Skip to contentSkip to conversation

tests/acceptance: Add a test with the Fedora 31 kernel and initrd

PATCH Superseded Thomas Huth 4 participants 10 messages Dec 17, 2020 – Dec 21, 2020

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 tests/acceptance: Add a test with the Fedora 31 kernel and initrd

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

 tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
 1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +111 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
 # This work is licensed under the terms of the GNU GPL, version 2 or
 # later.  See the COPYING file in the top-level directory.
 
+import os
+import re
+import tempfile
 
 from avocado_qemu import Test
 from avocado_qemu import exec_command_and_wait_for_pattern
 from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
 class S390CCWVirtioMachine(Test):
     KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
         self.vm.command('human-monitor-command', command_line='balloon 128')
         exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                           'MemTotal:         115640 kB')
+
+
+    def test_s390x_fedora(self):
+
+        """
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')
Signature · 1 line
2.27.0
#1 lore raw
Quoted 44 lines · On Thu, 17 Dec 2020 09:53:34 +0100 Thomas Huth <[email protected]> wrote:

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

 tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
 1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +4 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
 # This work is licensed under the terms of the GNU GPL, version 2 or
 # later.  See the COPYING file in the top-level directory.
 
+import os
+import re
+import tempfile
 
 from avocado_qemu import Test
 from avocado_qemu import exec_command_and_wait_for_pattern
 from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
 class S390CCWVirtioMachine(Test):
     KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
         self.vm.command('human-monitor-command', command_line='balloon 128')
         exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                           'MemTotal:         115640 kB')

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 166

Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +10 -0
+
+
+    def test_s390x_fedora(self):
+
+        """
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net

Hm. I'm wondering whether we should add some of these device: tags to the other test in this file as well?

Quoted 31 lines
tests/acceptance/machine_s390_ccw_virtio.py +31 -0
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 207

Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +10 -0
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',

Does the Fedora image have lscss? (As it seems to have lspci.) If yes, you could use it here; I only looked directly at the sysfs entry in the tests above because the environment there is rather minimal.

(We can easily change that on top.)

Quoted 38 lines
tests/acceptance/machine_s390_ccw_virtio.py +38 -0
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 255

Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +10 -0
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')

Do you want to add a device number and check for it here as well? And maybe wait for the CRWs here and below, although that might not be strictly needed?

Quoted 8 lines
tests/acceptance/machine_s390_ccw_virtio.py +8 -0
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')
#2 lore raw
Quoted 52 lines · Cornelia Huck
Quoted 49 lines · On Thu, 17 Dec 2020 09:53:34 +0100 Thomas Huth <[email protected]> wrote:

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

 tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
 1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +9 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
 # This work is licensed under the terms of the GNU GPL, version 2 or
 # later.  See the COPYING file in the top-level directory.
 
+import os
+import re
+import tempfile
 
 from avocado_qemu import Test
 from avocado_qemu import exec_command_and_wait_for_pattern
 from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
 class S390CCWVirtioMachine(Test):
     KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
         self.vm.command('human-monitor-command', command_line='balloon 128')
         exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                           'MemTotal:         115640 kB')
+
+
+    def test_s390x_fedora(self):
+
+        """
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 166

Quoted 5 lines
tests/acceptance/machine_s390_ccw_virtio.py +5 -0
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net

Hm. I'm wondering whether we should add some of these device: tags to the other test in this file as well?

Yes, we could do that (in a separate patch later)

Quoted 37 lines
Quoted 37 lines
tests/acceptance/machine_s390_ccw_virtio.py +37 -0
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 207

Quoted 4 lines
tests/acceptance/machine_s390_ccw_virtio.py +4 -0
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',

Does the Fedora image have lscss? (As it seems to have lspci.) If yes, you could use it here; I only looked directly at the sysfs entry in the tests above because the environment there is rather minimal.

Seems like lscss is available indeed. I'll try to change this...

Quoted 45 lines
Quoted 45 lines
tests/acceptance/machine_s390_ccw_virtio.py +44 -0
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 255

Quoted 4 lines
tests/acceptance/machine_s390_ccw_virtio.py +4 -0
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')

Do you want to add a device number and check for it here as well? And maybe wait for the CRWs here and below, although that might not be strictly needed?

I don't think that this is strictly necessary here, indeed, since the "Accelerator device is ready" already means that the device has been detected correctly.

