Re: [PATCH] tools lib traceevent: Add support for IP address formats

[Jiri Olsa <jolsa@redhat.com>]

On Thu, Dec 18, 2014 at 06:41:58PM -0700, David Ahern wrote:
> On 12/18/14 3:13 PM, David Ahern wrote:
> >Are you seeing something different?
> 
> I put a tracepoint in the ipv4/ipv6 add/del address code. The IPv4 one
> prints the address using pi4 and pI4 and the IPv6 using pi6, pI6 and pI6c.

apart from your tracepoints, is there another tracepoint using this?
I couldn't find any.. I dont think it's a problem.. just wanted to
try it ;-)

thanks,
jirka

> 
> Using ifconfig eth1 1.2.3.4/24
> 
> I get this for trace_pipe which uses the in-kernel printf code:
> 
> root@localhost:/sys/kernel/debug/tracing/events# cat ../trace_pipe
>         ifconfig-601   [000] ....   710.660611: ipv4_del_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig-602   [000] ....   720.833559: ipv4_add_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig-602   [000] .Ns.   720.833966: ipv6_add_addr:
> pi6=fe80000000000000000202fffe020211
> pI6=fe80:0000:0000:0000:0002:02ff:fe02:0211 pI6c=fe80::2:2ff:fe02:211
>         ifconfig-602   [000] ....   720.835413: ipv4_del_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig-602   [000] .N..   720.835451: ipv4_add_addr: pI4=1.2.3.4
> pi4=001.002.003.004
> 
> 
> And this using perf:
> root@localhost:~# perf record -a -e  ipv4:* -e ipv6:* --no-buffering | perf
> --no-pager script
> 
>        ifconfig   601 [000]   710.912862: ipv4:ipv4_del_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig   602 [000]   721.090776: ipv4:ipv4_add_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig   602 [000]   721.091177: ipv6:ipv6_add_addr:
> pi6=fe80000000000000000202fffe020211
> pI6=fe80:0000:0000:0000:0002:02ff:fe02:0211 pI6c=fe80::2:2ff:fe02:211
>         ifconfig   602 [000]   721.092626: ipv4:ipv4_del_addr: pI4=1.2.3.4
> pi4=001.002.003.004
>         ifconfig   602 [000]   721.092648: ipv4:ipv4_add_addr: pI4=1.2.3.4
> pi4=001.002.003.004
> 
> (Don't be fooled by the differing timestamps; data collected from both using
> the same exact command. timestamp differences are perf vs tracing.)
> 
> which suggest with the patch perf output matches tracing.
> 
> David
>