Only include kernel udebs for the highest ABI we know about

Don't waste space on installation media with older udebs from the
archive that have been kept around for other users.

debian/changelog

@@ -1,4 +1,5 @@
 debian-cd (3.1.26-buster) UNRELEASED; urgency=medium
+
   [ Steve McIntyre ]
   * Backports of fixes from debian-cd in unstable
     + Keep grub resolution in EFI boot, to avoid tiny fonts (closes: #935546).
@@ -7,6 +8,9 @@ debian-cd (3.1.26-buster) UNRELEASED; urgency=medium
     + Remove lilo-installer and elilo-installer for all arches
     + tools/update_popcon: cope with lower-case HTTP headers too; Debian
       just enabled http2 so this broke
+    + tools/generate_di_list: Only include kernel udebs for the
+      highest ABI we know about. Don't waste space on new installation
+      media with older udebs from the archive.
 
  -- Samuel Thibault <sthibault@debian.org>  Fri, 23 Aug 2019 21:51:50 +0200
 

tools/generate_di_list

@@ -1,12 +1,18 @@
-#!/usr/bin/perl -w
+#!/usr/bin/perl
 # Generate a list of packages required for debian-installer
 # This script makes use of the following variables that need to be preset:
 # MIRROR, DI_CODENAME, BASEDIR
+
+use strict;
+use warnings;
+
 die "Missing \$MIRROR variable" unless $ENV{MIRROR};
 die "Missing \$DI_CODENAME variable" unless $ENV{DI_CODENAME};
 die "Missing \$BASEDIR variable" unless $ENV{BASEDIR};
 die "Missing \$ARCHES variable" unless $ENV{ARCHES};
 
+my $catz = $ENV{BASEDIR} . "/tools/catz";
+
 # Early exit if we're building a source-only CD
 exit 0 if $ENV{ARCHES} =~ /^\s*source\s*$/;
 
@@ -38,6 +44,33 @@ print OUT << "EOF";
  */
 EOF
 
+sub di_ker_abi_to_number ($$$$) {
+    # Make up a version we can compare sensibly from
+    # the version and ABI
+    my $maj = shift;
+    my $min = shift;
+    my $patch = shift;
+    my $abi = shift;
+    my $kernel_ver = ($maj * 1000000000)
+	+ ($min * 1000000)
+	+ ($patch * 1000)
+	+ $abi;
+    return $kernel_ver;
+}
+
+sub number_to_di_ker_abi ($) {
+    # Convert back to a useful string
+    my $num = shift;
+    my $maj = int($num / 1000000000);
+    $num -= ($maj * 1000000000);
+    my $min = int($num / 1000000);
+    $num -= ($min * 1000000);
+    my $patch = int($num / 1000);
+    $num -= ($patch * 1000);
+    my $abi = $num;
+    return "$maj.$min.$patch-$abi";
+}
+
 my @common_excludes = read_exclude("exclude-udebs");
 my $mirror_path = "$ENV{MIRROR}/dists/$ENV{DI_CODENAME}";
 my @components = qw(main);
@@ -65,18 +98,71 @@ foreach my $arch (@ARCHES) {
         push @exclude, read_exclude("exclude-udebs-$arch")
             if -e exclude_path("exclude-udebs-$arch");
 
-	if (-f $pgz) {
+	my $pz = $pgz;
-	    foreach my $udeb (map { chomp; $_ } `zcat $pgz | awk '/^Package:/ {print \$2}'`) {
+	if (! -f $pz) {
-		$output .= "$udeb\n" unless grep { $udeb =~ /^${_}$/ } @exclude;
+	    $pz = $pxz;
 	}
-	} elsif (-f $pxz) {
+	if (! -f $pz) {
-	    foreach my $udeb (map { chomp; $_ } `xzcat $pxz | awk '/^Package:/ {print \$2}'`) {
-		$output .= "$udeb\n" unless grep { $udeb =~ /^${_}$/ } @exclude;
-	    }
-	} else {
             print "Missing package file for $arch/$component.\n";
             next;
 	}
+
+	# Two passes here
+	# First, need to find the highest kernel ABI in the archive
+	# Next, list all the udebs that:
+	#   1. are not otherwise excluded
+	#   2. are not kernel driver udebs, OR
+	#      are kernel driver udebs with the right kernel ABI
+
+	unless ( open (PZ, "$catz $pz |") ) {
+	    warn "failed to read package file $pz: $1";
+	    next;
+	}
+
+	my @output_udebs;
+	my %driver_udebs;
+	my $highest_kernel_ver = 0;
+
+	while (defined (my $line = <PZ>)) {
+	    chomp $line;
+	    if ($line =~ m/^Package: (\S+)/) {
+		my $udeb = $1;
+		if (grep { $udeb =~ /^${_}$/ } @exclude) {
+		    next;
+		}
+		if ($udeb =~ m/-modules-([0-9]*)\.([0-9]*)\.([0-9]*)-(\d+)-.*-di/) {
+		    my $kernel_ver = di_ker_abi_to_number($1, $2, $3, $4);
+		    if ($kernel_ver > $highest_kernel_ver) {
+			$highest_kernel_ver = $kernel_ver;
+		    }
+		    # Append this driver udeb to a list for that kernel_ver
+		    push(@{ $driver_udebs{$kernel_ver} }, $udeb);
+		} else {
+		    #print "found non-driver udeb $udeb\n";
+		    push @output_udebs, $udeb;
+		}
+	    }
+	}
+	close PZ;
+	$output .= "/* First, generic udebs */\n";
+	foreach my $udeb (@output_udebs) {
+	    $output .= "$udeb\n";
+	}
+	my $num_this_abi = scalar (@{ $driver_udebs{$highest_kernel_ver}});
+	$output .= "/* Next: $num_this_abi udebs for kernel/ABI version ";
+	$output .= number_to_di_ker_abi($highest_kernel_ver);
+	$output .= " */\n";
+	foreach my $udeb (@{ $driver_udebs{$highest_kernel_ver}}) {
+	    $output .= "$udeb\n";
+	}
+	foreach my $key (keys (%driver_udebs)) {
+	    if ($key != $highest_kernel_ver) {
+		$num_this_abi = scalar (@{ $driver_udebs{$key}});
+		$output .= "/* IGNORING $num_this_abi udebs for kernel/ABI version ";
+		$output .= number_to_di_ker_abi($key);
+		$output .= " */\n";
+	    }
+	}
     }
 
     next unless $output;