Skip to contentSkip to conversation

mlx4: info get fails in secondary process

dpdk-dev [email protected] 3 participants 3 messages Jul 12, 2019 – Jul 19, 2019

lore
Share & tools
lore
#0 lore raw

https://bugs.dpdk.org/show_bug.cgi?id=320

            Bug ID: 320
           Summary: mlx4: info get fails in secondary process
           Product: DPDK
           Version: 19.08
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: other
          Assignee: [email protected]
          Reporter: [email protected]

Target Milestone: ---

Attempts to do dev_info_get in secondary process fail because the driver handle to infiniband (priv->ctx) is not valid in secondary process. This leads to:

#0 0x00007ffff5d06ab4 in _IO_vfprintf_internal (s=s@entry=0x7fffffffe3c0, format=<optimized out>, format@entry=0x7ffff496a605 "%s/device/net", ap=ap@entry=0x7fffffffe528) at vfprintf.c:1635
#1 0x00007ffff5d32599 in _IO_vsnprintf (string=0x7fffffffe4b0 "", maxlen=<optimized out>, format=0x7ffff496a605 "%s/device/net", args=args@entry=0x7fffffffe528) at vsnprintf.c:119
#2 0x00007ffff5d0fd32 in __snprintf (s=s@entry=0x0, maxlen=maxlen@entry=0, format=format@entry=0x7ffff496a605 "%s/device/net") at snprintf.c:34 #3 0x00007ffff4720b8b in mlx4_get_ifname (priv=priv@entry=0xc0137e93c0, ifname=ifname@entry=0x7fffffffe690) at /usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:71 #4 0x00007ffff47214e4 in mlx4_dev_infos_get (dev=<optimized out>, info=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:583 #5 0x00007ffff44b896c in rte_eth_dev_info_get (port_id=port_id@entry=0, dev_info=dev_info@entry=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/lib/librte_ethdev/rte_ethdev.c:2532

Signature · 2 lines
You are receiving this mail because:
You are the assignee for the bug.
#1 lore raw
Quoted 28 lines · On Fri, 12 Jul 2019 20:04:26 +0000 [email protected] wrote:

https://bugs.dpdk.org/show_bug.cgi?id=320

            Bug ID: 320
           Summary: mlx4: info get fails in secondary process
           Product: DPDK
           Version: 19.08
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: other
          Assignee: [email protected]
          Reporter: [email protected]

Target Milestone: ---

Attempts to do dev_info_get in secondary process fail because the driver handle to infiniband (priv->ctx) is not valid in secondary process. This leads to:

#0 0x00007ffff5d06ab4 in _IO_vfprintf_internal (s=s@entry=0x7fffffffe3c0, format=<optimized out>, format@entry=0x7ffff496a605 "%s/device/net", ap=ap@entry=0x7fffffffe528) at vfprintf.c:1635
#1 0x00007ffff5d32599 in _IO_vsnprintf (string=0x7fffffffe4b0 "", maxlen=<optimized out>, format=0x7ffff496a605 "%s/device/net", args=args@entry=0x7fffffffe528) at vsnprintf.c:119
#2 0x00007ffff5d0fd32 in __snprintf (s=s@entry=0x0, maxlen=maxlen@entry=0, format=format@entry=0x7ffff496a605 "%s/device/net") at snprintf.c:34 #3 0x00007ffff4720b8b in mlx4_get_ifname (priv=priv@entry=0xc0137e93c0,

Quoted 9 lines

ifname=ifname@entry=0x7fffffffe690) at /usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:71 #4 0x00007ffff47214e4 in mlx4_dev_infos_get (dev=<optimized out>, info=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:583 #5 0x00007ffff44b896c in rte_eth_dev_info_get (port_id=port_id@entry=0, dev_info=dev_info@entry=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/lib/librte_ethdev/rte_ethdev.c:2532

I am testing a patch for this.

#2 lore raw

Is it really needed for mlx4? mlx4 uses neither Netlink nor shared context.

With best regards, Slava

Quoted 65 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Stephen Hemminger
Sent: Saturday, July 13, 2019 0:04
To: [email protected]
Cc: [email protected]
Subject: Re: [dpdk-dev] [Bug 320] mlx4: info get fails in secondary process

Quoted 1 line · On Fri, 12 Jul 2019 20:04:26 +0000 [email protected] wrote:

https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbugs

Quoted 1 line

.dpdk.org%2Fshow_bug.cgi%3Fid%3D320&amp;data=02%7C01%7Cviachesla vo%40m

Quoted 1 line

ellanox.com%7Cc75dc339868c42296e5c08d7070c7dbf%7Ca652971c7d2e4d 9ba6a4d

Quoted 1 line

149256f461b%7C0%7C0%7C636985622553681551&amp;sdata=hyzFADnV52 GNWXbCPFK

Quoted 14 lines

zmUZS6j6gPbzNAxIqLrM%2BLRM%3D&amp;reserved=0

            Bug ID: 320
           Summary: mlx4: info get fails in secondary process
           Product: DPDK
           Version: 19.08
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: other
          Assignee: [email protected]
          Reporter: [email protected]
Quoted 4 lines

Target Milestone: ---

Attempts to do dev_info_get in secondary process fail because the driver handle to infiniband (priv->ctx) is not valid in secondary process. This

leads to:

Quoted 14 lines

#0 0x00007ffff5d06ab4 in _IO_vfprintf_internal (s=s@entry=0x7fffffffe3c0, format=<optimized out>, format@entry=0x7ffff496a605 "%s/device/net", ap=ap@entry=0x7fffffffe528) at vfprintf.c:1635
#1 0x00007ffff5d32599 in _IO_vsnprintf (string=0x7fffffffe4b0 "", maxlen=<optimized out>, format=0x7ffff496a605 "%s/device/net", args=args@entry=0x7fffffffe528) at vsnprintf.c:119
#2 0x00007ffff5d0fd32 in __snprintf (s=s@entry=0x0, maxlen=maxlen@entry=0,
format=format@entry=0x7ffff496a605 "%s/device/net") at snprintf.c:34 #3 0x00007ffff4720b8b in mlx4_get_ifname (priv=priv@entry=0xc0137e93c0, ifname=ifname@entry=0x7fffffffe690) at

Quoted 9 lines

/usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:71 #4 0x00007ffff47214e4 in mlx4_dev_infos_get (dev=<optimized out>, info=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/drivers/net/mlx4/mlx4_ethdev.c:583 #5 0x00007ffff44b896c in rte_eth_dev_info_get (port_id=port_id@entry=0, dev_info=dev_info@entry=0x7fffffffe6e0) at /usr/src/redhat/BUILD/dpdk-18.11/lib/librte_ethdev/rte_ethdev.c:2532

I am testing a patch for this.