close the git add race

There's a race adding a new file to the annex: The file is moved to the
annex and replaced with a symlink, and then we git add the symlink. If
someone comes along in the meantime and replaces the symlink with
something else, such as a new large file, we add that instead. Which could
be bad..

This race is fixed by avoiding using git add, instead the symlink is
directly staged into the index.

It would be nice to make `git annex add` use this same technique.
I have not done so yet because it currently runs git update-index once per
file, which would slow does `git annex add`. A future enhancement would be
to extend the Git.Queue to include the ability to run update-index with
a list of Streamers.
This commit is contained in:
Joey Hess 2012-06-06 14:29:10 -04:00
parent 91db540769
commit b819f644ad
3 changed files with 47 additions and 20 deletions

View file

@ -94,8 +94,9 @@ undo file key e = do
src <- inRepo $ gitAnnexLocation key
liftIO $ moveFile src file
{- Creates the symlink to the annexed content. -}
link :: FilePath -> Key -> Bool -> Annex ()
{- Creates the symlink to the annexed content, and also returns the link's
- text. -}
link :: FilePath -> Key -> Bool -> Annex FilePath
link file key hascontent = handle (undo file key) $ do
l <- calcGitLink file key
liftIO $ createSymbolicLink l file
@ -109,11 +110,13 @@ link file key hascontent = handle (undo file key) $ do
mtime <- modificationTime <$> getFileStatus file
touch file (TimeSpec mtime) False
return l
{- Note: Several other commands call this, and expect it to
- create the symlink and add it. -}
cleanup :: FilePath -> Key -> Bool -> CommandCleanup
cleanup file key hascontent = do
link file key hascontent
_ <- link file key hascontent
params <- ifM (Annex.getState Annex.force)
( return [Param "-f"]
, return []