Add an initial version of some tools to compare (extended) K8 memory settings.
This generates (dirty) html with interpreted differences between PCI dumps,
based on the K8 socket F bkdg.
Signed-off-by: Ward Vandewege <ward@gnu.org>
Acked-by: Stepan Reinauer <stepan@coresystems.de>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@4886 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
diff --git a/util/amdtools/k8-interpret-extended-memory-settings.pl b/util/amdtools/k8-interpret-extended-memory-settings.pl
new file mode 100755
index 0000000..0bb4e62
--- /dev/null
+++ b/util/amdtools/k8-interpret-extended-memory-settings.pl
@@ -0,0 +1,248 @@
+#!/usr/bin/perl -w
+use Getopt::Long;
+
+use strict;
+
+my $NAME = $0;
+my $VERSION = '0.01';
+my $DATE = '2009-09-04';
+my $AUTHOR = "Ward Vandewege <ward\@jhvc.com>";
+my $COPYRIGHT = "2009";
+my $LICENSE = "GPL v3 - http://www.fsf.org/licenses/gpl.txt";
+my $URL = "http://coreboot.org";
+
+my $DEBUG = 0;
+
+our %info;
+
+$|=1;
+
+&main();
+
+sub version_information {
+ my ($NAME,$VERSION,$DATE,$COPYRIGHT,$AUTHOR,$LICENSE,$URL) = (shift,shift,shift,shift,shift,shift,shift);
+ print "\nThis is $NAME version $VERSION ($DATE)\n";
+ print "Copyright (c) $COPYRIGHT by $AUTHOR\n";
+ print "License: $LICENSE\n";
+ print "More information at $URL\n\n";
+ exit;
+}
+
+sub usage_information {
+ my $retval = "\n$NAME v$VERSION ($DATE)\n";
+ $retval .= "\nYou have not supplied all required parameters. $NAME takes these arguments:\n";
+ $retval .= " $NAME -f <filename1> -f <filename2>\n\n";
+ $retval .= " -f <filename1> is the name of a file with k8 memory configuration values\n";
+ $retval .= " -f <filename2> is the name of a second file with k8 memory configuration values, to compare with filename1\n";
+ $retval .= " -v (optional) provides version information\n";
+ $retval .= "\nSee the k8-read-mem-settings.sh script for an example of how to generate the input files to this script.\n\n";
+ print $retval;
+ exit;
+}
+
+sub parse_file {
+ my $register = '';
+ my $devreg = '';
+ my $filename = shift;
+ my %data = @_;
+ open(TMP, $filename) || die "Could not open $filename: $!\n";
+ while (<TMP>) {
+ chomp;
+ # Line format - pairs of lines:
+ # 0:18.2 98.l: 80000000
+ # 0:18.2 9C.l: 10111222
+ # First field is pci device. Second field is register offset (hex)
+ # where third field value (in hex) was read from.
