guest-test framework improvement and add 6 TDX guest BAT TCs

guest-test/README.md

@@ -17,7 +17,7 @@ No multi-VMs test scenarios covered/supported.
 
 In any case of issue debugging, please refer to above test log with VM QEMU config info and launch VM and debug issues manually running test scripts/binaries.
 
-A prepared Guest OS Image (qcow2 or raw image format) is requried with preset root account and password, several values/parameters in qemu.config.json highly depend on Guest OS Image, please accomodate accordingly.
+A prepared Guest OS Image (qcow2 or raw image format) is requried with preset root account and password, several values/parameters in common.json highly depend on Guest OS Image, please accomodate accordingly.
 
 ## Usage
 ### common.json and qemu.config.json description
@@ -46,11 +46,11 @@ group "common" includes all configurable values to be passed to group "vm", "tdx
 group "vm" includes all legacy vm launch qemu config options, which can be used to launch legacy vm standalone or as base part to launch tdx vm
 
 group "vm" 2nd-level-keys could be bypassed if not provided (file not exists)
-"cfg_var_6" & "cfg_var_10"
+"cfg_var_6", "cfg_var_10" could be bypassed separately, "cfg_var_5" & "cfg_var_9" could be bypassed in pair
 
 group "tdx" includes tdx vm specific qemu config options, which is used to launch tdx vm (with group "vm" as base) or as a based part to launch tdxio vm
 
-group "tdxio" includes tdxio vm specific qemu config options, which is used to launch tdxio vm (with group "vm" + "tdx" as base)
+group "tdxio" includes tdxio/tdx-connect vm specific qemu config options, which is used to launch tdxio/tdx-connect vm (with group "vm" + "tdx" as base)
 
 note about qemu.config.json:
 - no changes allowed on 1st-level-keys hierarchy
@@ -64,14 +64,17 @@ main test entrance, with following key args can be passed to override the values
   - `-m` $MEM memory size in GB
   - `-d` $DEBUG debug on/off
   - `-t` $VM_TYPE vm_type legacy/tdx/tdxio
+  - `-f` $FEATURE feature sub-folder for new feature vm test extension
   - `-x` $TESTCASE testcase pass to test_executor
   - `-c` $CMDLINE guest kernel extra commandline
   - `-p` $PMU guest pmu off/on
   - `-g` $GCOV code coverage test mode off/on
+  - `-i` $JSON_C file path under guest-test to standalone common.json file
+  - `-j` $JSON_Q file path under guest-test to standalone qemu.config.json file
 
-above key args will be recorded in a fresh new test_params.py for further import/source purpose accross scripts
+above key args will be recorded in a fresh new test_params.py for further import/source purpose accross python and shell scripts
 
-by enter each test, qemu_get_config.py will be called to get following pre-set parameters from qemu.config.json
+by enter each test, qemu_get_config.py will be called to get following pre-set parameters from common.json
   - $KERNEL_IMG values passed by group "common" key "kernel_img"
   - $INITRD_IMG values passed by group "common" key "initrd_img"
   - $BIOS_IMG values passed by group "common" key "bios_img"
@@ -82,9 +85,9 @@ by enter each test, qemu_get_config.py will be called to get following pre-set p
   - $SSHPASS values passed by group "common" key "guest_root_passwd"
 
 call guest.qemu_runner.sh and wait for $BOOT_PATTERN (shows VM boot up completed and ready for login) during VM boot
-  - $BOOT_PATTERN selected based on following CentOS Stream 8/9 boot log example: "*Kernel*on*x86_64*"
+  - $BOOT_PATTERN is selected based on following CentOS Stream 8/9 boot log example: "*Kernel*on*an*x86_64*"
 
