[dpdk-dev] [Bug 741] [dpdk-21.05]vhost relaunch meet Segmentation fault issue when virtio queues larger than vhost queues

[dpdk-dev] [Bug 741] [dpdk-21.05]vhost relaunch meet Segmentation fault issue when virtio queues larger than vhost queues

[bugzilla]

https://bugs.dpdk.org/show_bug.cgi?id=741

            Bug ID: 741
           Summary: [dpdk-21.05]vhost relaunch meet Segmentation fault
                    issue when virtio queues larger than vhost queues
           Product: DPDK
           Version: unspecified
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: vhost/virtio
          Assignee: dev@dpdk.org
          Reporter: weix.ling@intel.com
  Target Milestone: ---

Issue discription：
vhost relaunch meet segment fault issue from when virtio queues larger than
vhost queues.

Test Environment:
DPDK version: DPDK v21.05
OS: Linux 5.4
Compiler: gcc 9.3
CPU: 8280m

The reproduce step is :
1. launch two vhost ports :

./x86_64-native-linuxapp-gcc/app/dpdk-testpmd -n 4 -l 2-4 --file-prefix=vhost 
--vdev 'net_vhost0,iface=/tmp/s0,client=1,queues=1' --vdev
'net_vhost1,iface=/tmp/s1,client=1,queues=1' -- -i --nb-cores=1 --txd=1024
--rxd=1024
testpmd>start
2. launch one virtio-user:

./x86_64-native-linuxapp-gcc/app/dpdk-testpmd -n 4 -l 5-6 --no-pci
--file-prefix=virtio0 \
--vdev=net_virtio_user0,mac=00:01:02:03:04:05,path=/tmp/s0,server=1,mrg_rxbuf=0,in_order=0,queues=1
\
-- -i --tx-offloads=0x0 --enable-hw-vlan-strip --nb-cores=1 --txd=1024
--rxd=1024
testpmd>start
3. launch another virtio-user with queue number=8:

 ./x86_64-native-linuxapp-gcc/app/dpdk-testpmd -n 4 -l 7-8 --no-pci
--file-prefix=virtio1 \
--vdev=net_virtio_user1,mac=00:01:02:03:04:55,path=/tmp/s1,server=1,mrg_rxbuf=0,in_order=0,queues=8
\
-- -i --tx-offloads=0x0 --enable-hw-vlan-strip --nb-cores=1 --txd=1024
--rxd=1024
testpmd>start tx_first 32
4. relaunch two vhost ports :

./x86_64-native-linuxapp-gcc/app/dpdk-testpmd -n 4 -l 2-4 --file-prefix=vhost 
--vdev 'net_vhost0,iface=/tmp/s0,client=1,queues=1' --vdev
'net_vhost1,iface=/tmp/s1,client=1,queues=1' -- -i --nb-cores=1 --txd=1024
--rxd=1024

Result:
Checking link statuses...
VHOST_CONFIG: read message VHOST_USER_SET_VRING_ENABLE
VHOST_CONFIG: set queue enable: 0 to qp idx: 0
VHOST_CONFIG: read message VHOST_USER_SET_VRING_ENABLE
VHOST_CONFIG: set queue enable: 0 to qp idx: 0
VHOST_CONFIG: read message VHOST_USER_SET_VRING_ENABLE
VHOST_CONFIG: set queue enable: 0 to qp idx: 1
VHOST_CONFIG: read message VHOST_USER_SET_VRING_ENABLE
VHOST_CONFIG: set queue enable: 0 to qp idx: 1
VHOST_CONFIG: read message VHOST_USER_GET_VRING_BASE
VHOST_CONFIG: vring base idx:0 file:0
VHOST_CONFIG: read message VHOST_USER_GET_VRING_BASE
VHOST_CONFIG: vring base idx:0 file:0
VHOST_CONFIG: read message VHOST_USER_GET_VRING_BASE
VHOST_CONFIG: vring base idx:1 file:0
VHOST_CONFIG: read message VHOST_USER_GET_VRING_BASE
VHOST_CONFIG: vring base idx:1 file:0
VHOST_CONFIG: read message VHOST_USER_GET_FEATURES
VHOST_CONFIG: read message VHOST_USER_GET_VRING_BASE
Segmentation fault

Expected Result:
vhost port can relaunch success and 1 queue can wok.


Is this issue a regression: Y
Version the regression was introduced: Specify git id if known.
root@dpdk:~/dpdk# git bisect bad 
commit 7804bbd13aa583e3a28b08557f7a98fcbe7fc8a8 (HEAD -> main)
Author: Maxime Coquelin <maxime.coquelin@redhat.com>
Date:   Fri Nov 6 15:47:44 2020 +0100

    vhost: fix virtqueue initialization

    This patches fixes virtqueue initialization issue causing
    segfault or file descriptor being closed unexpectedly.

    The wrong index was passed to init_vring_queue() by
    alloc_vring_queue() when a hole in the virtqueue array was
    met.

    Fixes: 8acd7c213353 ("vhost: fix virtqueues metadata allocation")
    Cc: stable@dpdk.org

    Reported-by: Yu Jiang <yux.jiang@intel.com>
    Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
    Reviewed-by: David Marchand <david.marchand@redhat.com>
    Tested-by: Yu Jiang <yux.jiang@intel.com>

-- 
You are receiving this mail because:
You are the assignee for the bug.