Quoted 9 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 263

Quoted 8 lines
tests/acceptance/machine_s390_ccw_virtio.py +8 -0
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')

Thomas

#3 lore raw
Quoted 62 lines · On Mon, 21 Dec 2020 14:05:56 +0100 Thomas Huth <[email protected]> wrote:
Quoted 52 lines · Cornelia Huck
Quoted 49 lines · On Thu, 17 Dec 2020 09:53:34 +0100 Thomas Huth <[email protected]> wrote:

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

 tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
 1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +9 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
 # This work is licensed under the terms of the GNU GPL, version 2 or
 # later.  See the COPYING file in the top-level directory.
 
+import os
+import re
+import tempfile
 
 from avocado_qemu import Test
 from avocado_qemu import exec_command_and_wait_for_pattern
 from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
 class S390CCWVirtioMachine(Test):
     KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
         self.vm.command('human-monitor-command', command_line='balloon 128')
         exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                           'MemTotal:         115640 kB')
+
+
+    def test_s390x_fedora(self):
+
+        """
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 166

Quoted 5 lines
tests/acceptance/machine_s390_ccw_virtio.py +5 -0
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net  

Hm. I'm wondering whether we should add some of these device: tags to the other test in this file as well?

Quoted 1 line

Yes, we could do that (in a separate patch later)

Nod.

Quoted 47 lines
Quoted 38 lines
Quoted 38 lines
tests/acceptance/machine_s390_ccw_virtio.py +37 -0
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 207

Quoted 4 lines
tests/acceptance/machine_s390_ccw_virtio.py +4 -0
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',  

Does the Fedora image have lscss? (As it seems to have lspci.) If yes, you could use it here; I only looked directly at the sysfs entry in the tests above because the environment there is rather minimal.

Quoted 1 line

Seems like lscss is available indeed. I'll try to change this...

I think you want to check for the devno only in that case (i.e. not for the subchannel number... we cannot guarantee that QEMU's algorithm will assign the same subchannel number in all eternity.)

Quoted 55 lines
Quoted 46 lines
Quoted 46 lines
tests/acceptance/machine_s390_ccw_virtio.py +44 -0
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()
Quoted 8 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 255

Quoted 4 lines
tests/acceptance/machine_s390_ccw_virtio.py +4 -0
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')  

Do you want to add a device number and check for it here as well? And maybe wait for the CRWs here and below, although that might not be strictly needed?

Quoted 3 lines

I don't think that this is strictly necessary here, indeed, since the "Accelerator device is ready" already means that the device has been detected correctly.

Yes. It might only ease debugging (e.g. the CRW was received correctly, but for some reason the device was not properly set up.) Not sure if it is worth the extra hassle.

Quoted 12 lines
Quoted 10 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 263

Quoted 9 lines
tests/acceptance/machine_s390_ccw_virtio.py +8 -0
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')  

Thomas

Reviewed-by on v2 1/1
#4 lore raw
In reply to Thomas Huth

Hi,

Quoted 26 lines · Thomas Huth

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

  tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
  1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +0 -0
index abe25a08f0..a9a0f8a8e2 100644

tests/acceptance/machine_s390_ccw_virtio.py line 13

Quoted 5 lines
tests/acceptance/machine_s390_ccw_virtio.py +2 -0
@@ -9,10 +9,14 @@
  # This work is licensed under the terms of the GNU GPL, version 2 or
  # later.  See the COPYING file in the top-level directory.
 
+import os
+import re

're' is not used.

Quoted 95 lines
tests/acceptance/machine_s390_ccw_virtio.py +84 -0
+import tempfile
 
  from avocado_qemu import Test
  from avocado_qemu import exec_command_and_wait_for_pattern
  from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
  class S390CCWVirtioMachine(Test):
      KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
          self.vm.command('human-monitor-command', command_line='balloon 128')
          exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                            'MemTotal:         115640 kB')
+
+
+    def test_s390x_fedora(self):
+
+        """
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 248

Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +10 -0
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close

In case you don't want to deal with 2 tmp files (and manually close them), here goes a suggestion:

        with tempfile.NamedTemporaryFile(suffix='.ppm', prefix='qemu-scrdump-') as ppmfile:
            self.vm.command('screendump', filename=ppmfile.name) ppmfile.seek(0)
            line = ppmfile.readline()
            self.assertEqual(line, b"P6\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"1024 768\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"255\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"The quick fox jumps over a lazy dog\n")

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 250

Quoted 2 lines
tests/acceptance/machine_s390_ccw_virtio.py +2 -0
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted

Yet another suggestion. The other day Cornelia asked how to make it easier to debug those asserts since it's being tested various conditions on a single test method (for the sake of speed). One simple thing would be print 'info' to the test logs, e.g.:

self.log.info("Test hot-plug virtio-crypto device")

Quoted 13 lines
tests/acceptance/machine_s390_ccw_virtio.py +13 -0
+        self.clear_guest_dmesg()
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')
#5 lore raw
Quoted 24 lines · Wainer dos Santos Moschetta

Hi,

Quoted 21 lines · Thomas Huth

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
  Based-on: [email protected]

  v2:
  - Uncompress the initrd on the host already to speed up the test
  - Disable plymouth to seepd up the test
  - Check for device numbers
  - Added virtio-crypto test
  - Added ":avocado: tags=device:" lines

  tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++

  1 file changed, 111 insertions(+)
Quoted 11 lines
Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +0 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
   # This work is licensed under the terms of the GNU GPL, version 2 or
   # later.  See the COPYING file in the top-level directory.

  +import os
+import re

're' is not used.

Right, leftover from a first try to parse the screendump file. Will remove it.

Quoted 116 lines
Quoted 104 lines

+import tempfile
    from avocado_qemu import Test
  from avocado_qemu import exec_command_and_wait_for_pattern from avocado_qemu import wait_for_console_pattern +from avocado.utils import archive
    class S390CCWVirtioMachine(Test):
      KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '

tests/acceptance/machine_s390_ccw_virtio.py +0 -0
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):

          self.vm.command('human-monitor-command', command_line='balloon 128')
          exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo', 'MemTotal:         115640 kB')

tests/acceptance/machine_s390_ccw_virtio.py +23 -0
+
+
+    def test_s390x_fedora(self):
+
+        """
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url,

asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img') +        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 ' +                              'rd.plymouth=0 plymouth.enable=0 rd.rescue') +        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa', +                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876', +                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432') +        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode') +
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000') +        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address', +                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/', +                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/', +                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo', +                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo', +                             'Extended Name:   Some Guest Name') +        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo', +                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa') +
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when +        # writing to the framebuffer. Since the PPM is uncompressed, we then +        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected. +        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do ' +            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-') +        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()

Quoted 8 lines

+        self.assertEqual(line, "P6\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n") +        ppmfile.close

In case you don't want to deal with 2 tmp files (and manually close them), here goes a suggestion:

Quoted 12 lines

        with tempfile.NamedTemporaryFile(suffix='.ppm', prefix='qemu-scrdump-') as ppmfile:
            self.vm.command('screendump', filename=ppmfile.name) ppmfile.seek(0)
            line = ppmfile.readline()
            self.assertEqual(line, b"P6\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"1024 768\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"255\n")
            line = ppmfile.readline()
            self.assertEqual(line, b"The quick fox jumps over a lazy dog\n")

Nice, seems to work, thanks, I'll use that instead.

Quoted 9 lines
Quoted 2 lines

+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted

Yet another suggestion. The other day Cornelia asked how to make it easier to debug those asserts since it's being tested various conditions on a single test method (for the sake of speed). One simple thing would be print 'info' to the test logs, e.g.:

self.log.info("Test hot-plug virtio-crypto device")

Good idea, I'll add some of these log statements!

Quoted 1 line

Thanks!

Thomas

Tested-by Reviewed-by on v2 1/1
#6 lore raw
In reply to Thomas Huth
Quoted 129 lines · Thomas Huth

This initrd contains a virtio-net and a virtio-gpu kernel module, so we can check that we can set a MAC address for the network device and whether we can hot-plug and -unplug a virtio-crypto device. But the most interesting part is maybe that we can also successfully write some stuff into the emulated framebuffer of the virtio-gpu device and make sure that we can read back that data from a screenshot.

---
Based-on: [email protected]

v2:
- Uncompress the initrd on the host already to speed up the test
- Disable plymouth to seepd up the test
- Check for device numbers
- Added virtio-crypto test
- Added ":avocado: tags=device:" lines

  tests/acceptance/machine_s390_ccw_virtio.py | 111 ++++++++++++++++++++
  1 file changed, 111 insertions(+)
tests/acceptance/machine_s390_ccw_virtio.py +89 -0
index abe25a08f0..a9a0f8a8e2 100644
@@ -9,10 +9,14 @@
  # This work is licensed under the terms of the GNU GPL, version 2 or
  # later.  See the COPYING file in the top-level directory.
 