-  boot log quoted:
+  boot log example quoted:
   ```
     CentOS Stream 9
     Kernel 6.5.0-rc5-next-20230809-next-20230809 on an x86_64
@@ -100,12 +103,12 @@ if $ERR_STRx found, handle the error info accordingly (err_handlers)
 
 no matter what, in the end, pkill VM process to avoid any potential test step failures above
 
-Note: bydeault, $GCOV is off, if $GCOV is on, above VM life-cycles management logic will be bypassed to keep VM process alive for gcov code coverage data collection
+Note: by deault, $GCOV is off, if $GCOV is on, above VM life-cycles management logic will be bypassed to keep VM process alive for gcov code coverage data collection
 
 ### guest.qemu_runner description
 VM boot engine, with parames exported from qemu_get_config.py and test scenario config sourced from test_params.py
 
-before VM boot, for $VM_TYPE tdx or tdxio, tdx_pre_check will be called to make sure basic environment is ready for TDX/TDXIO launching
+before VM boot, for $VM_TYPE tdx or tdxio, tdx_pre_check and tdx_module_check will be called to make sure basic environment is ready for TDX/TDXIO launching
 
 VM boot is triggered by qemu_runner.py based on $VM_TYPE, with proper qemu config options applied
 
@@ -117,8 +120,10 @@ guest VM test execution basic framework implemented in guerst.test_executor.sh,
   - guest_test_close, function based on sshpass to close VM
 
 ## How to add new feature test
-as described above, if simply add new TCs to run based on current qemu.config.json format, just need to implement it in test_executor with new $TESTCASE branch,
+as described above, if simply add new TCs to run based on current common.json and qemu.config.json format, just need to implement it in $FEATURE.test_executor.sh under guest-test/$FEATURE subfolder,
+please leverage reference code "common variables example" and "common works example" in tdx/tdx.test_executor.sh, further implement "$FEATURE specific Functions" and "$FEATURE specific code path",
+common functions of guest.test_executor.sh should be good enough to prepare/run/close new $FEATURE specific tests implemented in $FEATURE.test_executor.sh;
 
-common functions of test_executor should be good enough to prepare/run/close new $TESTCASE
+it's allowed to customize and pass $FEATURE standalone common.json and qemu.config.json, please refer to ### common.json and qemu.config.json description and ### guest.test_launcher description for quick rule reference
 
-if qemu.config.json format will be revised due to feature changes on QEMU implementation, please update qemu.config.json and qemu_get_config.py accordingly
+furthermore, if common.json and qemu.config.json format will be revised due to feature changes on QEMU implementation, please update qemu_get_config.py accordingly

guest-test/guest.qemu_runner.sh

@@ -16,6 +16,49 @@ SCRIPT_DIR="$( cd "$( dirname "$0" )" && pwd )"
 echo "$SCRIPT_DIR"
 
 ###################### Functions ######################
+# function to remove 0x and prefix useless 0 for hex data
+version_check() {
+  local version_raw=$1
+  if [[ "$version_raw" = "0x00000000" ]]; then
+    local version=0x0
+  else
+    local version_hex
+    version_hex=${version_raw#0x}
+    version_hex=${version_hex##+(0)}
+    local version=$((16#$version_hex))
+  fi
+  echo "$version"
+}
+
+# function to do tdx module status and version check
+tdx_module_check() {
+  local tdx_module_path="/sys/firmware/tdx/tdx_module/"
+  local status
+  local attributes
+  local vendor_id
+  local major_version
+  local minor_version
+  local build_date
