Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

ras-mc-ctl: add option to exclude old events from reports #1

Merged
merged 1 commit into from
Apr 21, 2023
Merged

ras-mc-ctl: add option to exclude old events from reports #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
59 changes: 35 additions & 24 deletions util/ras-mc-ctl.in
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,6 +95,7 @@ Usage: $prog [OPTIONS...]
--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.
--since=YYYY-MM-DD Only include events since the date YYYY-MM-DD.
--vendor-errors-summary <platform-id> Presents a summary of the vendor-specific logged errors.
--vendor-errors <platform-id> Shows the vendor-specific errors stored in the error database.
--vendor-errors <platform-id> <module-name> Shows the vendor-specific errors for a specific module stored in the error database.
Expand Down Expand Up @@ -175,6 +176,7 @@ sub parse_cmdline
$conf{opt}{error_count} = 0;
$conf{opt}{vendor_errors_summary} = 0;
$conf{opt}{vendor_errors} = 0;
$conf{opt}{since} = '';
$conf{opt}{vendor_platforms} = 0;

my $rref = \$conf{opt}{report};
Expand All @@ -196,6 +198,7 @@ sub parse_cmdline
"error-count" => \$conf{opt}{error_count},
"vendor-errors-summary" => \$conf{opt}{vendor_errors_summary},
"vendor-errors" => \$conf{opt}{vendor_errors},
"since=s" => \$conf{opt}{since},
"vendor-platforms" => \$conf{opt}{vendor_platforms},
);

Expand All @@ -207,6 +210,14 @@ sub parse_cmdline
log_error ("Only use --delay with --register-labels\n");
exit (1);
}

if ($conf{opt}{since}) {
if ($conf{opt}{since} !~ /^20\d\d-[01]\d-[0-3]\d/) {
log_error ("--since requires a date like yyyy-mm-dd where yyyy is the year, mm the month, and dd the day\n");
exit (1);
}
$conf{opt}{since} = " where timestamp>='$conf{opt}{since}'";
}
}

sub usage
Expand Down Expand Up @@ -1168,7 +1179,7 @@ sub summary
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});

# Memory controller mc_event errors
$query = "select err_type, label, mc, top_layer,middle_layer,lower_layer, count(*) from mc_event group by err_type, label, mc, top_layer, middle_layer, lower_layer";
$query = "select err_type, label, mc, top_layer,middle_layer,lower_layer, count(*) from mc_event$conf{opt}{since} group by err_type, label, mc, top_layer, middle_layer, lower_layer";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_type, $label, $mc, $top, $mid, $low, $count));
Expand All @@ -1185,7 +1196,7 @@ sub summary

# PCIe AER aer_event errors
if ($has_aer == 1) {
$query = "select err_type, err_msg, count(*) from aer_event group by err_type, err_msg";
$query = "select err_type, err_msg, count(*) from aer_event$conf{opt}{since} group by err_type, err_msg";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_type, $msg, $count));
Expand All @@ -1203,7 +1214,7 @@ sub summary

# ARM processor arm_event errors
if ($has_arm == 1) {
$query = "select mpidr, count(*) from arm_event group by mpidr";
$query = "select mpidr, count(*) from arm_event$conf{opt}{since} group by mpidr";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($mpidr, $count));
Expand All @@ -1221,7 +1232,7 @@ sub summary

# extlog errors
if ($has_extlog == 1) {
$query = "select etype, severity, count(*) from extlog_event group by etype, severity";
$query = "select etype, severity, count(*) from extlog_event$conf{opt}{since} group by etype, severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($etype, $severity, $count));
Expand All @@ -1241,7 +1252,7 @@ sub summary

# devlink errors
if ($has_devlink == 1) {
$query = "select dev_name, count(*) from devlink_event group by dev_name";
$query = "select dev_name, count(*) from devlink_event$conf{opt}{since} group by dev_name";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($dev_name, $count));
Expand All @@ -1259,7 +1270,7 @@ sub summary

# Disk errors
if ($has_disk_errors == 1) {
$query = "select dev, count(*) from disk_errors group by dev";
$query = "select dev, count(*) from disk_errors$conf{opt}{since} group by dev";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($dev, $count));
Expand All @@ -1277,7 +1288,7 @@ sub summary

