  7. brian m. carlsonJan 19, 2018
  8. Duy NguyenJan 19, 2018
  9. Eric SunshineJan 19, 2018
  10. Duy NguyenJan 19, 2018
  11. Eric SunshineJan 19, 2018
  12. Todd ZullingerJan 19, 2018
  13. Eric SunshineJan 20, 2018
  14. brian m. carlsonJan 19, 2018
  15. Eric SunshineJan 20, 2018
  16. brian m. carlsonJan 20, 2018
  17. Eric SunshineJan 20, 2018
  18. repository: pre-initialize hash algo pointerbrian m. carlson, Jan 19, 2018
  19. Eric SunshineJan 19, 2018
  20. Junio C HamanoJan 19, 2018
  21. Eric SunshineJan 19, 2018
  22. Junio C HamanoJan 19, 2018
  23. brian m. carlsonJan 19, 2018
  24. Junio C HamanoJan 20, 2018
  25. Eric SunshineJan 20, 2018
  26. t: add clone test for files differing only in casebrian m. carlson, Jan 20, 2018
  27. Eric SunshineJan 21, 2018
  28. Junio C HamanoJan 21, 2018
  29. Eric SunshineJan 21, 2018
  30. Eric SunshineJan 21, 2018
  31. brian m. carlsonJan 21, 2018
  32. Duy NguyenJan 21, 2018
  33. Eric SunshineJan 21, 2018

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.