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

Re: [PATCH] builtin/clone: fix segfault when using --revision on some servers

From
Adrian Friedli <adrian.friedli@mt.com>
Date
Apr 9, 2026, 13:09 UTC
Message-ID
<20260409130904.1811-1-adrian.friedli@mt.com>
In-Reply-To
<DB8SPR01MB0009F61C462C819CDFF80ED7EA44A@DB8SPR01MB0009.eurprd03.prod.outlook.com>
Show 20 quoted lines
> > > In the bad case the server ignores
> > > `transport_ls_refs_options.ref_prefixes` and in `cmd_clone()` the
> > > linked list `refs` returned by `transport_get_remote_refs()` contains
> > > many items, amongst others "HEAD". `remote_head` returned by
> > > `find_ref_by_name()` is not NULL and `remote_head_points_at` returned
> > > by `guess_remote_head()` is not NULL but its field `peer_ref` is NULL.
> > > Because `remote_head_points_at` is not NULL the guard in
> > > `update_remote_refs()` does not skip the affected code and
> > > `remote_head_points_at->peer_ref->name` is accessed, which causes a
> > > segfault later on.
> >
> > The description makes it sound more like this code is perfectly fine, and the
> > problem is in guess_remote_head() that reads the refs list and includes such a
> > bogus thing with no peer_ref in the result of its guessing.  There are 4 direct
> > callers to guess_remote_head() including this one---wouldn't they also obtain
> > a list with such a ref entry?
> 
> I traced the 3 other callers to guess_remote_head() and none of them has a
> problem if peer_ref is NULL. In get_expanded_map() there is even a condition
> `if (cpy->peer_ref)`, which indicates peer_ref is allowed to be NULL.
The details of the 4 callers to guess_remote_head()

builtin/fetch.c:set_head() builtin/remote.c:get_head_names()

  refs from guess_remote_head() are only used with ref->name
builtin/clone.c:wanted_peer_refs()
  Does not use REMOTE_GUESS_HEAD_ALL flag in guess_remote_head() call. So only one ref is returned.
  calls get_fetch_map() with refs from guess_remote_head().
    may call get_expanded_map()
      safe if peer_ref is NULL, has a guard on the copy of ref `if (cpy->peer_ref)`.
    may call get_remote_ref()
      calls get_remote_ref()
        calls find_ref_by_name_abbrev()
	  only looks at name and next
builtin/clone.c:cmd_clone()
  Does not use REMOTE_GUESS_HEAD_ALL flag in guess_remote_head() call. So only one ref is returned.
  may call write_refspec_config() with refs from guess_remote_head()
    only uses name from refs
  calls update_remote_refs() with refs from guess_remote_head() 
  -> thats the failing one, no other uses
  may call update_head() with refs from guess_remote_head()
    only uses name and old_oid from refs

Kind regards Adrian

Previous: Friedli Adrian LCPF-CH
Message 4 of 4 in “builtin/clone: fix segfault when using --revision on some servers”
  1. builtin/clone: fix segfault when using --revision on some serversAdrian Friedli, Mar 6, 2026
  2. Junio C HamanoMar 6, 2026
  3. Friedli Adrian LCPF-CHMar 12, 2026
  4. Adrian FriedliApr 9, 2026

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.