From 3ca04e355a413975e55adf8b204d6962a9341d32 Mon Sep 17 00:00:00 2001
From: Sebastian Thiel <sebastian.thiel@icloud.com>
Date: Mon, 20 Dec 2021 16:21:39 +0800
Subject: [PATCH] read and validate fanout chunk (#279)

---
 git-chunk/src/lib.rs        | 47 ++++++++++++++++++++++++++++++++++---
 git-pack/src/multi_index.rs | 28 ++++++++++++++++++----
 2 files changed, 68 insertions(+), 7 deletions(-)

diff --git a/git-chunk/src/lib.rs b/git-chunk/src/lib.rs
index 12682866fcf..d5ef451b45d 100644
--- a/git-chunk/src/lib.rs
+++ b/git-chunk/src/lib.rs
@@ -4,12 +4,22 @@
 #![deny(unsafe_code)]
 #![deny(rust_2018_idioms, missing_docs)]
 
+use std::convert::TryInto;
+use std::ops::Range;
+
 /// An identifier to describe the kind of chunk, unique within a chunk file.
 pub type Kind = u32;
 
 /// A special value denoting the end of the chunk file table of contents.
 pub const SENTINEL: Kind = 0;
 
+/// Turn a u64 Range into a usize range safely, to make chunk ranges useful in memory mapped files.
+pub fn into_usize_range(Range { start, end }: Range<file::Offset>) -> Option<Range<usize>> {
+    let start = start.try_into().ok()?;
+    let end = end.try_into().ok()?;
+    Some(Range { start, end })
+}
+
 ///
 pub mod file {
     ///
@@ -18,7 +28,7 @@ pub mod file {
         use std::ops::Range;
 
         ///
-        pub mod not_found {
+        pub mod offset_by_kind {
             use std::fmt::{Display, Formatter};
 
             /// The error returned by [Index::offset_by_kind()][super::Index::offset_by_kind()].
@@ -42,6 +52,26 @@ pub mod file {
             impl std::error::Error for Error {}
         }
 
+        ///
+        pub mod data_by_kind {
+            use quick_error::quick_error;
+            quick_error! {
+                /// The error returned by [Index::data_by_kind()][super::Index::data_by_kind()].
+                #[derive(Debug)]
+                #[allow(missing_docs)]
+                pub enum Error {
+                    NotFound(err: super::offset_by_kind::Error) {
+                        display("The chunk wasn't found in the file index")
+                        from()
+                        source(err)
+                    }
+                    FileTooLarge {
+                        display("The offsets into the file couldn't be represented by usize")
+                    }
+                }
+            }
+        }
+
         /// An entry of a chunk file index
         pub struct Entry {
             /// The kind of the chunk file
@@ -61,11 +91,22 @@ pub mod file {
                 &self,
                 kind: crate::Kind,
                 name: &'static str,
-            ) -> Result<Range<crate::file::Offset>, not_found::Error> {
+            ) -> Result<Range<crate::file::Offset>, offset_by_kind::Error> {
                 self.chunks
                     .iter()
                     .find_map(|c| (c.kind == kind).then(|| c.offset.clone()))
-                    .ok_or_else(|| not_found::Error { kind, name })
+                    .ok_or_else(|| offset_by_kind::Error { kind, name })
+            }
+
+            /// Find a chunk of `kind` and return its data slice based on its offset.
+            pub fn data_by_kind<'a>(
+                &self,
+                data: &'a [u8],
+                kind: crate::Kind,
+                name: &'static str,
+            ) -> Result<&'a [u8], data_by_kind::Error> {
+                let offset = self.offset_by_kind(kind, name)?;
+                Ok(&data[crate::into_usize_range(offset).ok_or_else(|| data_by_kind::Error::FileTooLarge)?])
             }
         }
     }
diff --git a/git-pack/src/multi_index.rs b/git-pack/src/multi_index.rs
index b2c15d05006..02492ff087b 100644
--- a/git-pack/src/multi_index.rs
+++ b/git-pack/src/multi_index.rs
@@ -27,7 +27,7 @@ pub struct File {
     num_chunks: u8,
     /// The amount of pack files contained within
     num_packs: u32,
-    fanout: Range<git_chunk::file::Offset>,
+    fan: [u32; 256],
 }
 
 ///
@@ -49,7 +49,20 @@ mod chunk {
         pub const ID: git_chunk::Kind = 0x504e414d; /* "PNAM" */
     }
     pub mod fanout {
+        use std::convert::TryInto;
+
         pub const ID: git_chunk::Kind = 0x4f494446; /* "OIDF" */
+
+        pub fn from_slice(chunk: &[u8]) -> Option<[u32; 256]> {
+            if chunk.len() != 4 * 256 {
+                return None;
+            }
+            let mut out = [0; 256];
+            for (c, f) in chunk.chunks(4).zip(out.iter_mut()) {
+                *f = u32::from_be_bytes(c.try_into().unwrap());
+            }
+            out.into()
+        }
     }
     pub mod lookup {
         pub const ID: git_chunk::Kind = 0x4f49444c; /* "OIDL" */
@@ -87,7 +100,11 @@ pub mod init {
             #[error(transparent)]
             ChunkFileDecode(#[from] git_chunk::file::decode::Error),
             #[error(transparent)]
-            MissingChunk(#[from] git_chunk::file::index::not_found::Error),
+            MissingChunk(#[from] git_chunk::file::index::offset_by_kind::Error),
+            #[error(transparent)]
+            FileTooLarge(#[from] git_chunk::file::index::data_by_kind::Error),
+            #[error("The multi-pack fan doesn't have the correct size of 256 * 4 bytes")]
+            MultiPackFanSize,
         }
     }
     pub use error::Error;
@@ -152,7 +169,10 @@ pub mod init {
 
             let chunks = git_chunk::file::Index::from_bytes(&data, HEADER_LEN, num_chunks as u32)?;
             let pack_names = chunks.offset_by_kind(chunk::pack_names::ID, "PNAM")?;
-            let fanout = chunks.offset_by_kind(chunk::fanout::ID, "OIDF")?;
+
+            let fan = chunks.data_by_kind(&data, chunk::fanout::ID, "OIDF")?;
+            let fan = chunk::fanout::from_slice(fan).ok_or_else(|| Error::MultiPackFanSize)?;
+
             let lookup = chunks.offset_by_kind(chunk::lookup::ID, "OIDL")?;
             let offsets = chunks.offset_by_kind(chunk::offsets::ID, "OOFF")?;
             let large_offsets = chunks.offset_by_kind(chunk::large_offsets::ID, "LOFF").ok();
@@ -162,7 +182,7 @@ pub mod init {
                 path: path.to_owned(),
                 version,
                 hash_kind,
-                fanout,
+                fan,
                 num_chunks,
                 num_packs,
             })