tcpprobe display format for snd_nxt and snd_una

tcpprobe display format for snd_nxt and snd_una

[hiren panchasara]

[-- Attachment #1: Type: text/plain, Size: 1148 bytes --]

(New to linux and first-time poster so please guide me if needed.)
 
Upon using tcpprobe I realized that it prints snd_nxt and snd_una as hex
which makes it harder to read and compare with tcpdump for example.
 
Not sure if that is intentional. If not, a simple patch like this would
print them as decimals.
 
[PATCH] Display snd_nxt and snd_una as decimals for better
 readability.
 
---
 net/ipv4/tcp_probe.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
 
diff --git a/net/ipv4/tcp_probe.c b/net/ipv4/tcp_probe.c
index f6c50af..a8e66c1 100644
--- a/net/ipv4/tcp_probe.c
+++ b/net/ipv4/tcp_probe.c
@@ -191,7 +191,7 @@ static int tcpprobe_sprint(char *tbuf, int n)
                = ktime_to_timespec64(ktime_sub(p->tstamp, tcp_probe.start));
 
        return scnprintf(tbuf, n,
-                       "%lu.%09lu %pISpc %pISpc %d %#x %#x %u %u %u %u %u\n",
+                       "%lu.%09lu %pISpc %pISpc %d %u %u %u %u %u %u %u\n",
                        (unsigned long)ts.tv_sec,
                        (unsigned long)ts.tv_nsec,
                        &p->src, &p->dst, p->length, p->snd_nxt, p->snd_una,

Re: tcpprobe display format for snd_nxt and snd_una

[Stephen Hemminger]

On Wed, 5 Apr 2017 12:40:09 -0700
hiren panchasara <hiren@strugglingcoder.info> wrote:

> (New to linux and first-time poster so please guide me if needed.)
>  
> Upon using tcpprobe I realized that it prints snd_nxt and snd_una as hex
> which makes it harder to read and compare with tcpdump for example.
>  
> Not sure if that is intentional. If not, a simple patch like this would
> print them as decimals.
>  
> [PATCH] Display snd_nxt and snd_una as decimals for better
>  readability.
>  
> ---
>  net/ipv4/tcp_probe.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>  
> diff --git a/net/ipv4/tcp_probe.c b/net/ipv4/tcp_probe.c
> index f6c50af..a8e66c1 100644
> --- a/net/ipv4/tcp_probe.c
> +++ b/net/ipv4/tcp_probe.c
> @@ -191,7 +191,7 @@ static int tcpprobe_sprint(char *tbuf, int n)
>                 = ktime_to_timespec64(ktime_sub(p->tstamp, tcp_probe.start));
>  
>         return scnprintf(tbuf, n,
> -                       "%lu.%09lu %pISpc %pISpc %d %#x %#x %u %u %u %u %u\n",
> +                       "%lu.%09lu %pISpc %pISpc %d %u %u %u %u %u %u %u\n",
>                         (unsigned long)ts.tv_sec,
>                         (unsigned long)ts.tv_nsec,
>                         &p->src, &p->dst, p->length, p->snd_nxt, p->snd_una,
> --
>  
> Let me know if I am missing something obvious.

The output of tcpprobe is intended for consumption by programs.
Changing the output format would be considered a kernel ABI breakage which
is something Linux tries not to do. Therefore I would prefer it
not be changed. Sorry if this is inconvenient for you but breaking other
users scripts would be a bigger problem.

Also, your patch email is not formatted with subject [RFC] or [PATCH]
and is missing signed-off-by.

Re: tcpprobe display format for snd_nxt and snd_una

[hiren panchasara]

[-- Attachment #1: Type: text/plain, Size: 2731 bytes --]

On 04/06/17 at 05:45P, Stephen Hemminger wrote:
> On Wed, 5 Apr 2017 12:40:09 -0700
> hiren panchasara <hiren@strugglingcoder.info> wrote:
> 
> > (New to linux and first-time poster so please guide me if needed.)
> >  
> > Upon using tcpprobe I realized that it prints snd_nxt and snd_una as hex
> > which makes it harder to read and compare with tcpdump for example.
> >  
> > Not sure if that is intentional. If not, a simple patch like this would
> > print them as decimals.
> >  
> > [PATCH] Display snd_nxt and snd_una as decimals for better
> >  readability.
> >  
> > ---
> >  net/ipv4/tcp_probe.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >  
> > diff --git a/net/ipv4/tcp_probe.c b/net/ipv4/tcp_probe.c
> > index f6c50af..a8e66c1 100644
> > --- a/net/ipv4/tcp_probe.c
> > +++ b/net/ipv4/tcp_probe.c
> > @@ -191,7 +191,7 @@ static int tcpprobe_sprint(char *tbuf, int n)
> >                 = ktime_to_timespec64(ktime_sub(p->tstamp, tcp_probe.start));
> >  
> >         return scnprintf(tbuf, n,
> > -                       "%lu.%09lu %pISpc %pISpc %d %#x %#x %u %u %u %u %u\n",
> > +                       "%lu.%09lu %pISpc %pISpc %d %u %u %u %u %u %u %u\n",
> >                         (unsigned long)ts.tv_sec,
> >                         (unsigned long)ts.tv_nsec,
> >                         &p->src, &p->dst, p->length, p->snd_nxt, p->snd_una,
> > --
> >  
> > Let me know if I am missing something obvious.
> 
> The output of tcpprobe is intended for consumption by programs.
> Changing the output format would be considered a kernel ABI breakage which
> is something Linux tries not to do. Therefore I would prefer it
> not be changed.
Ah, I see.

> Sorry if this is inconvenient for you but breaking other
> users scripts would be a bigger problem.
No, I don't have any scripts depending on this. I just want to see this
o/p with naked eye and compare it with a pcap to see any possible
abnormalities of a "bad" connection. I guess I'd have to no write a
wrapper on top of this to achieve whhat

Can you point me to the programs/tools that consume this data? I am new
to Linux and want to learn tools that can help in troubleshooting. (In
FreeBSD land I can trivially dump this data with dtrace and I assume
something like that exists here too.)

I can always have a wrapper on top of this to do the conversion. Trivial
but inconvenient. :-)

> 
> Also, your patch email is not formatted with subject [RFC] or [PATCH]
> and is missing signed-off-by.

Apologies for that. I'll make sure I do the right thing next time as
this looks like a non-starter because of ABI concerns.

Thanks for taking time and responding.
Cheers,
Hiren

[-- Attachment #2: Type: application/pgp-signature, Size: 603 bytes --]