DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] IXGBE VF: segfault in ixgbe_dev_link_update
@ 2017-05-10  8:41 Gregory Etelson
  2017-05-10 10:08 ` Ferruh Yigit
  0 siblings, 1 reply; 2+ messages in thread
From: Gregory Etelson @ 2017-05-10  8:41 UTC (permalink / raw)
  To: dev; +Cc: Lu, Wenzhuo

Hello,

My application works with IXGBE VF 17.05-rc4.
After I shutdown switch port,
a call for rte_eth_link_get() crashes with SEGFAULT.
Apparently, ixgbe_dev_link_update() calls for non-existing in VF driver ixgbe_get_link_capabilities().

Is there a patch with get_link_capabilities() implementation for IXGBE VF I can use ?
or 
Can I safely disable call for ixgbe_get_link_capabilities() for VF ?

Regards,
Gregory 

#0  0x0000000000000000 in ?? ()
(gdb) bt
#0  0x0000000000000000 in ?? ()
#1  0x0000000000491c36 in ixgbe_dev_link_update (dev=0x751680, wait_to_complete=1) at /root/dpdk/drivers/net/ixgbe/ixgbe_ethdev.c:3802
#2  0x000000000043b753 in rte_eth_link_get (port_id=<value optimized out>, eth_link=0x7ffe991fff90) at /root/dpdk/lib/librte_ether/rte_ethdev.c:1306
(gdb) f 1
#1  0x0000000000491c36 in ixgbe_dev_link_update (dev=0x751680, wait_to_complete=1) at /root/dpdk/drivers/net/ixgbe/ixgbe_ethdev.c:3802
3802                    hw->mac.ops.get_media_type(hw) == ixgbe_media_type_fiber) {
(gdb) p speed
$1 = 0


s32 ixgbe_init_ops_vf(struct ixgbe_hw *hw)
{
	hw->mac.ops.get_link_capabilities = NULL;

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [dpdk-dev] IXGBE VF: segfault in ixgbe_dev_link_update
  2017-05-10  8:41 [dpdk-dev] IXGBE VF: segfault in ixgbe_dev_link_update Gregory Etelson
@ 2017-05-10 10:08 ` Ferruh Yigit
  0 siblings, 0 replies; 2+ messages in thread
From: Ferruh Yigit @ 2017-05-10 10:08 UTC (permalink / raw)
  To: Gregory Etelson, dev; +Cc: Lu, Wenzhuo

On 5/10/2017 9:41 AM, Gregory Etelson wrote:
> Hello,
> 
> My application works with IXGBE VF 17.05-rc4.
> After I shutdown switch port,
> a call for rte_eth_link_get() crashes with SEGFAULT.
> Apparently, ixgbe_dev_link_update() calls for non-existing in VF driver ixgbe_get_link_capabilities().

Hi Gregory,

This doesn't looks like because of the ixgbe_get_link_capabilities() but
hw->mac.ops.get_media_type.

Crash happens in line 3802, which has:
hw->mac.ops.get_media_type(hw) == ixgbe_media_type_fiber) {

In that stage "speed == 0" but that is expected.

You already aware of the issue and patch, but as reference, this issue
first reported in:
http://dpdk.org/ml/archives/dev/2017-May/065533.html

and proposed fix is in:
http://dpdk.org/ml/archives/dev/2017-May/065562.html

Thanks,
ferruh

> 
> Is there a patch with get_link_capabilities() implementation for IXGBE VF I can use ?
> or 
> Can I safely disable call for ixgbe_get_link_capabilities() for VF ?
> 
> Regards,
> Gregory 
> 
> #0  0x0000000000000000 in ?? ()
> (gdb) bt
> #0  0x0000000000000000 in ?? ()
> #1  0x0000000000491c36 in ixgbe_dev_link_update (dev=0x751680, wait_to_complete=1) at /root/dpdk/drivers/net/ixgbe/ixgbe_ethdev.c:3802
> #2  0x000000000043b753 in rte_eth_link_get (port_id=<value optimized out>, eth_link=0x7ffe991fff90) at /root/dpdk/lib/librte_ether/rte_ethdev.c:1306
> (gdb) f 1
> #1  0x0000000000491c36 in ixgbe_dev_link_update (dev=0x751680, wait_to_complete=1) at /root/dpdk/drivers/net/ixgbe/ixgbe_ethdev.c:3802
> 3802                    hw->mac.ops.get_media_type(hw) == ixgbe_media_type_fiber) {
> (gdb) p speed
> $1 = 0
> 
> 
> s32 ixgbe_init_ops_vf(struct ixgbe_hw *hw)
> {
> 	hw->mac.ops.get_link_capabilities = NULL;
> 

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-05-10 10:08 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-05-10  8:41 [dpdk-dev] IXGBE VF: segfault in ixgbe_dev_link_update Gregory Etelson
2017-05-10 10:08 ` Ferruh Yigit

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).