netdev - Re: [PATCH net] net/smc: fix unexpected SMC_CLC_DECL_ERR

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20220302075146.GA29189@e02h04389.eu6sqa>
Date:   Wed, 2 Mar 2022 15:51:46 +0800
From:   "D. Wythe" <alibuda@...ux.alibaba.com>
To:     Wen Gu <guwen@...ux.alibaba.com>
Cc:     kgraul@...ux.ibm.com, kuba@...nel.org, davem@...emloft.net,
        netdev@...r.kernel.org, linux-s390@...r.kernel.org,
        linux-rdma@...r.kernel.org
Subject: Re: [PATCH net] net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error

On Wed, Mar 02, 2022 at 03:28:32PM +0800, Wen Gu wrote:
> 
> 
> 在 2022/3/1 下午9:17, D. Wythe 写道:
> >From: "D. Wythe" <alibuda@...ux.alibaba.com>
> >
> >Remove connections from link group is not synchronous with handling
> >SMC_LLC_DELETE_RKEY, which means that even the number of connections is
> >less that SMC_RMBS_PER_LGR_MAX, it does not mean that the connection can
> >register rtoken successfully later, in other words, the rtoken entry may
> >have not been released. This will cause an unexpected
> >SMC_CLC_DECL_ERR_REGRMB to be reported, and then ths smc connection have
> >to fallback to TCP.
> >
> 
> 
> IMHO, if there are SMC_RMBS_PER_LGR_MAX connections in the link group now,
> one of them is being removed and here comes a new connection at this moment,
> then:
> 
> (1) without this patch, the new connection will be added into the old link group
>     but fallback if the removing connection has not finished unregistering its rmb.
> 
> (2) with this patch, a new link group will be created and the new connection
>     will be added into the new link group.
> 
> I am wondering if (1) should be considered as a issue, or just a bydesign?
> If it is a issue, I think this patch works, Thanks!


We should always be willing to improve the success rate of the SMC 
connection, creating a new group is not a side effect of this patch, it 
actually dues to the state bewteen connections that can not achieve 
clock synchronization. In fact, it can happen in any times.

Thanks.