about summary refs log tree commit diff
path: root/tvix/glue/src/tvix_io.rs
diff options
context:
space:
mode:
authorConnor Brewster <cbrewster@hey.com>2024-03-22T23·52-0500
committerConnor Brewster <cbrewster@hey.com>2024-04-09T17·31+0000
commit63116d8c21afdc50725ae93d13839fe1915b06b7 (patch)
tree4997838251dac809c2917b35e5d32224030ba595 /tvix/glue/src/tvix_io.rs
parent17849c5c0033fa1909f0403b5d5e6a5e018b7fee (diff)
fix(tvix): Avoid buffering file into memory in builtins.hashFile r/7882
Right now `builtins.hashFile` always reads the entire file into memory
before hashing, which is not ideal for large files. This replaces
`read_to_string` with `open_file` which allows calculating the hash of
the file without buffering it entirely into memory. Other callers can
continue to buffer into memory if they choose, but they still use the
`open_file` VM request and then call `read_to_string` or `read_to_end`
on the `std::io::Reader`.

Fixes b/380

Change-Id: Ifa1c8324bcee8f751604b0b449feab875c632fda
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11236
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Diffstat (limited to 'tvix/glue/src/tvix_io.rs')
-rw-r--r--tvix/glue/src/tvix_io.rs11
1 files changed, 5 insertions, 6 deletions
diff --git a/tvix/glue/src/tvix_io.rs b/tvix/glue/src/tvix_io.rs
index 95146df7287e..0e5f23b99093 100644
--- a/tvix/glue/src/tvix_io.rs
+++ b/tvix/glue/src/tvix_io.rs
@@ -8,7 +8,7 @@
 //! otherwise fundamental features like nixpkgs bootstrapping and hash
 //! calculation will not work.
 
-use std::io;
+use std::io::{self, Cursor};
 use std::path::{Path, PathBuf};
 use tvix_eval::{EvalIO, FileType};
 
@@ -44,7 +44,7 @@ where
         self.actual.as_ref().path_exists(path)
     }
 
-    fn read_to_end(&self, path: &Path) -> io::Result<Vec<u8>> {
+    fn open(&self, path: &Path) -> io::Result<Box<dyn io::Read>> {
         // Bundled version of corepkgs/fetchurl.nix. The counterpart
         // of this happens in [crate::configure_nix_path], where the `nix_path`
         // of the evaluation has `nix=/__corepkgs__` added to it.
@@ -52,13 +52,12 @@ where
         // This workaround is similar to what cppnix does for passing
         // the path through.
         //
-        // TODO: this comparison is bad and allocates, we should use
-        // the sane path library.
+        // TODO: this comparison is bad we should use the sane path library.
         if path.starts_with("/__corepkgs__/fetchurl.nix") {
-            return Ok(include_bytes!("fetchurl.nix").to_vec());
+            return Ok(Box::new(Cursor::new(include_bytes!("fetchurl.nix"))));
         }
 
-        self.actual.as_ref().read_to_end(path)
+        self.actual.as_ref().open(path)
     }
 
     fn read_dir(&self, path: &Path) -> io::Result<Vec<(bytes::Bytes, FileType)>> {