From: Junio C Hamano <gitster@pobox.com>
To: Wesley Smith <wsmith@greatergiving.com>
Cc: "git\@vger.kernel.org" <git@vger.kernel.org>
Subject: Re: Is finalize_object_file in sha1_file.c handling errno from "rename" correctly?
Date: Fri, 15 Sep 2017 15:50:35 +0900 [thread overview]
Message-ID: <xmqq4ls4xxok.fsf@gitster.mtv.corp.google.com> (raw)
In-Reply-To: <c9b3bc17110048f0b7943704cfbd8f68@APEXC04.auctionpay.com> (Wesley Smith's message of "Fri, 15 Sep 2017 06:18:13 +0000")
Wesley Smith <wsmith@greatergiving.com> writes:
> 1) This bug is triggered because "git fetch" is causing a pack
> file to be written when that same pack file already exists. It
> seems like this is harmless and shouldn't cause a problem. Is
> that correct?
The final name of the packfile is derived from the entire contents
of the packfile; it should be harmless when we attempt to rename a
new file, which has exactly the same contents as an existing file,
to the existing file and see a failure out of that attempt.
> 2) It seems that finalize_object_file is not accounting for the
> fact that "link" will return EEXIST if the destination file
> already exists but is not writeable, whereas "rename" will return
> EACCESS in this case. Is that correct? If so, should
> finalize_object_file be fixed to account for this? Perhaps it
> should check if the newfile exists before calling rename. Or,
> should the Windows mingw_rename function be modified to return
> EEXIST in this case, even though that's not the standard errno for
> that situation?
The codepath that is triggered by OBJECT_CREATION_USES_RENAMES ought
to behave correctly even on non-Windows platforms, so bending the
error code of rename() only on Windows to fit the existing error
handling would not be a smart thing to do. Rather, the rename()
emulation should leave a correct errno and the caller should be
updated to be aware of that error that is not EEXIST, which it
currently knows about.
Thanks for spotting a problem and digging to its cause.
This is a #leftoverbits tangent, and should be done separately from
your "OBJECT_CREATION_USES_RENAMES is broken" topic, but I think it
is a bug to use finalize_object_file() directly to "finalize"
anything but an individual loose object file in the first place.
We should create a new shared helper that does what the function
currently does, make finalize_object_file() call that new shared
helper, and make sure finalize_object_file() is called only on a
newly created loose object file. The codepath that creates a new
packfile and other things and moves them to the final name should
not call finalize_object_file() but the new shared helper instead.
That way, we could later implement the "collision? check" alluded by
the in-code comment in finailize_object_file(), and we won't have to
worry about affecting callers other than the one that creates a
loose object file with such an enhancement.
next prev parent reply other threads:[~2017-09-15 6:50 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-09-15 6:18 Is finalize_object_file in sha1_file.c handling errno from "rename" correctly? Wesley Smith
2017-09-15 6:50 ` Junio C Hamano [this message]
2017-09-15 18:18 ` Wesley Smith
2017-09-29 23:21 ` Johannes Schindelin
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
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=xmqq4ls4xxok.fsf@gitster.mtv.corp.google.com \
--to=gitster@pobox.com \
--cc=git@vger.kernel.org \
--cc=wsmith@greatergiving.com \
/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.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
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).