+  local build_num
+  status=$(cat "$tdx_module_path"status)
+  [[ "$status" = "initialized" ]] || \
+  die "TDX module not initialized correctly, \
+  please check host kernel tdx enabling setup."
+  test_print_trc "TDX module initialized correctly"
+  attributes=$(cat "$tdx_module_path"attributes)
+  attributes=$(version_check "$attributes")
+  vendor_id=$(cat "$tdx_module_path"vendor_id)
+  major_version=$(cat "$tdx_module_path"major_version)
+  major_version=$(version_check "$major_version")
+  minor_version=$(cat "$tdx_module_path"minor_version)
+  minor_version=$(version_check "$minor_version")
+  build_date=$(cat "$tdx_module_path"build_date)
+  build_num=$(cat "$tdx_module_path"build_num)
+  build_num=$(version_check "$build_num")
+  test_print_trc "TDX module: attributes $attributes, vendor_id $vendor_id, \
+  major_version $major_version, minor_version $minor_version, \
+  build_date $build_date, build_num $build_num"
+}
 
 # function to do TDX/TDXIO VM launching basic pre-check
 # list all the variables value
@@ -81,8 +124,9 @@ source "$SCRIPT_DIR"/test_params.py
 # do basic pre-check for TDX/TDXIO VM launching
 if [[ $VM_TYPE == "tdx" ]] || [[ $VM_TYPE == "tdxio" ]]; then
   tdx_pre_check
+  tdx_module_check
 fi
 
 # launch VM by qemu via qemu_runner.py
 test_print_trc "qemu_runner start to launch $VM_TYPE VM"
-python3 "$SCRIPT_DIR"/qemu_runner.py
+python3 "$SCRIPT_DIR"/qemu_runner.py

guest-test/guest.test_executor.sh

@@ -46,17 +46,16 @@ EOF
     source $GUEST_TEST_DIR/common.sh
     cd $GUEST_TEST_DIR
     cd $1
-    dnf list installed gcc || dnf install -y gcc || \
-    { die "Failed to install gcc in guest os"; return 1; }
-    dnf list installed glibc-static || dnf install -y glibc-static || \
-    { die "Failed to install glibc-static in guest os"; return 1; }
-    make || { die "Failed to compile source code $1"; return 1; }
+    dnf list installed gcc || dnf install -y gcc > /dev/null 2>&1
+    dnf list installed glibc-static || dnf install -y glibc-static > /dev/null 2>&1
+    apt list installed gcc | grep "installed" || apt-get install -y gcc > /dev/null 2>&1
+    apt list installed libc6-dev | grep "installed" || apt-get install -y libc6-dev > /dev/null 2>&1
+    make || die "Failed to compile source code $1"
     if [ -f $2 ]; then
       chmod a+x $2
       cp $2 $GUEST_TEST_DIR
     else
       die "Can't find test binary $2"
-      return 1
     fi
 EOF
   ERR_NUM=$?
@@ -90,14 +89,18 @@ guest_test_close() {
   sshpass -e ssh -p "$PORT" -o StrictHostKeyChecking=no root@localhost << EOF
     source $GUEST_TEST_DIR/common.sh
     test_print_trc "guest test complete, close VM now"
-    systemctl reboot --reboot-argument=now
+    if [[ "$VM_TYPE" = "legacy" ]]; then
+      shutdown now
+    else
+      systemctl reboot --reboot-argument=now
+    fi
 EOF
   ERR_NUM=$?
   if [ $ERR_NUM -eq 0 ]; then
     test_print_trc "Guest VM closed properly after test"
     return 0
   else
-    return 1
+    die "Failed on close guest VM"
   fi
 }
 
@@ -114,5 +117,5 @@ if [ "${BASH_SOURCE[0]}" -ef "$0" ]; then
   cd "$SCRIPT_DIR" || die "fail to switch to $SCRIPT_DIR"
   # select specific "$FEATURE.test_executor.sh" by $FEATURE
   "$FEATURE"/"$FEATURE".test_executor.sh || \
-  { die "Failed on $TESTCASE of $FEATURE"; return 1; }
+  die "Failed on $TESTCASE of $FEATURE"
 fi

guest-test/guest.test_launcher.sh

@@ -77,9 +77,14 @@ guest_kernel_check() {
 EOF
 }
 
-guest_kernel_reboot() {
+guest_kernel_shutdown() {
   sshpass -e ssh -p "$PORT" -o StrictHostKeyChecking=no root@localhost << EOF
