Merge pull request #53785 from danbst/get-hash-doc

manual: document ways of obtaining source hashes

2 files changed, 122 insertions, 3 deletions

diff --git a/doc/Makefile b/doc/Makefile
index c6aed62a939..91b62fe138b 100644
--- a/doc/Makefile
+++ b/doc/Makefile
@@ -9,8 +9,10 @@ debug:
 
 .PHONY: format
 format:
-	find . -iname '*.xml' -type f -print0 | xargs -0 -I{} -n1 \
-		xmlformat --config-file "$$XMLFORMAT_CONFIG" -i {}
+	find . -iname '*.xml' -type f | while read f; do \
+		echo $$f ;\
+		xmlformat --config-file "$$XMLFORMAT_CONFIG" -i $$f ;\
+	done
 
 .PHONY: fix-misc-xml
 fix-misc-xml:
diff --git a/doc/coding-conventions.xml b/doc/coding-conventions.xml
index a8a4557b461..d2c7a1baae9 100644
--- a/doc/coding-conventions.xml
+++ b/doc/coding-conventions.xml
@@ -814,7 +814,7 @@ args.stdenv.mkDerivation (args // {
 
   <para>
    There are multiple ways to fetch a package source in nixpkgs. The general
-   guideline is that you should package sources with a high degree of
+   guideline is that you should package reproducible sources with a high degree of
    availability. Right now there is only one fetcher which has mirroring
    support and that is <literal>fetchurl</literal>. Note that you should also
    prefer protocols which have a corresponding proxy environment variable.
@@ -876,6 +876,123 @@ src = fetchFromGitHub {
    </itemizedlist>
   </para>
  </section>
+ <section xml:id="sec-source-hashes">
+  <title>Obtaining source hash</title>
+
+  <para>
+   Preferred source hash type is sha256. There are several ways to get it.
+  </para>
+
+  <orderedlist>
+   <listitem>
+    <para>
+     Prefetch URL (with <literal>nix-prefetch-<replaceable>XXX</replaceable>
+     <replaceable>URL</replaceable></literal>, where
+     <replaceable>XXX</replaceable> is one of <literal>url</literal>,
+     <literal>git</literal>, <literal>hg</literal>, <literal>cvs</literal>,
+     <literal>bzr</literal>, <literal>svn</literal>). Hash is printed to
+     stdout.
+    </para>
+   </listitem>
+   <listitem>
+    <para>
+     Prefetch by package source (with <literal>nix-prefetch-url
+     '&lt;nixpkgs&gt;' -A <replaceable>PACKAGE</replaceable>.src</literal>,
+     where <replaceable>PACKAGE</replaceable> is package attribute name). Hash
+     is printed to stdout.
+    </para>
+    <para>
+     This works well when you've upgraded existing package version and want to
+     find out new hash, but is useless if package can't be accessed by
+     attribute or package has multiple sources (<literal>.srcs</literal>,
+     architecture-dependent sources, etc).
+    </para>
+   </listitem>
+   <listitem>
+    <para>
+     Upstream provided hash: use it when upstream provides
+     <literal>sha256</literal> or <literal>sha512</literal> (when upstream
+     provides <literal>md5</literal>, don't use it, compute
+     <literal>sha256</literal> instead).
+    </para>
+    <para>
+     A little nuance is that <literal>nix-prefetch-*</literal> tools produce
+     hash encoded with <literal>base32</literal>, but upstream usually provides
+     hexadecimal (<literal>base16</literal>) encoding. Fetchers understand both
+     formats. Nixpkgs does not standardize on any one format.
+    </para>
+    <para>
+     You can convert between formats with nix-hash, for example:
+<screen>
+$ nix-hash --type sha256 --to-base32 <replaceable>HASH</replaceable>
+</screen>
+    </para>
+   </listitem>
+   <listitem>
+    <para>
+     Extracting hash from local source tarball can be done with
+     <literal>sha256sum</literal>. Use <literal>nix-prefetch-url
+     file:///path/to/tarball </literal> if you want base32 hash.
+    </para>
+   </listitem>
+   <listitem>
+    <para>
+     Fake hash: set fake hash in package expression, perform build and extract
+     correct hash from error Nix prints.
+    </para>
+    <para>
+     For package updates it is enough to change one symbol to make hash fake.
+     For new packages, you can use <literal>lib.fakeSha256</literal>,
+     <literal>lib.fakeSha512</literal> or any other fake hash.
+    </para>
+    <para>
+     This is last resort method when reconstructing source URL is non-trivial
+     and <literal>nix-prefetch-url -A</literal> isn't applicable (for example,
+     <link xlink:href="https://github.com/NixOS/nixpkgs/blob/d2ab091dd308b99e4912b805a5eb088dd536adb9/pkgs/applications/video/kodi/default.nix#L73">
+     one of <literal>kodi</literal> dependencies</link>). The easiest way then
+     would be replace hash with a fake one and rebuild. Nix build will fail and
+     error message will contain desired hash.
+    </para>
+    <warning><para>This method has security problems. Check below for details.</para></warning>
+   </listitem>
+  </orderedlist>
+
+  <section xml:id="sec-source-hashes-security">
+   <title>Obtaining hashes securely</title>
+   <para>
+    Let's say Man-in-the-Middle (MITM) sits close to your network. Then instead of fetching
+    source you can fetch malware, and instead of source hash you get hash of malware. Here are
+    security considerations for this scenario:
+   </para>
+   <itemizedlist>
+    <listitem>
+     <para>
+      <literal>http://</literal> URLs are not secure to prefetch hash from;
+     </para>
+    </listitem>
+    <listitem>
+     <para>
+      hashes from upstream (in method 3) should be obtained via secure protocol;
+     </para>
+    </listitem>
+    <listitem>
+     <para>
+      <literal>https://</literal> URLs are secure in methods 1, 2, 3;
+     </para>
+    </listitem>
+    <listitem>
+     <para>
+      <literal>https://</literal> URLs are not secure in method 5. When obtaining hashes
+      with fake hash method, TLS checks are disabled. So
+      refetch source hash from several different networks to exclude MITM scenario.
+      Alternatively, use fake hash method to make Nix error, but instead of extracting
+      hash from error, extract <literal>https://</literal> URL and prefetch it
+      with method 1.
+     </para>
+    </listitem>
+   </itemizedlist>
+  </section>
+ </section>
  <section xml:id="sec-patches">
   <title>Patches</title>