# Memory failure errors
if ($has_mem_failure == 1) {
$query = "select action_result, count(*) from memory_failure_event group by action_result";
$query = "select action_result, count(*) from memory_failure_event$conf{opt}{since} group by action_result";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($action_result, $count));
Expand All @@ -1295,7 +1306,7 @@ sub summary

# MCE mce_record errors
if ($has_mce == 1) {
$query = "select error_msg, count(*) from mce_record group by error_msg";
$query = "select error_msg, count(*) from mce_record$conf{opt}{since} group by error_msg";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($msg, $count));
Expand Down Expand Up @@ -1328,7 +1339,7 @@ sub errors
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbname", "", "", {});

# Memory controller mc_event errors
$query = "select id, timestamp, err_count, err_type, err_msg, label, mc, top_layer,middle_layer,lower_layer, address, grain, syndrome, driver_detail from mc_event order by id";
$query = "select id, timestamp, err_count, err_type, err_msg, label, mc, top_layer,middle_layer,lower_layer, address, grain, syndrome, driver_detail from mc_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
if (!$query_handle) {
log_error ("mc_event table missing from $dbname. Run 'rasdaemon --record'.\n");
Expand All @@ -1349,7 +1360,7 @@ sub errors

# PCIe AER aer_event errors
if ($has_aer == 1) {
$query = "select id, timestamp, dev_name, err_type, err_msg from aer_event order by id";
$query = "select id, timestamp, dev_name, err_type, err_msg from aer_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $time, $devname, $type, $msg));
Expand All @@ -1367,7 +1378,7 @@ sub errors

# ARM processor arm_event errors
if ($has_arm == 1) {
$query = "select id, timestamp, error_count, affinity, mpidr, running_state, psci_state from arm_event order by id";
$query = "select id, timestamp, error_count, affinity, mpidr, running_state, psci_state from arm_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $error_count, $affinity, $mpidr, $r_state, $psci_state));
Expand All @@ -1391,7 +1402,7 @@ sub errors

# Extlog errors
if ($has_extlog == 1) {
$query = "select id, timestamp, etype, severity, address, fru_id, fru_text, cper_data from extlog_event order by id";
$query = "select id, timestamp, etype, severity, address, fru_id, fru_text, cper_data from extlog_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $etype, $severity, $addr, $fru_id, $fru_text, $cper_data));
Expand All @@ -1418,7 +1429,7 @@ sub errors

# devlink errors
if ($has_devlink == 1) {
$query = "select id, timestamp, bus_name, dev_name, driver_name, reporter_name, msg from devlink_event order by id";
$query = "select id, timestamp, bus_name, dev_name, driver_name, reporter_name, msg from devlink_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $bus_name, $dev_name, $driver_name, $reporter_name, $msg));
Expand All @@ -1442,7 +1453,7 @@ sub errors

# Disk errors
if ($has_disk_errors == 1) {
$query = "select id, timestamp, dev, sector, nr_sector, error, rwbs, cmd from disk_errors order by id";
$query = "select id, timestamp, dev, sector, nr_sector, error, rwbs, cmd from disk_errors$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $dev, $sector, $nr_sector, $error, $rwbs, $cmd));
Expand All @@ -1467,7 +1478,7 @@ sub errors

# Memory failure errors
if ($has_mem_failure == 1) {
$query = "select id, timestamp, pfn, page_type, action_result from memory_failure_event order by id";
$query = "select id, timestamp, pfn, page_type, action_result from memory_failure_event$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $pfn, $page_type, $action_result));
Expand All @@ -1486,7 +1497,7 @@ sub errors