+import os
+import re
+import tempfile
 
  from avocado_qemu import Test
  from avocado_qemu import exec_command_and_wait_for_pattern
  from avocado_qemu import wait_for_console_pattern
+from avocado.utils import archive
 
  class S390CCWVirtioMachine(Test):
      KERNEL_COMMON_COMMAND_LINE = 'printk.time=0 '
@@ -150,3 +154,110 @@ class S390CCWVirtioMachine(Test):
          self.vm.command('human-monitor-command', command_line='balloon 128')
          exec_command_and_wait_for_pattern(self, 'head -n 1 /proc/meminfo',
                                            'MemTotal:         115640 kB')
+
+
+    def test_s390x_fedora(self):
+
+        """
+        :avocado: tags=arch:s390x
+        :avocado: tags=machine:s390-ccw-virtio
+        :avocado: tags=device:virtio-gpu
+        :avocado: tags=device:virtio-crypto
+        :avocado: tags=device:virtio-net
+        """
+
+        kernel_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/kernel.img')
+        kernel_hash = 'b93d1efcafcf29c1673a4ce371a1f8b43941cfeb'
+        kernel_path = self.fetch_asset(kernel_url, asset_hash=kernel_hash)
+
+        initrd_url = ('https://archives.fedoraproject.org/pub/archive'
+                      '/fedora-secondary/releases/31/Server/s390x/os'
+                      '/images/initrd.img')
+        initrd_hash = '3de45d411df5624b8d8ef21cd0b44419ab59b12f'
+        initrd_path_xz = self.fetch_asset(initrd_url, asset_hash=initrd_hash)
+        initrd_path = os.path.join(self.workdir, 'initrd-raw.img')
+        archive.lzma_uncompress(initrd_path_xz, initrd_path)
+
+        self.vm.set_console()
+        kernel_command_line = (self.KERNEL_COMMON_COMMAND_LINE + ' audit=0 '
+                              'rd.plymouth=0 plymouth.enable=0 rd.rescue')
+        self.vm.add_args('-nographic',
+                         '-smp', '4',
+                         '-m', '512',
+                         '-name', 'Some Guest Name',
+                         '-uuid', '30de4fd9-b4d5-409e-86a5-09b387f70bfa',
+                         '-kernel', kernel_path,
+                         '-initrd', initrd_path,
+                         '-append', kernel_command_line,
+                         '-device', 'zpci,uid=7,target=n',
+                         '-device', 'virtio-net-pci,id=n,mac=02:ca:fe:fa:ce:12',
+                         '-device', 'virtio-rng-ccw,devno=fe.1.9876',
+                         '-device', 'virtio-gpu-ccw,devno=fe.2.5432')
+        self.vm.launch()
+        self.wait_for_console_pattern('Entering emergency mode')
+
+        # Some tests to see whether the CLI options have been considered:
+        exec_command_and_wait_for_pattern(self, 'lspci',
+                             '0007:00:00.0 Class 0200: Device 1af4:1000')
+        exec_command_and_wait_for_pattern(self,
+                             'cat /sys/class/net/enP7p0s0/address',
+                             '02:ca:fe:fa:ce:12')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.1.9876')
+        exec_command_and_wait_for_pattern(self, 'ls /sys/bus/ccw/devices/',
+                             '0.2.5432')
+        exec_command_and_wait_for_pattern(self, 'cat /proc/cpuinfo',
+                             'processors    : 4')
+        exec_command_and_wait_for_pattern(self, 'grep MemTotal /proc/meminfo',
+                             'MemTotal:         499848 kB')
+        exec_command_and_wait_for_pattern(self, 'grep Name /proc/sysinfo',
+                             'Extended Name:   Some Guest Name')
+        exec_command_and_wait_for_pattern(self, 'grep UUID /proc/sysinfo',
+                             '30de4fd9-b4d5-409e-86a5-09b387f70bfa')
+
+        # Disable blinking cursor, then write some stuff into the framebuffer.
+        # QEMU's PPM screendumps contain uncompressed 24-bit values, while the
+        # framebuffer uses 32-bit, so we pad our text with some spaces when
+        # writing to the framebuffer. Since the PPM is uncompressed, we then
+        # can simply read the written "magic bytes" back from the PPM file to
+        # check whether the framebuffer is working as expected.
+        exec_command_and_wait_for_pattern(self,
+            'echo -e "\e[?25l" > /dev/tty0', ':/#')
+        exec_command_and_wait_for_pattern(self, 'for ((i=0;i<250;i++)); do '
+            'echo " The  qu ick  fo x j ump s o ver  a  laz y d og" >> fox.txt;'
+            'done',
+            ':/#')
+        exec_command_and_wait_for_pattern(self,
+            'dd if=fox.txt of=/dev/fb0 bs=1000 oflag=sync,nocache ; rm fox.txt',
+            '12+0 records out')
+        tmpfile = tempfile.NamedTemporaryFile(suffix='.ppm',
+                                              prefix='qemu-scrdump-')
+        self.vm.command('screendump', filename=tmpfile.name)
+        ppmfile = open(tmpfile.name, "r")
+        tmpfile.close
+        line = ppmfile.readline()
+        self.assertEqual(line, "P6\n")