+ my @tmp = split(/ /);
+ $tmp[1] =~ s/:$//; # strip optional trailing colon on second field
+
+ my $device = $tmp[0];
+ my $packed = pack("H*",$tmp[2]); # Pack our number so we can easily represent it in binary
+ my $binrep = unpack("B*", $packed); # Binary string representation
+
+ if ($tmp[1] eq '98.l') {
+ $register = ($tmp[2] =~ /(..)$/)[0]; # last 2 digits are (hex) of what we wrote to the register, if second field is 98.l
+ $devreg = "$device $register";
+ if ("$binrep" =~ /^1/) {
+ # bit 31 *must* be 1 if readout is to be correct
+ print "$tmp[0] - $register<br>\n" if ($DEBUG);
+ } else {
+ print "ERROR: we read too fast: $tmp[2] does not have bit 31 set ($binrep)\n";
+ exit;
+ }
+ } else {
+ # last field is register value (hex)
+ print "$tmp[2]h ($binrep)<br>\n" if ($DEBUG);
+ $data{$devreg} = {} if (!defined($data{$devreg}));
+ $data{$devreg}{$filename} = $packed;
+ }
+ }
+ return %data;
+}
+
+sub interpret_differences {
+ my $reg = shift;
+ $reg = sprintf("%02s",$reg);
+ my $tag1 = shift;
+ my $val1 = shift;
+ my $tag2 = shift;
+ my $val2 = shift;
+ my $retval = '';
+ my $retval2 = '';
+
+ # XOR values together - the positions with 1 after the XOR are the ones with the differences
+ my $xor = $val1 ^ $val2;
+
+ my @val1 = split(//,unpack("B*",$val1));
+ my @val2 = split(//,unpack("B*",$val2));
+ my @xor = split(//,unpack("B*",$xor));
+
+ my %changed;
+
+ if (!exists($info{$reg})) {
+ print STDERR "MISSING DATA for register $reg\n";
+ return '';
+ }
+
+ for (my $i=0; $i<=$#xor;$i++) {
+ my $invi = 31 - $i;
+ if ($xor[$i] eq '1') {
+#print STDERR "REG: $reg INVI: $invi\n";
+#print STDERR $info{$reg}{'fields'}{$invi} . "\n";
+#print STDERR $info{$reg}{'fields'}{$invi}{'range'} . "\n";
+ my $r = $info{$reg}{'fields'}{$invi}{'range'};
+# if (!exists($changed{$r})) {
+# $changed{$r}{'v1'} = '';
+# $changed{$r}{'v2'} = '';
+# }
+# $changed{$r}{'v1'} .= $val1[$i];
+# $changed{$r}{'v2'} .= $val2[$i];
+ $changed{$r}{'v1'} = 1;
+ $changed{$r}{'v2'} = 1;
+ }
+ }
+
+ foreach my $r (keys %changed) {
+ my $width = $info{$reg}{'ranges'}{$r}{'width'};
+ #$changed{$r}{'v1'} = sprintf("%0" . $width . "sb",$changed{$r}{'v1'});
+ #$changed{$r}{'v2'} = sprintf("%0" . $width . "sb",$changed{$r}{'v2'});
+ #my $v1 = $changed{$r}{'v1'};
+ #my $v2 = $changed{$r}{'v2'};
+ my $v1 = substr(unpack("B*",$val1),31-$info{$reg}{'ranges'}{$r}{'end'},$info{$reg}{'ranges'}{$r}{'width'}) . 'b';
+ my $v2 = substr(unpack("B*",$val2),31-$info{$reg}{'ranges'}{$r}{'end'},$info{$reg}{'ranges'}{$r}{'width'}) . 'b';
+
+ my $desc = $info{$reg}{'ranges'}{$r}{'description'};
+ $desc =~ s/\n+/<br>/g;
+
+ $retval2 .= $info{$reg}{'ranges'}{$r}{'function'} . " (" . $info{$reg}{'ranges'}{$r}{'mnemonic'} . ") - Bits ($r)" . "<br>";
+ $retval2 .= " <i>$desc</i><p>" if ($desc ne '');
+
+ $v1 = $v1 . " (" . $info{$reg}{'ranges'}{$r}{'values'}{$v1} . ")" if (exists($info{$reg}{'ranges'}{$r}{'values'}{$v1}));
