ltp_tpci: testcase 12 related to virtio releasing and reassigning resource leads to guest hung

ENV: kernel 6.0.7-301.fc37.x86_64, qemu-7.0.0-9.fc37 VM: virsh xml: machine='pc-i440fx-6.2'

<os>
    <type arch='x86_64' machine='pc-i440fx-6.2'>hvm</type>
    <boot dev='hd'/>
</os>
<devices>
    <disk type='file' device='disk'>
      <driver name='qemu' type='qcow2' iothread='1'/>
      <source file='replace your vm image path'/>
      <backingStore/>
      <target dev='vda' bus='virtio'/>
      <boot order='1'/>
      <alias name='virtio-disk0'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x08' function='0x0'/>
    </disk>
......

I executed the testcase in the virtualization environment of fedora37 qemu 7.0 (I440FX): ./runltp -f kernel_misc -s tpci after run the testcase 12, leads to guest hung(io hung), the test releases virtio device resource and then reassign. [ 59.043221][ T2431] ltp_tpci: test-case 12 [ 59.043845][ T2431] ltp_tpci: assign resources [ 59.044499][ T2431] ltp_tpci: assign resource #0 [ 59.045176][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 262401, start 0xc000, end 0xc07f [ 59.046456][ T2431] ltp_tpci: assign resource #1 [ 59.047112][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 262656, start 0xfebd3000, end 0xfebd3fff [ 59.048517][ T2431] ltp_tpci: assign resource #2 [ 59.049369][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 0, start 0x0, end 0x0 [ 59.050593][ T2431] ltp_tpci: assign resource #3 [ 59.051316][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 0, start 0x0, end 0x0 [ 59.052413][ T2431] ltp_tpci: assign resource #4 [ 59.053103][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 1319436, start 0xfe008000, end 0xfe00bfff [ 59.054522][ T2431] virtio-pci 0000:00:06.0: BAR 4: releasing [mem 0xfe008000-0xfe00bfff 64bit pref] [ 59.087851][ T2431] virtio-pci 0000:00:06.0: BAR 4: assigned [mem 0x440008000-0x44000bfff 64bit pref] [ 59.090712][ T2431] ltp_tpci: assign resource to '4', ret '0' [ 59.091576][ T2431] ltp_tpci: assign resource #5 [ 59.092315][ T2431] ltp_tpci: name = (null), flags = 0, start 0x0, end 0x0 [ 59.093312][ T2431] ltp_tpci: assign resource #6 [ 59.094047][ T2431] ltp_tpci: name = 0000:00:06.0, flags = 0, start 0x0, end 0x0

virtio-pci BAR 4： before test: virtio-pci 0000:00:06.0: BAR 4: releasing [mem 0xfe008000-0xfe00bfff 64bit pref] after test: virtio-pci 0000:00:06.0: BAR 4: assigned [mem 0x440008000-0x44000bfff 64bit pref]

https://lists.linux.it/pipermail/ltp/2021-June/023296.html I found that someone had reported a similar problem to me before [https://lists.linux.it/pipermail/ltp/2021-June/023296.html] The modified method of react is：If I use this modification method it can pass

 .../device-drivers/pci/tpci_kernel/ltp_tpci.c     | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c b/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c
index f2d4a4ba497c..41462d4ead6d 100644
--- a/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c
+++ b/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c
@@ -108,8 +108,9 @@ static int probe_pci_dev(unsigned int bus, unsigned int slot)
    if (!dev || !dev->driver)
        return -ENODEV;

-   prk_info("found pci_dev '%s', bus %u, devfn %u",
-       pci_name(dev), bus, slot);
+   prk_info("found pci_dev '%s', driver '%s', bus %u, devfn %u",
+       pci_name(dev), (dev->driver) ? dev->driver->name : "",
+       bus, slot);

    ltp_pci.dev = dev;
    ltp_pci.bus = dev->bus;
@@ -444,7 +445,15 @@ static int test_assign_resources(void)
            r->flags & IORESOURCE_PREFETCH) {
            ret = pci_assign_resource(dev, i);
            prk_info("assign resource to '%d', ret '%d'", i, ret);
-           rc |= (ret < 0 && ret != -EBUSY) ? TFAIL : TPASS;
+           if (dev->driver && !strncmp(dev->driver->name, "virtio-pci",
+                           strlen("virtio-pci"))) {
+               if (ret < 0 && ret != -EBUSY && ret != -ENOMEM)
+                   rc |= TFAIL;
+               else
+                   rc |= TPASS;
+           } else {
+               rc |= (ret < 0 && ret != -EBUSY) ? TFAIL : TPASS;
+           }
        }
    }

But the modification method of the ltp community is： [https://github.com/linux-test-project/ltp/commit/d631e9caef6818cffae6e028191787efcf65b3da] https://github.com/linux-test-project/ltp/blame/master/testcases/kernel/device-drivers/pci/tpci_kernel/ltp_tpci.c#L445-L446

Can you tell me whether the use of i440fx by vm is caused by the problem of ltp use cases? Looking forward to your reply

linux-test-project / ltp

ltp_tpci: testcase 12 related to virtio releasing and reassigning resource leads to guest hung #1061