DirectIO is working! Also a bit more cleanup.

src/io.rs

@@ -1,7 +1,6 @@
-use std::ffi::CString;
 use std::fs::{File, OpenOptions};
 use std::io::{self, Seek, SeekFrom};
-use std::os::fd::{AsRawFd, FromRawFd, IntoRawFd};
+use std::os::unix::fs::OpenOptionsExt;
 
 use crate::cli::CONFIG;
 
@@ -21,22 +20,12 @@ pub fn get_stream_length<S: Seek>(stream: &mut S) -> io::Result<u64> {
     len
 }
 
-/*
- * IO Error Poisoning:
- *
- *
- * Attempt 1:
- *
- * Wrap C calls w/ `ffi`. Most importantly, execute the `close()` through C.
- *
- * I cannot seem to execute `close()`... one hopes that reassigning will drop
- * calling `close()` on the old fd, but I don't know.
- */
-
 pub fn load_input() -> anyhow::Result<File> {
-    let path = CString::new(CONFIG.input.to_str().unwrap().to_owned())?;
-    let flags = libc::O_RDONLY | libc::O_DIRECT;
-    Ok(unsafe { File::from_raw_fd(libc::open(path.as_ptr(), flags)) })
+    OpenOptions::new()
+        .read(true)
+        .custom_flags(libc::O_DIRECT)
+        .open(&CONFIG.input)
+        .with_context(|| format!("Failed to open input file: {}", &CONFIG.input.display()))
 }
 
 pub fn load_output() -> anyhow::Result<File> {
@@ -72,3 +61,18 @@ pub fn load_map_write() -> anyhow::Result<File> {
             )
         })
 }
+
+#[repr(C, align(512))]
+pub struct DirectIOBuffer(pub [u8; crate::MAX_BUFFER_SIZE]);
+
+impl Default for DirectIOBuffer {
+    fn default() -> Self {
+        Self([crate::FB_NULL_VALUE; _])
+    }
+}
+
+impl From<[u8; crate::MAX_BUFFER_SIZE]> for DirectIOBuffer {
+    fn from(value: [u8; crate::MAX_BUFFER_SIZE]) -> Self {
+        Self(value)
+    }
+}

src/main.rs

@@ -11,6 +11,8 @@ use anyhow;
 const FB_SECTOR_SIZE: usize = 2048;
 const FB_NULL_VALUE: u8 = 0;
 
+const MAX_BUFFER_SIZE: usize = FB_SECTOR_SIZE * 16;
+
 fn main() -> anyhow::Result<()> {
     let mut recover_tool = Recover::new()?;
     recover_tool.run()?;

src/mapping/cluster.rs

@@ -21,6 +21,7 @@ impl Default for Cluster {
 impl Cluster {
     /// Breaks apart into a vec of clusters,
     /// each of cluster_size, excepting last.
+    #[allow(dead_code)]
     pub fn subdivide(&mut self, cluster_len: usize) -> Vec<Cluster> {
         let domain_len = self.domain.len();
         let mut start = self.domain.start;
@@ -49,6 +50,8 @@ impl Cluster {
         clusters
     }
 
+    // This is used in unit tests at present. Ideally it probably shouldn't exist.
+    #[allow(dead_code)]
     pub fn set_stage(&mut self, stage: Stage) -> &mut Self {
         self.stage = stage;
         self

src/mapping/domain.rs

@@ -16,6 +16,7 @@ impl Default for Domain {
 
 impl Domain {
     /// Return length of domain in sectors.
+    #[allow(dead_code)]
     pub fn len(self) -> usize {
         self.end - self.start
     }

src/recovery.rs

@@ -2,13 +2,13 @@ use std::fs::File;
 use std::io::{BufWriter, Read, Seek, SeekFrom, Write};
 use std::usize;
 
-use anyhow::{Context, anyhow};
+use anyhow::Context;
 
 use crate::cli::CONFIG;
+use crate::io::DirectIOBuffer;
 use crate::mapping::prelude::*;
 
 #[derive(Debug)]
-#[allow(dead_code)]
 pub struct Recover {
     input: File,
     output: BufWriter<File>,
@@ -90,11 +90,12 @@ impl Recover {
 
     /// Attempt to copy all untested blocks.
     fn copy_untested(&mut self) -> anyhow::Result<()> {
+        let mut buf = DirectIOBuffer::default();
+
         for untested in self.map.get_clusters(Stage::Untested) {
             // Caching.
             let mut read_position: usize;
             let mut cluster: Cluster;
-            let mut buf: Vec<u8>;
             let mut buf_capacity = self.get_buf_capacity() as usize;
 
             dbg!(untested.domain);
@@ -104,7 +105,6 @@ impl Recover {
                 dbg!(read_position);
 
                 buf_capacity = buf_capacity.min(untested.domain.end - read_position);
-                buf = vec![crate::FB_NULL_VALUE; buf_capacity];
 
                 cluster = Cluster {
                     domain: Domain {
@@ -114,7 +114,7 @@ impl Recover {
                     stage: Stage::Intact,
                 };
 
-                if let Err(err) = self.input.read_exact(&mut buf) {
+                if let Err(err) = self.input.read_exact(&mut buf.0) {
                     // If buf were zeroed out before every read, one could theoretically recover
                     // part of that read given the assumption that all null values from the end to
                     // the first non-null value are unread, and some further padding from the last
@@ -141,7 +141,7 @@ impl Recover {
 
                 if cluster.stage == Stage::Intact {
                     self.output
-                        .write_all(&buf[0..buf_capacity])
+                        .write_all(&buf.0[0..buf_capacity])
                         .context("Failed to write data to output file")?;
                 }
 
@@ -151,6 +151,8 @@ impl Recover {
             }
         }
 
+        drop(buf);
+
         self.map.write_to(&mut crate::io::load_map_write()?)?;
 
         Ok(())
@@ -159,7 +161,7 @@ impl Recover {
     /// Set buffer capacity as cluster length in bytes.
     /// Varies depending on the recovery stage.
     fn get_buf_capacity(&mut self) -> u64 {
-        CONFIG.sector_size as u64 * CONFIG.cluster_length as u64
+        crate::MAX_BUFFER_SIZE.min(CONFIG.sector_size * CONFIG.cluster_length) as u64
     }
 
     /// Reloads the input and restores the seek position.