Merge pull request #1852 from hj-johannes-lee/PR-2024-006

 qat, initcontainer: add enablement of auto_reset

Author: hj-johannes-lee

build/docker/intel-qat-initcontainer.Dockerfile

@@ -64,5 +64,6 @@ LABEL summary='Intel® QAT initcontainer for Kubernetes'
 LABEL description='Intel QAT initcontainer initializes devices'
 COPY --from=builder /install_root /
 COPY demo/qat-init.sh /usr/local/bin/
+COPY demo/qat-autoreset.sh /usr/local/bin/
 WORKDIR /qat-init
-ENTRYPOINT ["/usr/local/bin/qat-init.sh"]
+ENTRYPOINT ["bash", "-c", "/usr/local/bin/qat-init.sh && /usr/local/bin/qat-autoreset.sh"]

build/docker/templates/intel-qat-initcontainer.Dockerfile.in

@@ -22,6 +22,8 @@ COPY --from=builder /install_root /
 
 COPY demo/qat-init.sh /usr/local/bin/
 
+COPY demo/qat-autoreset.sh /usr/local/bin/
+
 WORKDIR /qat-init
 
-ENTRYPOINT ["/usr/local/bin/qat-init.sh"]
+ENTRYPOINT ["bash", "-c", "/usr/local/bin/qat-init.sh && /usr/local/bin/qat-autoreset.sh"]

cmd/qat_plugin/README.md

@@ -131,6 +131,7 @@ In addition to the default configuration, you can add device-specific configurat
 | Device | Possible Configuration | How To Customize | Options | Notes |
 |:-------|:-----------------------|:-----------------|:--------|:------|
 | 4xxx, 401xx, 402xx, 420xx | [cfg_services](https://github.com/torvalds/linux/blob/v6.6-rc5/Documentation/ABI/testing/sysfs-driver-qat) reports the configured services (crypto services or compression services) of the QAT device. | `ServicesEnabled=<value>` | compress:`dc`, crypto:`sym;asym`, <br>crypto+compress:`asym;dc`,<br>crypto+compress:`sym;dc` | 4xxx/401xx/402xx: Linux 6.0+ kernel. 420xx: Linux 6.8+ kernel. |
+| 4xxx, 401xx, 402xx, 420xx | [auto_reset](https://github.com/torvalds/linux/blob/a38297e3fb012ddfa7ce0321a7e5a8daeb1872b6/Documentation/ABI/testing/sysfs-driver-qat#L145) reports the setting of the QAT device's automatic error recovery functionality. | `AutoresetEnabled=<value>` | `on`, `off`, | Linux 6.8+ kernel. |
 
 To create a provisioning `configMap`, run the following command before deploying initcontainer:
 

demo/qat-autoreset.sh

@@ -0,0 +1,45 @@
+#!/usr/bin/env bash
+NODE_NAME="${NODE_NAME:-}"
+ENABLED_QAT_PF_PCIIDS=${ENABLED_QAT_PF_PCIIDS:-37c8 4940 4942 4944 4946}
+DEVS=$(for pf in $ENABLED_QAT_PF_PCIIDS; do lspci -n | grep -e "$pf" | grep -o -e "^\\S*"; done)
+
+AUTORESET_ENABLED="NONE"
+AUTORESET_ENABLED_FOUND="FALSE"
+AUTORESET_OPTIONS_LIST="on off"
+
+check_config() {
+  [ -f "conf/qat.conf" ] && AUTORESET_ENABLED=$(grep "^AutoresetEnabled=" conf/qat.conf | cut -d= -f 2 | grep '\S')
+  [ -f "conf/qat-$NODE_NAME.conf" ] && AUTORESET_ENABLED=$(grep "^AutoresetEnabled=" conf/qat-"$NODE_NAME".conf | cut -d= -f 2 | grep '\S')
+
+  if [ "$AUTORESET_ENABLED" != "NONE" ]; then
+    AUTORESET_ENABLED_FOUND="FALSE"
+    for OPTION in $AUTORESET_OPTIONS_LIST
+    do
+      if [ "$OPTION" = "$AUTORESET_ENABLED" ]; then
+        AUTORESET_ENABLED_FOUND="TRUE"
+        break
+      fi
+    done
+  fi
+}
+
+enable_auto_reset() {
+  if [ "$AUTORESET_ENABLED_FOUND" = "TRUE" ]; then
+    for dev in $DEVS; do
+      devpath="/sys/bus/pci/devices/0000:$dev"
+      autoreset_path="$devpath/qat/auto_reset"
+      if ! test -w "$autoreset_path"; then
+        echo "error: $autoreset_path is not found or not writable. Check if QAT driver module is loaded. Skipping..."
+        exit 1
+      fi
+      if [ "$(cat "$autoreset_path")" = "$AUTORESET_ENABLED" ]; then
+        echo "$devpath's auto reset is already $AUTORESET_ENABLED"
+      else
+        echo "$AUTORESET_ENABLED" > "$autoreset_path" && echo "$devpath's auto reset has been set $AUTORESET_ENABLED"
+      fi
+    done
+  fi
+}
+
+check_config
+enable_auto_reset

demo/qat-init.sh

@@ -12,8 +12,8 @@ SERVICES_ENABLED="NONE"
 SERVICES_ENABLED_FOUND="FALSE"
 
 check_config() {
-  [ -f "conf/qat.conf" ] && SERVICES_ENABLED=$(cut -d= -f 2 conf/qat.conf | grep '\S')
-  [ -f "conf/qat-$NODE_NAME.conf" ] && SERVICES_ENABLED=$(cut -d= -f 2 conf/qat-$NODE_NAME.conf | grep '\S')
+  [ -f "conf/qat.conf" ] && SERVICES_ENABLED=$(grep "^ServicesEnabled=" conf/qat.conf | cut -d= -f 2 | grep '\S')
+  [ -f "conf/qat-$NODE_NAME.conf" ] && SERVICES_ENABLED=$(grep "^ServicesEnabled=" conf/qat-"$NODE_NAME".conf | cut -d= -f 2 | grep '\S')
 
   if [ "$SERVICES_ENABLED" != "NONE" ]; then
     SERVICES_ENABLED_FOUND="FALSE"