From 289670b313fe029161a872a250403f2d041f8a95 Mon Sep 17 00:00:00 2001
From: "Peter J. Holzer" <hjp@hjp.at>
Date: Sun, 25 Jul 2021 12:19:18 +0200
Subject: [PATCH] Treat zero length files specially for better performance

---
 lib/Simba/CA.pm | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/lib/Simba/CA.pm b/lib/Simba/CA.pm
index 15a624e..051587c 100644
--- a/lib/Simba/CA.pm
+++ b/lib/Simba/CA.pm
@@ -818,6 +818,25 @@ sub store_file {
             $self->log(5, "cannot link $self->{last_backup}/$f->{name} to $self->{this_backup}/$f->{name}: $!");
         }
     }
+    # If size is zero, check if we have seen a matching file before. If we have, link to it.
+    # Ubuntu contains a lot of zero sized files (about 8000 per installed kernel).
+    # Searching for them in the database is slow, so we special-case that here.
+    # We could generalize that, but I don't think that there will ever be enough identical
+    # non-empty files to make that worthwhile.
+    # XXX - not yet implemented.
+    if ($f->{s} == 0 && $f->{t} eq 'f') {
+        no warnings 'uninitialized'; # unix bits may not exist
+        my $k = "$f->{m} $f->{o} $f->{g} $f->{acl} $f->{setuid} $f->{setgid} $f->{sticky}";
+        if ($self->{null_files}{$k}) {
+            my $oldfile = $self->{null_files}{$k}{name};
+            my $backup_filename = "$self->{this_backup}/$f->{name}";
+            if (link($oldfile, $backup_filename)) {
+                $self->log(10, "linked (empty)");
+                $self->{counts}{dup10}++;
+                return $success;
+            }
+        }
+    }
 
     # else request from da
     unless ($self->{file_pid}) {
@@ -884,6 +903,11 @@ sub store_file {
             $self->setmeta($f);
             $self->log(10, "stored");
         }
+        if ($f->{s} == 0 && $f->{t} eq 'f') {
+            no warnings 'uninitialized'; # unix bits may not exist
+            my $k = "$f->{m} $f->{o} $f->{g} $f->{acl} $f->{setuid} $f->{setgid} $f->{sticky}";
+            $self->{null_files}{$k}{name} = $backup_filename;
+        }
     } else {
         $self->log(5, "unexpected header $header\n");
         $self->close_file_connection;