-    systemctl reboot --reboot-argument=now
+    echo "$VM_TYPE VM guest kernel shutdown now"
+    if [[ "$VM_TYPE" = "legacy" ]]; then
+      shutdown now
+    else
+      systemctl reboot --reboot-argument=now
+    fi
 EOF
 }
 
@@ -251,7 +256,9 @@ while read -r line; do
   fi
 done < <(
   if [ "$GCOV" == "off" ]; then
-    timeout "$TIMEOUT" ./guest.qemu_runner.sh
+    # keep timeout process run foreground for direct script execution correctness
+    # handle timeout effect case SIGTERM impact on terminal no type-in prompt issue
+    timeout --foreground "$TIMEOUT" ./guest.qemu_runner.sh || reset
   else
     test_print_trc "${VM_TYPE}vm_$PORT keep alive for gcov data collection" && ./guest.qemu_runner.sh
   fi
@@ -308,17 +315,17 @@ if ! guest_kernel_check; then
     pkill "${VM_TYPE}vm_$PORT"
     die "$VM_TYPE VM test seems fail at beginning, please check test log"
   fi
-# guest_kernel_kernel function zero return value shows TDVM is still accessible handling
-# handling: no matter why it's still accessible, close it by guest_kernel_reboot function
+# guest_kernel_check function zero return value shows TDVM is still accessible handling
+# handling: no matter why it's still accessible, close it by guest_kernel_shutdown function
 elif [ "$GCOV" == "off" ]; then
-  if ! guest_kernel_reboot; then
+  if guest_kernel_shutdown; then
     test_print_trc "$VM_TYPE VM is still up"
     test_print_trc "time: $SECONDS"
     test_print_trc "SSHPASS: $SSHPASS"
     test_print_trc "PORT: $PORT"
     test_print_trc "$VM_TYPE VM closed"
     # must die here since TDVM should be closed and not accessible if test complete all correctly
-    # else it's due to test die before reaching final close point td_test_close function
+    # else it's due to test die before reaching final close point guest_test_close function
     die "$VM_TYPE VM test fail, please check test log"
   fi
 else # [ $GCOV == "on" ] || [ guest_kernel_check return 0 ]

guest-test/qemu.config.json

@@ -21,7 +21,7 @@
     "cfg_var_6": "-initrd $INITRD_IMG ",
     "cfg_var_7": "-netdev user,id=mynet0,hostfwd=tcp::$PORT-:22 ",
     "cfg_var_8": "-drive file=$GUEST_IMG,if=virtio,format=$IMG_FORMAT ",
-    "cfg_var_9": "-append \"root=/dev/vda3 ro console=hvc0 earlyprintk=ttyS0 ignore_loglevel debug earlyprintk l1tf=off initcall_debug log_buf_len=200M nokaslr tsc=reliable efi=debug mce=off efi=debug $CMDLINE\" ",
+    "cfg_var_9": "-append \"root=/dev/vda3 ro console=hvc0 earlyprintk=ttyS0 ignore_loglevel debug earlyprintk l1tf=off initcall_debug log_buf_len=200M nokaslr tsc=reliable efi=debug mce=off $CMDLINE\" ",
     "cfg_var_10": "-bios $BIOS_IMG ",
     "cfg_var_11": "-monitor telnet:127.0.0.1:$PORT_TEL,server,nowait "
   },

guest-test/qemu_runner.py

@@ -17,6 +17,7 @@
 import subprocess as sp
 from qemu_get_config import *
 from test_params import *
+from signal import signal, SIGPIPE, SIG_DFL
 
 ###################### Variables ######################
 # all variables imported from qemu_get_config and test_params
@@ -25,6 +26,8 @@
 # all work done in qemu_get_config.py
 
 ###################### Do Works ######################
+# igore SIGPIPE in case of broken pipe errno 32 case
+signal(SIGPIPE, SIG_DFL)
 # launch legacy common vm based on vm_type config
 if vm_type == "legacy":
   command = '{} {}'.format(qemu_img, vm_cfg)
@@ -38,4 +41,4 @@
 # launch tdxio vm based on vm_type config
 if vm_type == "tdxio":
   command = '{} {} {} {}'.format(qemu_img, vm_cfg, tdx_cfg, tdxio_cfg)
-  sp.run(command, shell=True)
+  sp.run(command, shell=True)