[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <YYIywxTUwl2xp4+/@TonyMac-Alibaba>
Date: Wed, 3 Nov 2021 14:57:07 +0800
From: Tony Lu <tonylu@...ux.alibaba.com>
To: Karsten Graul <kgraul@...ux.ibm.com>
Cc: kuba@...nel.org, davem@...emloft.net, guwen@...ux.alibaba.com,
dust.li@...ux.alibaba.com, netdev@...r.kernel.org,
linux-s390@...r.kernel.org, linux-rdma@...r.kernel.org
Subject: Re: [PATCH net-next 3/3] net/smc: Introduce tracepoint for smcr link
down
On Tue, Nov 02, 2021 at 10:30:22AM +0100, Karsten Graul wrote:
> On 01/11/2021 08:39, Tony Lu wrote:
> > +
> > + TP_printk("lnk=%p lgr=%p state=%d dev=%s location=%p",
> > + __entry->lnk, __entry->lgr,
> > + __entry->state, __get_str(name),
> > + __entry->location)
>
> The location is printed as pointer (which might even be randomized?),
> is it possible to print the function name of the caller, as described
> here: https://stackoverflow.com/questions/4141324/function-caller-in-linux-kernel
>
> printk("Caller is %pS\n", __builtin_return_address(0));
>
> Not sure if this is possible with the trace points, but it would be
> easier to use. You plan to use a dump to find out about the function caller?
Yes, I am going to find out where caused the SMC-R link down. In our
test environment, the tracepoint for link down help me to debug the root
cause of link termination, without eBPF or systemtap.
By using "%pS", it makes the trace log easy to show the function caller
name without additional translating.
<idle>-0 [000] ..s. 69.087164: smcr_link_down: lnk=00000000dab41cdc lgr=000000007d5d8e24 state=0 rc=1 dev=mlx5_0 location=smc_wr_tx_tasklet_fn+0x5ef/0x6f0 [smc]
I will improve it in the next patch.
Cheers,
Tony Lu
Powered by blists - more mailing lists