From 60a91e4da4f2daf2b10143fc148a8043312b61e5 Mon Sep 17 00:00:00 2001 From: Aristeu Rozanski Date: Wed, 1 Aug 2018 16:29:58 -0400 Subject: [PATCH] rasdaemon: ras-mc-ctl: add option to show error counts In some scenarios it might not be desirable to have a daemon running to parse and store the errors provided by EDAC and only having the number of CEs and UEs is enough. This patch implements this feature as an ras-mc-ctl option. Signed-off-by: Aristeu Rozanski Signed-off-by: Mauro Carvalho Chehab --- util/ras-mc-ctl.in | 75 ++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 73 insertions(+), 2 deletions(-) diff --git a/util/ras-mc-ctl.in b/util/ras-mc-ctl.in index 38b7824..aee431a 100755 --- a/util/ras-mc-ctl.in +++ b/util/ras-mc-ctl.in @@ -50,6 +50,8 @@ my %dimm_location = (); my %csrow_size = (); my %rank_size = (); my %csrow_ranks = (); +my %dimm_ce_count = (); +my %dimm_ue_count = (); my @layers; my @max_pos; @@ -76,6 +78,7 @@ Usage: $prog [OPTIONS...] --layout Display the memory layout. --summary Presents a summary of the logged errors. --errors Shows the errors stored at the error database. + --error-count Shows the corrected and uncorrected error counts using sysfs. --help This help message. EOF @@ -83,7 +86,7 @@ parse_cmdline(); if ( $conf{opt}{mainboard} || $conf{opt}{print_labels} || $conf{opt}{register_labels} || $conf{opt}{display_memory_layout} - || $conf{opt}{guess_dimm_label}) { + || $conf{opt}{guess_dimm_label} || $conf{opt}{error_count}) { get_mainboard_info(); @@ -105,6 +108,9 @@ if ( $conf{opt}{mainboard} || $conf{opt}{print_labels} if ($conf{opt}{guess_dimm_label}) { guess_dimm_label (); } + if ($conf{opt}{error_count}) { + display_error_count (); + } } if ($conf{opt}{status}) { @@ -134,6 +140,7 @@ sub parse_cmdline $conf{opt}{guess_dimm_label} = 0; $conf{opt}{summary} = 0; $conf{opt}{errors} = 0; + $conf{opt}{error_count} = 0; my $rref = \$conf{opt}{report}; my $mref = \$conf{opt}{mainboard}; @@ -150,7 +157,8 @@ sub parse_cmdline "status" => \$conf{opt}{status}, "layout" => \$conf{opt}{display_memory_layout}, "summary" => \$conf{opt}{summary}, - "errors" => \$conf{opt}{errors} + "errors" => \$conf{opt}{errors}, + "error-count" => \$conf{opt}{error_count} ); usage(1) if !$rc; @@ -284,6 +292,30 @@ sub parse_dimm_nodes $dimm_label_file{$str_loc} = $file; $dimm_location{$str_loc} = $location; + my $count; + + $file =~s/dimm_label/dimm_ce_count/; + if (-e $file) { + open IN, $file; + chomp($count = ); + close IN; + } else { + log_error ("dimm_ce_count not found in sysfs. Old kernel?\n"); + exit -1; + } + $dimm_ce_count{$str_loc} = $count; + + $file =~s/dimm_ce_count/dimm_ue_count/; + if (-e $file) { + open IN, $file; + chomp($count = ); + close IN; + } else { + log_error ("dimm_ue_count not found in sysfs. Old kernel?\n"); + exit -1; + } + $dimm_ue_count{$str_loc} = $count; + return; } } @@ -906,6 +938,45 @@ sub display_memory_layout dimm_display_mem(); } +sub display_error_count +{ + my $sysfs_dir = "/sys/devices/system/edac/mc"; + my $key; + my $max_width = 0; + my %dimm_labels = (); + + find ({wanted => \&parse_dimm_nodes, no_chdir => 1}, $sysfs_dir); + + if (!scalar(keys %dimm_node)) { + log_error ("No DIMMs found in /sys or new sysfs EDAC interface not found.\n"); + exit -1; + } + + foreach $key (keys %dimm_node) { + my $label_width; + + open IN, $dimm_label_file{$key}; + chomp(my $label = ); + close IN; + $label_width = length $label; + + if ($label_width > $max_width) { + $max_width = $label_width; + } + $dimm_labels{$key} = $label; + } + my $string = "Label"; + $string .= " " x ($max_width - length $string); + print($string . "\tCE\tUE\n"); + + foreach $key (keys %dimm_node) { + my $ce_count = $dimm_ce_count{$key}; + my $ue_count = $dimm_ue_count{$key}; + + print("$dimm_labels{$key}\t$ce_count\t$ue_count\n"); + } +} + sub find_prog { my ($file) = @_; -- 1.8.3.1