# MCE mce_record errors
if ($has_mce == 1) {
$query = "select id, timestamp, mcgcap, mcgstatus, status, addr, misc, ip, tsc, walltime, cpu, cpuid, apicid, socketid, cs, bank, cpuvendor, bank_name, error_msg, mcgstatus_msg, mcistatus_msg, user_action, mc_location from mce_record order by id";
$query = "select id, timestamp, mcgcap, mcgstatus, status, addr, misc, ip, tsc, walltime, cpu, cpuid, apicid, socketid, cs, bank, cpuvendor, bank_name, error_msg, mcgstatus_msg, mcistatus_msg, user_action, mc_location from mce_record$conf{opt}{since} order by id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $time, $mcgcap,$mcgstatus, $status, $addr, $misc, $ip, $tsc, $walltime, $cpu, $cpuid, $apicid, $socketid, $cs, $bank, $cpuvendor, $bank_name, $msg, $mcgstatus_msg, $mcistatus_msg, $user_action, $mc_location));
Expand Down Expand Up @@ -1555,7 +1566,7 @@ sub vendor_errors_summary
# HiSilicon KunPeng9xx errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select err_severity, module_id, count(*) from hip08_oem_type1_event_v2 group by err_severity, module_id";
$query = "select err_severity, module_id, count(*) from hip08_oem_type1_event_v2$conf{opt}{since} group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_severity, $module_id, $count));
Expand All @@ -1573,7 +1584,7 @@ sub vendor_errors_summary
}
$query_handle->finish;

$query = "select err_severity, module_id, count(*) from hip08_oem_type2_event_v2 group by err_severity, module_id";
$query = "select err_severity, module_id, count(*) from hip08_oem_type2_event_v2$conf{opt}{since} group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_severity, $module_id, $count));
Expand All @@ -1591,7 +1602,7 @@ sub vendor_errors_summary
}
$query_handle->finish;

$query = "select err_severity, sub_module_id, count(*) from hip08_pcie_local_event_v2 group by err_severity, sub_module_id";
$query = "select err_severity, sub_module_id, count(*) from hip08_pcie_local_event_v2$conf{opt}{since} group by err_severity, sub_module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_severity, $sub_module_id, $count));
Expand All @@ -1609,7 +1620,7 @@ sub vendor_errors_summary
}
$query_handle->finish;

$query = "select err_severity, module_id, count(*) from hisi_common_section_v2 group by err_severity, module_id";
$query = "select err_severity, module_id, count(*) from hisi_common_section_v2$conf{opt}{since} group by err_severity, module_id";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($err_severity, $module_id, $count));
Expand Down Expand Up @@ -1663,7 +1674,7 @@ sub vendor_errors
# HiSilicon KunPeng9xx errors
if ($platform_id eq HISILICON_KUNPENG_9XX) {
$found_platform = 1;
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type1_event_v2 order by id, module_id, err_severity";
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type1_event_v2$conf{opt}{since} order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $module_id, $sub_module_id, $err_severity, $regs));
Expand All @@ -1688,7 +1699,7 @@ sub vendor_errors
}
$query_handle->finish;

$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type2_event_v2 order by id, module_id, err_severity";
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, module_id, sub_module_id, err_severity, regs_dump from hip08_oem_type2_event_v2$conf{opt}{since} order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $module_id, $sub_module_id, $err_severity, $regs));
Expand All @@ -1713,7 +1724,7 @@ sub vendor_errors
}
$query_handle->finish;

$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, sub_module_id, core_id, port_id, err_severity, err_type, regs_dump from hip08_pcie_local_event_v2 order by id, sub_module_id, err_severity";
$query = "select id, timestamp, version, soc_id, socket_id, nimbus_id, sub_module_id, core_id, port_id, err_severity, err_type, regs_dump from hip08_pcie_local_event_v2$conf{opt}{since} order by id, sub_module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $nimbus_id, $sub_module_id, $core_id, $port_id, $err_severity, $err_type, $regs));
Expand All @@ -1740,7 +1751,7 @@ sub vendor_errors
}
$query_handle->finish;

$query = "select id, timestamp, version, soc_id, socket_id, totem_id, nimbus_id, sub_system_id, module_id, sub_module_id, core_id, port_id, err_type, pcie_info, err_severity, regs_dump from hisi_common_section_v2 order by id, module_id, err_severity";
$query = "select id, timestamp, version, soc_id, socket_id, totem_id, nimbus_id, sub_system_id, module_id, sub_module_id, core_id, port_id, err_type, pcie_info, err_severity, regs_dump from hisi_common_section_v2$conf{opt}{since} order by id, module_id, err_severity";
$query_handle = $dbh->prepare($query);
$query_handle->execute();
$query_handle->bind_columns(\($id, $timestamp, $version, $soc_id, $socket_id, $totem_id, $nimbus_id, $sub_system_id, $module_id, $sub_module_id, $core_id, $port_id, $err_type, $pcie_info, $err_severity, $regs));
Expand Down