Re: [PATCH v7 2/4] lib: fix comparison between devices

[Stephen Hemminger <stephen@networkplumber.org>]

On Thu, 6 Mar 2025 16:26:50 +0000
Shani Peretz <shperetz@nvidia.com> wrote:

> > -----Original Message-----
> > From: Stephen Hemminger <stephen@networkplumber.org>
> > Sent: Thursday, 20 February 2025 20:33
> > To: Shani Peretz <shperetz@nvidia.com>
> > Cc: dev@dpdk.org; Tyler Retzlaff <roretzla@linux.microsoft.com>; Parav Pandit
> > <parav@nvidia.com>; Xueming Li <xuemingl@nvidia.com>; Nipun Gupta
> > <nipun.gupta@amd.com>; Nikhil Agarwal <nikhil.agarwal@amd.com>; Hemant
> > Agrawal <hemant.agrawal@nxp.com>; Sachin Saxena
> > <sachin.saxena@nxp.com>; Rosen Xu <rosen.xu@intel.com>; Chenbo Xia
> > <chenbox@nvidia.com>; Tomasz Duszynski <tduszynski@marvell.com>;
> > Chengwen Feng <fengchengwen@huawei.com>; NBU-Contact-longli
> > (EXTERNAL) <longli@microsoft.com>; Wei Hu <weh@microsoft.com>; Bruce
> > Richardson <bruce.richardson@intel.com>; Kevin Laatz
> > <kevin.laatz@intel.com>; Jan Blunck <jblunck@infradead.org>
> > Subject: Re: [PATCH v7 2/4] lib: fix comparison between devices
> > 
> > External email: Use caution opening links or attachments
> > 
> > 
> > On Wed, 12 Feb 2025 18:38:33 +0200
> > Shani Peretz <shperetz@nvidia.com> wrote:
> >   
> > > DPDK supports multiple formats for specifying buses, (such as
> > > "0000:08:00.0" and "08:00.0" for PCI).
> > > This flexibility can lead to inconsistencies when using one format
> > > while running testpmd, then attempts to use the other format in a
> > > later command, resulting in a failure.
> > >
> > > The issue arises from the find_device function, which compares the
> > > user-provided string directly with the device->name in the rte_device
> > > structure.
> > > If we want to accurately compare these names, we'll need to bring both
> > > sides to the same representation by invoking the parse function on the
> > > user input.  
> > 
> > Could you give an example where this happens please?
> > Shouldn't find_device string always be changed into canonical form in
> > find_device handler?  
> 
> The flow I was dealing with was attach_port -> rte_dev_probe - > local_dev_probe -> find_device. 
> The string passed to attach_port was the short version, directly from the user.
> 
> So, to clarify - you're saying that find_device simply need to accept the string in its canonical form? Which means we'll only need to fix local_dev_probe to bring it to the canonical form before calling find_device?
> I tried it but then I noticed that there's no function that gets the user-provided string and returns it's string canonical form. The closest to this is parse, but what it eventually returns is not necessarily a string - it can be anything - for instance pci_parse will give you back a struct rte_pci_addr.

Not sure at this point. There are two options. One would be fixup in attach_port the other would be allowing
short form in PCI part of find_device. Since the strings from command line are put in canonical form
for devargs, it seems logical to do it in attach_port path.