From 35f23240e77d5f2582ae6124dadadc28cdbeefaa Mon Sep 17 00:00:00 2001
From: m5r <mokhtar@mial.net>
Date: Sun, 2 Aug 2026 16:47:14 +0200
Subject: [PATCH] fix blocklist fetcher aborting the process on a buffered read

the response reader was constructed over transfer_buf and then read into that
same buffer, so readSliceShort memcpy'd the reader's buffer onto itself and
panicked on any read that found bytes already buffered. stream the body
straight into the caller's writer instead: no second copy, no aliasing.
---
 src/filter/fetcher.zig | 95 ++++++++++++++++++++++++++++++++++++------
 1 file changed, 82 insertions(+), 13 deletions(-)

diff --git a/src/filter/fetcher.zig b/src/filter/fetcher.zig
index 5a97192..56a03c0 100644
--- a/src/filter/fetcher.zig
+++ b/src/filter/fetcher.zig
@@ -101,22 +101,45 @@ pub const Fetcher = struct {
         }
 
         const body = resp.reader(self.transfer_buf);
-        var total: u64 = 0;
-        while (true) {
-            const n = body.readSliceShort(self.transfer_buf) catch |err|
-                return mapError(err, .receive);
-            if (n == 0) break;
-            total += n;
-            if (total > max_body_bytes) return error.BodyTooLarge;
-            // The caller owns `w` and can read the concrete failure from its
-            // own writer; this taxonomy has no member for a failing sink.
-            w.writeAll(self.transfer_buf[0..n]) catch return error.Unexpected;
-        }
-
-        return .{ .bytes_read = total, .status = resp.head.status };
+        return .{ .bytes_read = try pumpBody(body, w, max_body_bytes), .status = resp.head.status };
     }
 };
 
+/// Streams the body into `w`, refusing more than `cap` bytes. `fetch` always
+/// passes `max_body_bytes`; the parameter exists so the refusal is testable
+/// without a 64 MiB body.
+///
+/// `w` is the only destination: `Reader.stream` hands the reader's buffered
+/// bytes straight to the writer, so the body is never copied twice and the
+/// reader's own buffer is never a destination slice. Reading into that buffer
+/// instead — `readSliceShort(self.transfer_buf)` on a reader constructed over
+/// `self.transfer_buf` — makes `@memcpy` copy the buffer onto itself
+/// (Reader.zig:677) and panics the process on the first read that finds bytes
+/// already buffered.
+fn pumpBody(body: *std.Io.Reader, w: *std.Io.Writer, cap: usize) Error!u64 {
+    var total: usize = 0;
+    while (total < cap) {
+        total += body.stream(w, .limited(cap - total)) catch |err| switch (err) {
+            error.EndOfStream => return total,
+            // The caller owns `w` and can read the concrete failure from its
+            // own writer; this taxonomy has no member for a failing sink.
+            error.WriteFailed => return error.Unexpected,
+            else => |e| return mapError(e, .receive),
+        };
+    }
+
+    // `cap` written exactly. One more byte separates a body that fits from one
+    // that was cut off, and it must not reach `w`.
+    var probe_buf: [1]u8 = undefined;
+    var probe: std.Io.Writer.Discarding = .init(&probe_buf);
+    const extra = body.stream(&probe.writer, .limited(1)) catch |err| switch (err) {
+        error.EndOfStream => return total,
+        error.WriteFailed => return error.Unexpected,
+        else => |e| return mapError(e, .receive),
+    };
+    return if (extra == 0) total else error.BodyTooLarge;
+}
+
 /// A scheme other than `http`/`https`, an unparseable URL and a URL with no
 /// host are one fault to the operator: the source row is unusable.
 fn parseUrl(url: []const u8) Error!std.Uri {
@@ -170,6 +193,52 @@ fn narrowLocal(local: transport.ExchangeError) Error {
 
 const testing = std.testing;
 
+// `Reader.fixed` hands back a reader whose whole payload is already buffered
+// (seek 0, end len) — the state that made the previous implementation copy the
+// reader's buffer onto itself and abort the process mid-download.
+
+test "pumpBody streams a fully buffered body without aliasing its source" {
+    const payload = "0.0.0.0 ads.example.com\n0.0.0.0 tracker.example\n";
+    var body: std.Io.Reader = .fixed(payload);
+    var out: [payload.len]u8 = undefined;
+    var sink: std.Io.Writer = .fixed(&out);
+
+    const n = try pumpBody(&body, &sink, max_body_bytes);
+    try testing.expectEqual(@as(u64, payload.len), n);
+    try testing.expectEqualStrings(payload, out[0..payload.len]);
+}
+
+test "pumpBody accepts a body of exactly the cap" {
+    const payload = "abcdefgh";
+    var body: std.Io.Reader = .fixed(payload);
+    var out: [payload.len]u8 = undefined;
+    var sink: std.Io.Writer = .fixed(&out);
+
+    const n = try pumpBody(&body, &sink, payload.len);
+    try testing.expectEqual(@as(u64, payload.len), n);
+    try testing.expectEqualStrings(payload, out[0..payload.len]);
+}
+
+test "pumpBody refuses a body one byte over the cap" {
+    const payload = "abcdefgh";
+    var body: std.Io.Reader = .fixed(payload);
+    var out: [payload.len]u8 = undefined;
+    var sink: std.Io.Writer = .fixed(&out);
+
+    try testing.expectError(
+        error.BodyTooLarge,
+        pumpBody(&body, &sink, payload.len - 1),
+    );
+}
+
+test "pumpBody reports an empty body as zero bytes" {
+    var body: std.Io.Reader = .fixed("");
+    var sink_buf: [0]u8 = .{};
+    var discarding: std.Io.Writer.Discarding = .init(&sink_buf);
+
+    try testing.expectEqual(@as(u64, 0), try pumpBody(&body, &discarding.writer, max_body_bytes));
+}
+
 fn undefinedFetcher(transfer_buf: []u8, redirect_buf: []u8) Fetcher {
     // `http` is never driven: every test below asserts a rejection that
     // happens before the first client call.