Re: [PATCH v5 6/7] app/testpmd: add commands to dump mbuf history

[Thomas Monjalon <thomas@monjalon.net>]

14/10/2025 11:48, Bruce Richardson:
> On Tue, Oct 14, 2025 at 11:43:28AM +0200, Thomas Monjalon wrote:
> > 14/10/2025 10:45, Morten Brørup:
> > > > From: Thomas Monjalon [mailto:thomas@monjalon.net]
> > > > Sent: Tuesday, 14 October 2025 08.59
> > > > 
> > > > Dump the mbuf history to the console or to a file.
> > > > 
> > > > The dump will contain:
> > > > - Operation history for each mbuf
> > > > - Summary and statistics about all mbufs
> > > > 
> > > > Dump the history of all mbufs:
> > > > 	testpmd> dump mbuf history all [file]
> > > > 
> > > > Dump the history of one mbuf pool:
> > > > 	testpmd> dump mbuf history mempool <mp_name> [file]
> > > > 
> > > > Dump the history of one mbuf:
> > > > 	testpmd> dump mbuf history one <mbuf_addr> [file]
> > > 
> > > The "one" seems superfluous. I think this would suffice:
> > > 	testpmd> dump mbuf history <mbuf_addr> [file]
> > 
> > It's not superfluous :)
> > Without it, the cmdline library cannot distinguish easily
> > between
> > 	dump mbuf history <mbuf_addr> [file]
> > and
> > 	dump mbuf history mempool <mp_name>
> > 
> The latter might read better as "dump mbuf pool history <mp_name>", which
> would also resolve the ambiguity.

I'll try that.

> Even without cmdline lib resolving the ambiguity, it should be easy enough
> for the cmdline callback to do strcmp() on the 4th parameter to check if
> it's the string "mempool" and then direct appropriately.

It is already doing strcmp(), but first, cmdline parsing must not be ambiguous.