git-annex-shell, remotedaemon, git remote: Fix some memory DOS attacks.

The attacker could just send a very lot of data, with no \n and it would
all be buffered in memory until the kernel killed git-annex or perhaps OOM
killed some other more valuable process.

This is a low impact security hole, only affecting communication between
local git-annex and git-annex-shell on the remote system. (With either
able to be the attacker). Only those with the right ssh key can do it. And,
there are probably lots of ways to construct git repositories that make git
use a lot of memory in various ways, which would have similar impact as
this attack.

The fix in P2P/IO.hs would have been higher impact, if it had made it to a
released version, since it would have allowed DOSing the tor hidden
service without needing to authenticate.

(The LockContent and NotifyChanges instances may not be really
exploitable; since the line is read and ignored, it probably gets read
lazily and does not end up staying buffered in memory.)

1 files changed, 34 insertions, 10 deletions

diff --git a/Utility/SimpleProtocol.hs b/Utility/SimpleProtocol.hs
index 473129218..7ab3c8c77 100644
--- a/Utility/SimpleProtocol.hs
+++ b/Utility/SimpleProtocol.hs
@@ -1,6 +1,6 @@
 {- Simple line-based protocols.
  -
- - Copyright 2013-2014 Joey Hess <id@joeyh.name>
+ - Copyright 2013-2016 Joey Hess <id@joeyh.name>
  -
  - License: BSD-2-clause
  -}
@@ -20,6 +20,7 @@ module Utility.SimpleProtocol (
 	parse2,
 	parse3,
 	dupIoHandles,
+	getProtocolLine,
 ) where
 
 import Data.Char
@@ -48,6 +49,16 @@ class Serializable a where
 	serialize :: a -> String
 	deserialize :: String -> Maybe a
 
+instance Serializable [Char] where
+	serialize = id
+	deserialize = Just
+
+instance Serializable ExitCode where
+	serialize ExitSuccess = "0"
+	serialize (ExitFailure n) = show n
+	deserialize "0" = Just ExitSuccess
+	deserialize s = ExitFailure <$> readish s
+
 {- Parsing the parameters of messages. Using the right parseN ensures
  - that the string is split into exactly the requested number of words,
  - which allows the last parameter of a message to contain arbitrary
@@ -93,12 +104,25 @@ dupIoHandles = do
 	stderr `hDuplicateTo` stdout
 	return (readh, writeh)
 
-instance Serializable [Char] where
-	serialize = id
-	deserialize = Just
-
-instance Serializable ExitCode where
-	serialize ExitSuccess = "0"
-	serialize (ExitFailure n) = show n
-	deserialize "0" = Just ExitSuccess
-	deserialize s = ExitFailure <$> readish s
+{- Reads a line, but to avoid super-long lines eating memory, returns
+ - Nothing if 32 kb have been read without seeing a '\n'
+ -
+ - If there is a '\r' before the '\n', it is removed, to support
+ - systems using "\r\n" at ends of lines 
+ -
+ - This implementation is not super efficient, but as long as the Handle
+ - supports buffering, it avoids reading a character at a time at the
+ - syscall level.
+ -}
+getProtocolLine :: Handle -> IO (Maybe String)
+getProtocolLine h = go (32768 :: Int) []
+  where
+	go 0 _ = return Nothing
+	go n l = do
+		c <- hGetChar h
+		if c == '\n'
+			then return $ Just $ reverse $ 
+				case l of
+					('\r':rest) -> rest
+					_ -> l
+			else go (n-1) (c:l)