From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: multipart/mixed; boundary="===============3883019095038932840=="
MIME-Version: 1.0
From: Niu, Yawei <yawei.niu at intel.com>
Subject: [SPDK] Re: Potential defect in pci_unplug()
Date: Tue, 29 Sep 2020 01:17:40 +0000
Message-ID: <E3D5E4F4-FE73-4861-B80A-15110705948C@intel.com>
In-Reply-To: 748FDEF7-6BB0-4D27-B630-4C991D06B6F8@intel.com
List-ID: <spdk@lists.01.org>
To: spdk@lists.01.org

--===============3883019095038932840==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

Thanks a lot, Jim. I've opened SPDK issue #1605 to track this, and the mail=
 is forwarded to dev(a)dpdk.org.

Thanks
-Niu

=EF=BB=BFOn 2020/9/28, 11:44 PM, "Harris, James R" <james.r.harris(a)intel.=
com> wrote:

    Hi Niu,
    =

    I agree, this doesn't look right.  Could you file an SPDK issue for thi=
s to make sure we track it?  And then try sending an e-mail to the DPDK mai=
ling list?  I'm open to submitting a patch to our DPDK submodule short-term=
, but only if we get agreement with DPDK community that this fix is correct.
    =

    Thanks,
    =

    Jim
    =

    =

    On 9/28/20, 12:17 AM, "Niu, Yawei" <yawei.niu(a)intel.com> wrote:
    =

        Hi,
    =

        In the pci_unplug() (dpdk/drivers/bus/pci/pci_common.c), why do we =
call rte_devargs_remove() to remove the saved device args? That looks a def=
ect to me, since that=E2=80=99ll result in the hot removed device failed to=
 be detected when it=E2=80=99s plugged back (when white list option was pro=
vided), the situation is described following:
    =

          1.  App starts with white list option provided, let=E2=80=99s sup=
pose the device in white list is: 0000:81:00.0;
          2.  rte_devargs_add() is called to add this device arg on rte_eal=
_init();
          3.  Issue =E2=80=9Cecho 1 > /sys/bus/pci/devices/0000\:81\:00.0/r=
emove=E2=80=9D to hot remove the device;
          4.  pci_unplug() is called to remove the device, and rte_devargs_=
remove() is called, so this device arg for white list is remove too;
          5.  Issue =E2=80=9Cecho 1 > /sys/bus/pci/rescan=E2=80=9D then =E2=
=80=9CPCI_WHITELIST=3D=E2=80=9D0000:81:00.0=E2=80=9D spdk/script/setup.sh=
=E2=80=9D to do hot plug;
          6.  In pci_scan_one(), new dev is generated, however, the dev->de=
vice.devargs is set NULL since the device args has been removed on device u=
nplug;
          7.  rte_pci_probe() does white list scan, however, it unexpectedl=
y skips the device because the devargs of the device is NULL now;
    =

        I don=E2=80=99t fully understand the DPDK code, but this rte_devarg=
s_remove() in pci_unplug() doesn=E2=80=99t make sense to me (when I comment=
ed it out, seems everything will work as expected). Is this a glitch or I m=
issed anything?
    =

        Thanks
        -Niu
    =

    =

        _______________________________________________
        SPDK mailing list -- spdk(a)lists.01.org
        To unsubscribe send an email to spdk-leave(a)lists.01.org
    =

    _______________________________________________
    SPDK mailing list -- spdk(a)lists.01.org
    To unsubscribe send an email to spdk-leave(a)lists.01.org
    =



--===============3883019095038932840==--