+ $v2 = $v2 . " (" . $info{$reg}{'ranges'}{$r}{'values'}{$v2} . ")" if (exists($info{$reg}{'ranges'}{$r}{'values'}{$v2}));
+ $retval2 .= sprintf("<b><a href=\"$tag1\">%44s</a>: %s</b>\n",$tag1, $v1);
+ $retval2 .= sprintf("<b><a href=\"$tag2\">%44s</a>: %s</b>\n",$tag2, $v2);
+ $retval2 .= "<p>";
+ }
+
+
+# this prints out the bitwise differences. TODO: clean up
+
+# for (my $i=0; $i<=$#xor;$i++) {
+# my $invi = 31 - $i;
+# if ($xor[$i] eq '1') {
+# my $m = $info{$reg}{'fields'}{$invi}{'mnemonic'};
+# my $f = $info{$reg}{'fields'}{$invi}{'function'};
+# my $range = $info{$reg}{'fields'}{$invi}{'range'};
+# if ($m && $f) {
+# $retval2 .= "Bit $invi ($info{$reg}{'fields'}{$invi}{'mnemonic'} - $info{$reg}{'fields'}{$invi}{'function'}):\n";
+# $retval2 .= sprintf("%32s: %d\n",$tag1, $val1[$i]);
+# $retval2 .= sprintf("%32s: %d\n",$tag2, $val2[$i]);
+# } else {
+# $retval2 .= "Bit $invi:\n";
+# $retval2 .= sprintf("%32s: %d\n",$tag1, $val1[$i]);
+# $retval2 .= sprintf("%32s: %d\n",$tag2, $val2[$i]);
+# }
+# }
+# }
+
+ $retval .= "\n";
+ if ($retval2 ne '') {
+ $retval .= "\n\n$retval2\n";
+ my $n = $info{$reg}{'name'};
+ my $d = $info{$reg}{'description'};
+ $n ||= '';
+ $d ||= '';
+ my $old = $retval;
+ $retval = '';
+ $retval .= sprintf("%40s -> %s<br>\n","XOR",unpack("B*",$xor)) if ($DEBUG);
+ $retval .= "\n$n\n" if ($n ne '');
+ $retval .= " $d" if ($d ne '');
+ $retval .= $old;
+ $retval .= "\n";
+ }
+
+ return "<pre>$retval</pre>";
+}
+
+sub load_datafile {
+ my $file = 'bkdg.data';
+ my $return = '';
+
+ if (-f $file) {
+ unless ($return = do $file) {
+ warn "couldn't parse $file: $@" if $@;
+ warn "couldn't do $file: $!" unless defined $return;
+ warn "couldn't run $file" unless $return;
+ }
+ } else {
+ print "Warning: data file '$file' not found - $0 will only report on differing bits without explanation.\n";
+ }
+
+}
+
+sub main {
+ my @filenames;
+ my $version = 0;
+ my %data;
+
+ GetOptions ("filename=s" => \@filenames, "version" => \$version);
+
+ &version_information($NAME,$VERSION,$DATE,$COPYRIGHT,$AUTHOR,$LICENSE,$URL) if ($version);
+
+ &usage_information() if ($#filenames < 1);
+
+ &load_datafile();
+
+ foreach my $file (@filenames) {
+ print STDERR "processing $file\n";
+ %data = &parse_file($file,%data);
+ }
+
+ print "<html>\n<body>\n";
+
+ foreach my $key (sort keys %data) {
+ my $first = pack("H*",'00000000');
+ my $firstfile = '';
+ foreach my $k2 (reverse sort keys %{$data{$key}}) {
+ if (unpack("H*",$first) eq '00000000') {
+ $first = $data{$key}{$k2};
+ $firstfile = $k2;
+ }
+ if (unpack("H*",$first) ne unpack("H*",$data{$key}{$k2})) {
+ my $reg = ($key =~ /\s+([a-z0-9]+)$/i)[0];
+ print "$key\n";
+ if ($DEBUG) {
+ print "<pre>";
+ printf("%44s -> %s (%s)\n",$firstfile,unpack("B*",$first),unpack("H*",$first));
+ printf("%44s -> %s (%s)\n",$k2,unpack("B*",$data{$key}{$k2}),unpack("H*",$data{$key}{$k2}));
+ print "</pre>";
+ }
+
+ print &interpret_differences($reg,$firstfile,$first,$k2,$data{$key}{$k2});
+ }
+ }
+ }
+ print "</body>\n</html>\n";
+
+}
+