Avoid using a lot of memory when large objects are present in the git repository

.. and have to be checked to see if they are a pointed to an annexed file.

Cases where such memory use could occur included, but were not limited to:
  - git commit -a of a large unlocked file (in v5 mode)
  - git-annex adjust when a large file was checked into git directly
Generally, any use of catKey was a potential problem.

Fix by using git cat-file --batch-check to check size before catting.
This adds another git batch process, which is included in the CatFileHandle
for simplicity.

There could be performance impact, anywhere catKey is used. Particularly
likely to affect adjusted branch generation speed, and operations on
unlocked files in v6 mode. Hopefully since the --batch-check and
--batch read the same data, disk buffering will avoid most overhead.
Leaving only the overhead of talking to the process over the pipe and
whatever computation --batch-check needs to do.

This commit was sponsored by Bruno BEAUFILS on Patreon.

2 files changed, 16 insertions, 4 deletions

diff --git a/Annex/CatFile.hs b/Annex/CatFile.hs
index 99d301b4b..b1d8fba28 100644
--- a/Annex/CatFile.hs
+++ b/Annex/CatFile.hs
@@ -49,6 +49,11 @@ catObject ref = do
 	h <- catFileHandle
 	liftIO $ Git.CatFile.catObject h ref
 
+catObjectMetaData :: Git.Ref -> Annex (Maybe (Integer, ObjectType))
+catObjectMetaData ref = do
+	h <- catFileHandle
+	liftIO $ Git.CatFile.catObjectMetaData h ref
+
 catTree :: Git.Ref -> Annex [(FilePath, FileMode)]
 catTree ref = do
 	h <- catFileHandle
@@ -89,7 +94,14 @@ catFileStop = do
 
 {- From ref to a symlink or a pointer file, get the key. -}
 catKey :: Ref -> Annex (Maybe Key)
-catKey ref = parseLinkOrPointer <$> catObject ref
+catKey ref = go =<< catObjectMetaData ref
+  where
+	go (Just (sz, _))
+		-- Avoid catting large files, that cannot be symlinks or
+		-- pointer files, which would require buffering their
+		-- content in memory, as well as a lot of IO.
+		| sz <= maxPointerSz = parseLinkOrPointer <$> catObject ref
+	go _ = return Nothing
 
 {- Gets a symlink target. -}
 catSymLinkTarget :: Sha -> Annex String
diff --git a/Annex/Link.hs b/Annex/Link.hs
index af20ae30d..90312a04a 100644
--- a/Annex/Link.hs
+++ b/Annex/Link.hs
@@ -26,7 +26,6 @@ import Annex.HashObject
 import Utility.FileMode
 
 import qualified Data.ByteString.Lazy as L
-import Data.Int
 
 type LinkTarget = String
 
@@ -137,7 +136,8 @@ writePointerFile file k mode = do
  - Only looks at the first line, as pointer files can have subsequent
  - lines. -}
 parseLinkOrPointer :: L.ByteString -> Maybe Key
-parseLinkOrPointer = parseLinkOrPointer' . decodeBS . L.take maxPointerSz
+parseLinkOrPointer = parseLinkOrPointer' 
+	. decodeBS . L.take (fromIntegral maxPointerSz)
   where
 
 {- Want to avoid buffering really big files in git into
@@ -146,7 +146,7 @@ parseLinkOrPointer = parseLinkOrPointer' . decodeBS . L.take maxPointerSz
  - 8192 bytes is plenty for a pointer to a key.
  - Pad some more to allow for any pointer files that might have
  - lines after the key explaining what the file is used for. -}
-maxPointerSz :: Int64
+maxPointerSz :: Integer
 maxPointerSz = 81920
 
 parseLinkOrPointer' :: String -> Maybe Key