From: John Cai <johncai86@gmail.com>
To: Junio C Hamano <gitster@pobox.com>
Cc: "John Cai via GitGitGadget" <gitgitgadget@gmail.com>,
git@vger.kernel.org, "Phillip Wood" <phillip.wood123@gmail.com>,
"Kristoffer Haugsbakk" <code@khaugsbakk.name>,
"Jeff King" <peff@peff.net>, "Patrick Steinhardt" <ps@pks.im>,
"Jean-Noël Avila" <avila.jn@gmail.com>
Subject: Re: [PATCH v2 1/3] refs: keep track of unresolved reference value in iterator
Date: Tue, 09 Apr 2024 16:29:48 -0400 [thread overview]
Message-ID: <DDF183A6-2840-4225-B459-8D9B4287D98F@gmail.com> (raw)
In-Reply-To: <xmqq34rv791a.fsf@gitster.g>
Hi Junio
On 8 Apr 2024, at 19:02, Junio C Hamano wrote:
>> diff --git a/refs.h b/refs.h
>> index 298caf6c618..2e740c692ac 100644
>> --- a/refs.h
>> +++ b/refs.h
>> @@ -71,9 +71,10 @@ struct pack_refs_opts {
>> struct ref_exclusions *exclusions;
>> struct string_list *includes;
>> };
>> -
>> const char *refs_resolve_ref_unsafe(struct ref_store *refs,
>> +
>> const char *refname,
>> + char **referent,
>> int resolve_flags,
>> struct object_id *oid,
>> int *flags);
>
> If referent is meant to be an out-parameter, it should sit next to
> oid that is also an out-parameter. And as a late-comer sibling, it
> should sit after its elder brother.
>
> Also, I do not see the reason for the shuffling of blank lines.
> Shouldn't it be the other way around, even? After an unrelated
> definition of "struct pack_refs_opts", there is (and should be)
> a blank line, then the first line of the declaration of function.
>
> Perhaps some fat-fingering.
This was indeed a case of fat-fingering.
>
>> @@ -1928,6 +1928,7 @@ int refs_read_symbolic_ref(struct ref_store *ref_store, const char *refname,
>>
>> const char *refs_resolve_ref_unsafe(struct ref_store *refs,
>> const char *refname,
>> + char **referent,
>> int resolve_flags,
>> struct object_id *oid,
>> int *flags)
>> @@ -1989,6 +1990,8 @@ const char *refs_resolve_ref_unsafe(struct ref_store *refs,
>> }
>>
>> *flags |= read_flags;
>> + if (referent && read_flags & REF_ISSYMREF && sb_refname.len > 0)
>> + *referent = sb_refname.buf;
>
> Is this safe? After this assignment, which "return" in this loop
> are you expecting to return from this function? If you fail to
> return from the function during this iteration, you'll clobber the
> same strbuf with the next refs_read_raw_ref(), but I do not see how
> you are ensuring that you'll return from the function without such
> corruption happening.
>
> This assignment happens only when read_flags has REF_ISSYMREF set,
> so the next "if it is not, then return refname" would not even
> trigger. If RESOLVE_REF_NO_RECURSE bit is on in resolve_flags,
> then we'd return without further dereferencing, but if that is the
> only safe exit from the function, shouldn't you be guarding the
> function with something like
>
> if (referent && !(resolve_flags & RESOLVE_REF_NO_RECURSE))
> BUG("recursive dereference can will clobber *referent");
>
> to protect its callers from their own mistakes?
>
> Another return before we start the next iteration of the loop and
> clobber sb_refname with another call to refs_read_raw_ref() is the
> error return codepath at the end of the loop, but that is a totally
> uninteresting case.
>
> Or am I totally confused?
Yeah good point. This probably not a good idea. In fact, perhaps adding another
argument to this function is unnecessary. We already have refs_read_symbolic_ref
and we can just make a separate call to that once we know that a ref is a
symbolic reference. Though a separate call is less efficient, I'm not sure it's
worth adding this parameter.
thanks
John
next prev parent reply other threads:[~2024-04-09 20:29 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-04 22:51 [PATCH] show-ref: add --unresolved option John Cai via GitGitGadget
2024-03-04 23:23 ` Junio C Hamano
2024-03-05 20:56 ` John Cai
2024-03-05 21:29 ` Junio C Hamano
2024-03-05 15:30 ` Phillip Wood
2024-03-05 17:01 ` Kristoffer Haugsbakk
2024-03-06 0:33 ` Jeff King
2024-03-06 2:19 ` Junio C Hamano
2024-03-06 0:41 ` Jeff King
2024-03-06 7:31 ` Patrick Steinhardt
2024-03-06 7:51 ` Jeff King
2024-03-06 16:48 ` Junio C Hamano
2024-03-06 9:36 ` Jean-Noël Avila
2024-04-08 17:38 ` [PATCH v2 0/3] show-ref: add --symbolic-name option John Cai via GitGitGadget
2024-04-08 17:38 ` [PATCH v2 1/3] refs: keep track of unresolved reference value in iterator John Cai via GitGitGadget
2024-04-08 23:02 ` Junio C Hamano
2024-04-09 20:29 ` John Cai [this message]
2024-04-10 6:52 ` Patrick Steinhardt
2024-04-10 15:26 ` Junio C Hamano
2024-04-11 9:11 ` Patrick Steinhardt
2024-04-08 17:38 ` [PATCH v2 2/3] refs: add referent to each_repo_ref_fn John Cai via GitGitGadget
2024-04-08 17:38 ` [PATCH v2 3/3] show-ref: add --symbolic-name option John Cai via GitGitGadget
2024-04-09 15:25 ` Phillip Wood
2024-04-11 19:57 ` John Cai
2024-04-12 9:37 ` phillip.wood123
2024-04-10 6:53 ` Patrick Steinhardt
2024-04-10 15:27 ` Junio C Hamano
2024-04-12 15:23 ` John Cai
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DDF183A6-2840-4225-B459-8D9B4287D98F@gmail.com \
--to=johncai86@gmail.com \
--cc=avila.jn@gmail.com \
--cc=code@khaugsbakk.name \
--cc=git@vger.kernel.org \
--cc=gitgitgadget@gmail.com \
--cc=gitster@pobox.com \
--cc=peff@peff.net \
--cc=phillip.wood123@gmail.com \
--cc=ps@pks.im \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).