Re: [dpdk-stable] patch 'ethdev: fix port id allocation' has been queued to LTS release 17.11.1

[Yuanhan Liu <yliu@fridaylinux.org>]

On Sun, Feb 11, 2018 at 06:40:27AM +0000, Matan Azrad wrote:
> Ho Yuanhan
> 
> From: Yuanhan Liu. Sent: Sunday, February 11, 2018 4:47 AM
> > On Thu, Feb 01, 2018 at 05:47:54PM +0800, Yuanhan Liu wrote:
> > > Hi,
> > >
> > > FYI, your patch has been queued to LTS release 17.11.1
> > >
> > > Note it hasn't been pushed to
> > https://emea01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fdpd
> > k.org%2Fbrowse%2Fdpdk-
> > stable&data=02%7C01%7Cmatan%40mellanox.com%7C1d21efdca33143ad70f
> > 808d570f9b78f%7Ca652971c7d2e4d9ba6a4d149256f461b%7C0%7C0%7C63653
> > 9140229336610&sdata=2Pgtg0sgIoqPmYRPirEjMi2dmJKVKyFljRZHGLIcz5M%3
> > D&reserved=0 yet.
> > > It will be pushed if I get no objections before 02/03/18. So please
> > > shout if anyone has objections.
> > 
> > It introduced an crash at port allocation.
> > And I have now reverted it. Please do a backport if necessary.
> > 
> 
> Please read the commit log and let me know if it necessary for this version - some secondary scenarios may fail without this patch.
> Please send me the crash details, maybe it exposes new bugs in the stable branch.

I didn't dig, the error happened at tring to access rte_eth_dev_data[0].name.

	if (rte_eth_dev_data[i].name[0] == '\0') {

What I observed is "rte_eth_dev_data" is NULL, looks like it haven't been
initiated yet.

	--yliu
> 
> 
> > Thanks.
> > 
> > 	--yliu
> > >
> > > ---
> > > >From 67abe74f3c09e6f166902c9fc5f4d7f65f7b3007 Mon Sep 17 00:00:00
> > > >2001
> > > From: Matan Azrad <matan@mellanox.com>
> > > Date: Mon, 22 Jan 2018 16:38:18 +0000
> > > Subject: [PATCH] ethdev: fix port id allocation
> > >
> > > [ upstream commit 8ee892a2385c50427c03db5cef1789babceb5999 ]
> > >
> > > rte_eth_dev_find_free_port() found a free port by state checking.
> > > The state field are in local process memory, so other DPDK processes
> > > may get the same port ID because their local states may be different.
> > >
> > > Replace the state checking by the ethdev port name checking, so, if
> > > the name is an empty string the port ID will be detected as unused.
> > >
> > > Fixes: d948f596fee2 ("ethdev: fix port data mismatched in multiple
> > > process model")
> > >
> > > Suggested-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> > > Signed-off-by: Matan Azrad <matan@mellanox.com>
> > > Acked-by: Thomas Monjalon <thomas@monjalon.net>
> > > Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> > > ---
> > >  lib/librte_ether/rte_ethdev.c | 6 +++++-
> > >  1 file changed, 5 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/lib/librte_ether/rte_ethdev.c
> > > b/lib/librte_ether/rte_ethdev.c index 9dcb0d2..8ea85cd 100644
> > > --- a/lib/librte_ether/rte_ethdev.c
> > > +++ b/lib/librte_ether/rte_ethdev.c
> > > @@ -192,8 +192,12 @@ rte_eth_dev_find_free_port(void)
> > >  	unsigned i;
> > >
> > >  	for (i = 0; i < RTE_MAX_ETHPORTS; i++) {
> > > -		if (rte_eth_devices[i].state == RTE_ETH_DEV_UNUSED)
> > > +		/* Using shared name field to find a free port. */
> > > +		if (rte_eth_dev_data[i].name[0] == '\0') {
> > > +			RTE_ASSERT(rte_eth_devices[i].state ==
> > > +				   RTE_ETH_DEV_UNUSED);
> > >  			return i;
> > > +		}
> > >  	}
> > >  	return RTE_MAX_ETHPORTS;
> > >  }
> > > --
> > > 2.7.4