From 2694a500787781ec237e39d9ab4e40aa6325572b Mon Sep 17 00:00:00 2001
From: Richard Kiss <him@richardkiss.com>
Date: Tue, 18 Jul 2023 16:11:12 -0700
Subject: [PATCH] Fix ROM deserializer bug.

---
 cl/chialisp_deserialisation.clsp | 72 ++++++++++++++++++++++++++++++++
 src/generator_rom.rs             |  4 +-
 2 files changed, 74 insertions(+), 2 deletions(-)
 create mode 100644 cl/chialisp_deserialisation.clsp

diff --git a/cl/chialisp_deserialisation.clsp b/cl/chialisp_deserialisation.clsp
new file mode 100644
index 000000000..63653d31e
--- /dev/null
+++ b/cl/chialisp_deserialisation.clsp
@@ -0,0 +1,72 @@
+;; this code deserializes blobs of length up to 1<<20 = 1048576 bytes
+
+(mod (input)
+
+  (defconstant MAX_SINGLE_BYTE 0x7F)
+  (defconstant MAX_TWO_BYTE 0xbf)
+  (defconstant MAX_THREE_BYTE 0xdf)
+  (defconstant MAX_FOUR_BYTE 0xef)
+  (defconstant MAX_FIVE_BYTE 0xf7)
+  (defconstant MAX_SIX_BYTE 0xfb)
+  (defconstant CONS_BOX_MARKER 0xFF)
+
+  (defun sexp_from_stream (input_stream)
+    (if (= (substr input_stream 0 1) CONS_BOX_MARKER)
+        (cons_sexp_from_stream (sexp_from_stream (substr input_stream 1)))
+        (atom_from_stream (substr input_stream 1) (substr input_stream 0 1))
+    )
+  )
+
+  (defun cons_sexp_from_stream (left_sexp_with_input)
+    (cons_return (f left_sexp_with_input) (sexp_from_stream (f (r left_sexp_with_input))))
+  )
+
+  (defun cons_return (left_sexp right_sexp_with_input)
+    (list (c left_sexp (f right_sexp_with_input)) (f (r right_sexp_with_input)))
+  )
+
+  (defun atom_from_stream (input_file input_bits)
+    (if (= input_bits (quote 0x80))
+        (list () input_file)
+        (if (>s input_bits MAX_SINGLE_BYTE)
+            (atom_from_stream_part_two (get_bitcount input_bits input_file))
+            (list input_bits input_file)
+        )
+    )
+  )
+
+  ; Note that we reject any serialized atom here with more than 4 bytes of
+  ; encoded length prefix, even though the Rust and Python CLVM interpreters
+  ; and deserializers support more.
+  ; This allows 6 + 7 + 7 = 20 bits = 1048576-byte atoms
+  ; Also note that this does not limit intermediate atom length. Those limits
+  ; are implemented in the clvm interpreters theselves
+  (defun-inline get_bitcount (input_bits input_file)
+    (if (>s input_bits MAX_TWO_BYTE)
+        (if (>s input_bits MAX_THREE_BYTE)
+            ; we don't use the constant here so costs are the same as the
+            ; original (buggy) version of this code which was limited
+            ; to 8192-byte atoms.
+            ; Adding the constant to the constant tree changes locations
+            ; of objects in the tree, which changes costs.
+            (if (>s input_bits (q . 0xef))
+                    (x)
+              ;four byte length prefix
+              (c (concat 0x00 (logand (quote 0x7) input_bits) (substr input_file 0 3)) (substr input_file 3))
+            )
+            ;three byte length prefix
+            (list (concat (logand (quote 0x1f) input_bits) (substr input_file 0 1)) (substr input_file 1))
+        )
+        ;two byte length prefix
+        (list (logand (quote 0x3f) input_bits) input_file)
+    )
+  )
+
+  (defun atom_from_stream_part_two ((size_to_read input_file))
+    (list (substr input_file 0 size_to_read) (substr input_file size_to_read))
+  )
+
+  ; main
+  (f (sexp_from_stream input))
+
+)
diff --git a/src/generator_rom.rs b/src/generator_rom.rs
index 92a545c90..03d8fa791 100644
--- a/src/generator_rom.rs
+++ b/src/generator_rom.rs
@@ -32,9 +32,9 @@ pub const GENERATOR_ROM: [u8; 737] = hex!(
 
 // the CLVM deserializer from:
 // https://github.com/Chia-Network/chia-blockchain/blob/main/chia/wallet/puzzles/chialisp_deserialisation.clsp.hex
-pub const CLVM_DESERIALIZER: [u8; 471] = hex!(
+pub const CLVM_DESERIALIZER: [u8; 498] = hex!(
 "
-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"
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"
 );
 
 // constant from the main chia blockchain: