urosm
/
ikiwiki
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

* Avoid a race in the git rcs_commit function, by not assuming HEAD will 

stay the same for the duration of the function.
master
Joey Hess 2007-10-31 17:17:03 -04:00
parent 2a6e353c20
commit c5d9c0d6b6
3 changed files with 51 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
IkiWiki/Rcs/git.pm
View File

@ -165,7 +165,10 @@ sub _parse_diff_tree ($@) { #{{{
# Identification lines for the commit.
while (my $line = shift @{ $dt_ref }) {
# Regexps are semi-stolen from gitweb.cgi.
if ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
if ($line =~ m/^commit ([0-9a-fA-F]{40})$/) {
$ci{'commit'} = $1;
}
elsif ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
$ci{'tree'} = $1;
}
elsif ($line =~ m/^parent ([0-9a-fA-F]{40})$/) {
@ -431,12 +434,9 @@ sub rcs_notify () { #{{{
#
# Here, we rely on a simple fact: we can extract all parts of the
# notification content by parsing the "HEAD" commit (which also
# triggers a refresh of IkiWiki pages) and we can obtain the diff
# by comparing HEAD and HEAD^ (the previous commit).
# triggers a refresh of IkiWiki pages).
my $sha1 = 'HEAD'; # the commit which triggers this action
my $ci = git_commit_info($sha1);
my $ci = git_commit_info('HEAD');
return if !defined $ci;
my @changed_pages = map { $_->{'file'} } @{ $ci->{'details'} };
@ -451,6 +451,8 @@ sub rcs_notify () { #{{{
$message = join "\n", @{ $ci->{'comment'} };
}
my $sha1 = $ci->{'commit'};
require IkiWiki::UserInfo;
send_commit_mails(
sub {

4
debian/changelog vendored
View File

@ -5,8 +5,10 @@ ikiwiki (2.12) UNRELEASED; urgency=low
page name to be expired and reused for several distinct guids. When this
happened, the expiry code counted each past guid that had used that page
name as a currently existing page, and thus expired too many pages.
* Avoid a race in the git rcs_commit function, by not assuming HEAD will
stay the same for the duration of the function.
-- Joey Hess <joeyh@debian.org> Tue, 30 Oct 2007 22:47:36 -0400
-- Joey Hess <joeyh@debian.org> Wed, 31 Oct 2007 17:11:12 -0400
ikiwiki (2.11) unstable; urgency=low

40
doc/bugs/git_mail_notification_race.mdwn 100644
View File

@ -0,0 +1,40 @@
I was suprised to receive two mails from ikiwiki about one web edit:
1 F Oct 30 To joey+ikiwiki update of ikiwiki's plugins/contrib/gallery.mdwn by http://arpitjain11.myopenid.com/
1 F Oct 30 To joey+ikiwiki update of ikiwiki's plugins/contrib/gallery.mdwn by http://arpitjain11.myopenid.com/
The first of these had the correct diff for the changes made by the web
edit (00259020061577316895370ee04cf00b634db98a).
But the second had a diff for modifications I made to ikiwiki code
around the same time (2a6e353c205a6c2c8b8e2eaf85fe9c585c1af0cd).
I'm now fairly sure a race is involved. Note that the name of the modified
page is correct, while the diff was not. The git rcs_commit code first
gets the list of modified page(s) and then gets the diff, and apparently my
other commit happened in the meantime, so HEAD changed.
Seems that the rcs_notify for git assumes that HEAD is the commit
that triggered the ikiwiki run, and that it won't change while the function is
running. Both assumptions are bad. Git does no locking and another commit
can come in at any time.
Notice that the equivilant code for subversion is careful to look at $REV.
git's post-update hook doesn't have an equivilant of $REV. Its post-receive hook
does, but I'm not sure if that's an appropriate hook for ikiwiki to be using
for this. Switching existing wikis to use a different hook would be tricky..
I've avoided part of the race; now the code does:
1. Get commit info for HEAD commit.
2. Extract file list from that.
3. Extract sha1 of commit from it too.
4. git diff sha1^ sha1
Still seems like there can be a race if another commit comes in before step #1.
In this case, two post-receive hooks could run at the same time, and both
see the same sha1 as HEAD, so two diffs might be sent for second commit and no
diff for the first commit.
Ikiwiki's own locking prevents this from happenning if both commits are web
edits. At least one of the two commits has to be a non-web commit.