From 7ea8f69a06300eba5d42b46a6f85daeacc2430a3 Mon Sep 17 00:00:00 2001
From: Natanael Copa <ncopa@alpinelinux.org>
Date: Wed, 5 Apr 2023 18:45:32 +0200
Subject: [PATCH] Fix user-data decompression on all shells

yash[1] and zsh does not handle 8-bit bytes in strings good. Work around it
by using printf ... | cmp ... to compare the header. This should be 100%
posix compatible.

Also fix some of the compression header magic bytes:

- unxz: Use octal aas the string '\3757zXZ\000' was misinterpreted by some
  shells.
- lzma: the third byte represents compression mode[2], and we want
  support all compression modes not only '8', so we only check the first
  three bytes instead of 4.

[1]: Upstream report: https://osdn.net/projects/yash/ticket/47772
[2]: https://github.com/frizb/FirmwareReverseEngineering/blob/master/IdentifyingCompressionAlgorithms.md#lzma
---
 .gitlab-ci.yml           |  2 +-
 lib/tiny-cloud/init-main | 38 ++++++++++++++++++--------------------
 tests/init-main.test     |  5 +++--
 3 files changed, 22 insertions(+), 23 deletions(-)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 1519c23..93ae0d3 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -2,7 +2,7 @@ test-default:
   image: alpine:latest
   stage: test
   script:
-    - apk add make kyua yx
+    - apk add make kyua yx xz lz4 zstd
     - make -j $(nproc) check
   tags:
     - docker-alpine
diff --git a/lib/tiny-cloud/init-main b/lib/tiny-cloud/init-main
index 2788dc8..1728d3a 100644
--- a/lib/tiny-cloud/init-main
+++ b/lib/tiny-cloud/init-main
@@ -40,11 +40,6 @@ set_ssh_keys() {
     imds @ssh-keys > "$keys_file"
 }
 
-match_header() {
-    local bytes=$(echo -en "$1")
-    [ "$bytes" = "$(dd bs=1 count=${#bytes} if="$2" 2>/dev/null)" ]
-}
-
 save_userdata() {
     skip_action save_userdata && return
 
@@ -55,22 +50,25 @@ save_userdata() {
     imds -e @userdata > "$tmpfile"
     cmd="cat"
     if ! skip_action decompress_userdata; then
-        if match_header '\037\213\010' "$tmpfile"; then
-            cmd="gzip -dc"
-        elif match_header 'BZh' "$tmpfile"; then
-            cmd="bzip2 -dc"
-        elif match_header '\3757zXZ\000' "$tmpfile"; then
-            cmd="unxz -c"
-        elif match_header '\135\0\0\0' "$tmpfile"; then
-            cmd="lzma -dc"
-        elif match_header '\211\114\132' "$tmpfile"; then
-            cmd="lzop -dc"
-        elif match_header '\004\042\115\030' "$tmpfile"; then
-            cmd="lz4 -dc"
-        elif match_header '(\265/\375' "$tmpfile"; then
-            cmd="zstd -dc"
+        if printf '\037\213\010' | cmp -s -n 3 "$tmpfile"; then
+            gzip -dc "$tmpfile" > "$userdata"
+        elif printf 'BZh' | cmp -s -n 3 "$tmpfile"; then
+            bzip2 -dc "$tmpfile" > "$userdata"
+        elif printf '\375\067\172\130\132\000' | cmp -s -n 6 "$tmpfile"; then
+            unxz -c "$tmpfile" > "$userdata"
+        elif printf '\135\000\000' | cmp -s -n 3 "$tmpfile"; then
+            lzma -dc "$tmpfile" > "$userdata"
+        elif printf '\211\114\132' | cmp -s -n 3 "$tmpfile"; then
+            lzop -dc "$tmpfile" > "$userdata"
+        elif printf '\004\042\115\030' | cmp -s -n 4 "$tmpfile"; then
+            lz4 -dc "$tmpfile" > "$userdata"
+        elif printf '(\265/\375' | cmp -s -n 4 "$tmpfile"; then
+            zstd -dc "$tmpfile" > "$userdata"
+        else
+            cp "$tmpfile" "$userdata"
         fi
+    else
+        cp "$tmpfile" "$userdata"
     fi
-    $cmd "$tmpfile" > "$userdata"
     rm "$tmpfile"
 }
diff --git a/tests/init-main.test b/tests/init-main.test
index 3350f8c..0206416 100755
--- a/tests/init-main.test
+++ b/tests/init-main.test
@@ -99,8 +99,9 @@ save_userdata_compressed_body() {
 		set_nocloud_userdata_from_file tmpfile
 		CLOUD="nocloud" atf_check \
 			sh -c ". \"$lib\"; save_userdata"
-		grep "^#cloud-config" var/lib/cloud/user-data \
-			|| atf_fail "$comp failed"
+		if ! grep "^#cloud-config" var/lib/cloud/user-data; then
+			atf_fail "$comp failed"
+		fi
 	done
 }