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 251

Quoted 10 lines
tests/acceptance/machine_s390_ccw_virtio.py +10 -0
+        line = ppmfile.readline()
+        self.assertEqual(line, "1024 768\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "255\n")
+        line = ppmfile.readline()
+        self.assertEqual(line, "The quick fox jumps over a lazy dog\n")
+        ppmfile.close
+
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()

Your previous patch "[PATCH 1/3] tests/acceptance: Extract the code to clear dmesg and wait for CRW reports" defined the method as "clear_guests_dmesg". After fixing this in the code:

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 263

Quoted 13 lines
tests/acceptance/machine_s390_ccw_virtio.py +12 -0
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')
#7 lore raw
Quoted 8 lines · On Fri, 18 Dec 2020 15:23:32 -0300 Willian Rampazzo <[email protected]> wrote:

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 251

Quoted 2 lines · Thomas Huth
tests/acceptance/machine_s390_ccw_virtio.py +2 -0
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()  

Your previous patch "[PATCH 1/3] tests/acceptance: Extract the code to clear dmesg and wait for CRW reports" defined the method as "clear_guests_dmesg". After fixing this in the code:

Hm, I see clear_guest_dmesg in the code I have on my s390-next branch?

Quoted 19 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 263

Quoted 13 lines
tests/acceptance/machine_s390_ccw_virtio.py +12 -0
+        self.vm.command('object-add', qom_type='cryptodev-backend-builtin',
+                        id='cbe0')
+        self.vm.command('device_add', driver='virtio-crypto-ccw', id='crypdev0',
+                        cryptodev='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Accelerator.device) ; do'
+                        ' sleep 1 ; done', 'Accelerator device is ready')
+        self.vm.command('device_del', id='crypdev0')
+        self.vm.command('object-del', id='cbe0')
+        exec_command_and_wait_for_pattern(self,
+                        'while ! (dmesg -c | grep Start.virtcrypto_remove) ; do'
+                        ' sleep 1 ; done', 'Start virtcrypto_remove.')
#8 lore raw
Quoted 5 lines · Cornelia Huck
Quoted 1 line · On Fri, 18 Dec 2020 15:23:32 -0300 Willian Rampazzo <[email protected]> wrote:

On 12/17/20 5:53 AM, Thomas Huth wrote:

Quoted 8 lines
Quoted 6 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 251

Quoted 2 lines
tests/acceptance/machine_s390_ccw_virtio.py +2 -0
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()  

Your previous patch "[PATCH 1/3] tests/acceptance: Extract the code to clear dmesg and wait for CRW reports" defined the method as "clear_guests_dmesg". After fixing this in the code:

Hm, I see clear_guest_dmesg in the code I have on my s390-next branch?

I likely used the wrong "Based-on" line in the header, sorry.

Thomas

#9 lore raw
In reply to Cornelia Huck
Quoted 6 lines · Cornelia Huck
Quoted 1 line · On Fri, 18 Dec 2020 15:23:32 -0300 Willian Rampazzo <[email protected]> wrote:

On 12/17/20 5:53 AM, Thomas Huth wrote:

Quoted 8 lines
Quoted 6 lines

tests/acceptance/machine_s390_ccw_virtio.py class S390CCWVirtioMachine(Test): · line 251

Quoted 2 lines
tests/acceptance/machine_s390_ccw_virtio.py +2 -0
+        # Hot-plug a virtio-crypto device and see whether it gets accepted
+        self.clear_guest_dmesg()

Your previous patch "[PATCH 1/3] tests/acceptance: Extract the code to clear dmesg and wait for CRW reports" defined the method as "clear_guests_dmesg". After fixing this in the code:

Hm, I see clear_guest_dmesg in the code I have on my s390-next branch?

Ah, I had his v0 locally
(https://patchew.org/QEMU/[email protected]/[email protected]/).

The v1 is, indeed, clear_guest_dmesg, so, we should be good to go!