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

Re: [PATCH v5 3/5] pseudorefs: create and use pseudoref update and delete functions

From
Junio C Hamano <gitster@pobox.com>
Date
Aug 11, 2015, 22:47 UTC
Message-ID
<xmqqbnedo31z.fsf@gitster.dls.corp.google.com>
In-Reply-To
<xmqqh9o5o3r2.fsf@gitster.dls.corp.google.com>
Junio C Hamano <gitster@pobox.com> writes:
Show 9 quoted lines
> David Turner <dturner@twopensource.com> writes:
>
>> On Fri, 2015-07-31 at 16:40 -0700, Stefan Beller wrote:
>>> I am sorry for being late to the review, I looked into coverity today as Duy
>>> bugged me to fix the memory allocation stuff[1]
>>
>> Thanks. Junio, can you pleas substitute the attached patch instead?
>
> No.  The topic is already in 'next', no?
Yes, the topic is already in 'next'.  A follow-up fix would be good.

The patch didn't apply cleanly on top of 74ec19d^ to replace 74ec19d anyway, so I was about to discard it, but after conflict resolution, the interdiff turns out just these two hunks.

-- >8 --
Subject: pseudoref: check return values from read_ref()
From: David Turner <dturner@twopensource.com>
Date: Wed, 15 Jul 2015 18:05:28 -0400

These codepaths attempt to compare the "expected" current value with the actual current value, but did not check if we successfully read the current value before comparison.

Signed-off-by: David Turner <dturner@twopensource.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 * It is likely that we would end up comparing the expected value with
   garbage when the read fails, and the most likely outcome is that
   they do not match and we fail the transaction, which is all fine.
   So in that sense, this is not all that urgent, but it is nice to
   fix it when we know the code is not kosher.
 refs.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/refs.c b/refs.c
index 522b19b..1db3654 100644
--- a/refs.c
+++ b/refs.c
@@ -2868,7 +2868,9 @@ static int write_pseudoref(const char *pseudoref, const unsigned char *sha1,
 
 	if (old_sha1) {
 		unsigned char actual_old_sha1[20];
-		read_ref(pseudoref, actual_old_sha1);
+
+		if (read_ref(pseudoref, actual_old_sha1))
+			die("could not read ref '%s'", pseudoref);
 		if (hashcmp(actual_old_sha1, old_sha1)) {
 			strbuf_addf(err, "Unexpected sha1 when writing %s", pseudoref);
 			rollback_lock_file(&lock);
@@ -2904,7 +2906,8 @@ static int delete_pseudoref(const char *pseudoref, const unsigned char *old_sha1
 					       LOCK_DIE_ON_ERROR);
 		if (fd < 0)
 			die_errno(_("Could not open '%s' for writing"), filename);
-		read_ref(pseudoref, actual_old_sha1);
+		if (read_ref(pseudoref, actual_old_sha1))
+			die("could not read ref '%s'", pseudoref);
 		if (hashcmp(actual_old_sha1, old_sha1)) {
 			warning("Unexpected sha1 when deleting %s", pseudoref);
 			rollback_lock_file(&lock);
Previous: Junio C HamanoNext: David Turner
Message 10 of 13 in “refs: introduce pseudoref and per-worktree ref concepts”
  1. 1/5 refs: introduce pseudoref and per-worktree ref conceptsDavid Turner, Jul 31, 2015
  2. 2/5 refs: add ref_type functionDavid Turner, Jul 31, 2015
  3. Duy NguyenAug 3, 2015
  4. David TurnerAug 3, 2015
  5. David TurnerAug 11, 2015
  6. 3/5 pseudorefs: create and use pseudoref update and delete functionsDavid Turner, Jul 31, 2015
  7. Stefan BellerJul 31, 2015
  8. David TurnerAug 11, 2015
  9. Junio C HamanoAug 11, 2015
  10. Junio C HamanoAug 11, 2015
  11. David TurnerAug 11, 2015
  12. 4/5 bisect: use update_refDavid Turner, Jul 31, 2015
  13. 5/5 sequencer: replace write_cherry_pick_head with update_refDavid Turner, Jul 31, 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.