From 00aeb8181190a2f731ea48d52841898d70595778 Mon Sep 17 00:00:00 2001
From: Daniel Barlow <dan@telent.net>
Date: Fri, 19 May 2023 23:49:11 +0100
Subject: [PATCH] min-list-garbage: check all store paths against file

this is step 1 of min-collect-garbage, no point implementing
deletion ourselves when rm -r exists

(arguably no point in implementing any of it, but this is the bit we
can't do efficiently in bourne shell - it means we're reading the
store-paths list once instead of grepping it afresh for every entry in
/nix/store/)
---
 pkgs/min-copy-closure/default.nix        |  8 ++-
 pkgs/min-copy-closure/min-list-garbage.c | 79 ++++++++++++++++++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
 create mode 100644 pkgs/min-copy-closure/min-list-garbage.c

diff --git a/pkgs/min-copy-closure/default.nix b/pkgs/min-copy-closure/default.nix
index f0dcf6b..1e07897 100644
--- a/pkgs/min-copy-closure/default.nix
+++ b/pkgs/min-copy-closure/default.nix
@@ -8,9 +8,13 @@
   buildInputs = [ ];
   propagatedBuildInputs = [ cpio openssh nix ];
   src = ./.;
+  makeFlags = [ "min-list-garbage" ];
   installPhase = ''
     mkdir -p $out/bin
-    cp min-copy-closure.sh $out/bin/min-copy-closure
-    cp liminix-rebuild.sh $out/bin/liminix-rebuild
+    for i in min-copy-closure liminix-rebuild; do
+      echo $i
+      cp ''${i}.sh $out/bin/$i
+    done
+    cp min-list-garbage $out/bin
   '';
 }
diff --git a/pkgs/min-copy-closure/min-list-garbage.c b/pkgs/min-copy-closure/min-list-garbage.c
new file mode 100644
index 0000000..a260a8b
--- /dev/null
+++ b/pkgs/min-copy-closure/min-list-garbage.c
@@ -0,0 +1,79 @@
+#include <dirent.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <string.h>
+#include <unistd.h>
+
+typedef char hash_t[32];
+hash_t *hash_list = NULL;
+int hash_list_entries = 0;
+int hash_list_size = 0;
+
+int add_list_entry(char *name)
+{
+    name += strlen("/nix/store/");
+    if(hash_list_entries >= hash_list_size) {
+	if(hash_list_size == 0) hash_list_size =  4;
+	hash_list_size = hash_list_size * 2;
+	hash_list = realloc(hash_list, hash_list_size * sizeof(hash_t));
+    }
+    strncpy(hash_list[hash_list_entries++], name, 32);
+    printf("%d %d\n", hash_list_entries, hash_list_size);
+}
+
+int read_list(char *filename)
+{
+    char s[1024];
+    FILE *fp;
+    if(fp = fopen(filename, "r")) {
+	while(fgets(s, 1024, fp)) {
+	    if(strrchr(s, '\n')) {
+		add_list_entry(s);
+	    } else {
+		puts("hash list entry too long");
+		exit(1);
+	    }
+	}
+    }
+    fclose(fp);
+}
+
+int on_list(char *name)
+{
+    int i;
+    for(i = 0; i < hash_list_entries; i++) {
+	if(!strncmp(hash_list[i], name, 32)) {
+	    return 1;
+	}
+    }
+    return 0;
+}
+
+int main(int argc, char * argv[])
+{
+    struct dirent * de;
+    DIR  * dirp;
+    char hash[32];
+
+    if(argc < 2) {
+	puts("Usage: min-list-garbage store-paths-file\n\nChecks all store paths against the list of expected paths in store-paths-file,\n and prints any which are present unexpectedly");
+	exit(1);
+    }
+
+    read_list(argv[1]);
+    
+    if((dirp = opendir("/nix/store")) == NULL) {
+	puts("can't open /nix/store");
+	exit(1);
+    }
+
+     while(de = readdir(dirp)) {
+	if(strlen(de->d_name) >= 32) {
+	    strncpy(hash, de->d_name, 32);
+	    if(!on_list(hash)) puts(de->d_name);
+	}
+    }
+
+    closedir(dirp);
+}