Question on a patch in lib/ethdev/ethdev_driver.c

Question on a patch in lib/ethdev/ethdev_driver.c

[Thierry Herbelot]

Hello,

Commit 9a9eb104ed ("ethdev: parse multiple representor devargs") removes
the following variable initialization in rte_eth_devargs_parse():

	memset(eth_da, 0, sizeof(*eth_da));

This in turn causes a failure in ixgbe PCI probe on Arm server:

	eth_ixgbe_pci_probe(): unsupported representor type:

Indeed, nb_representor_ports was supposed to be correctly defined when
calling rte_eth_devargs_parse().

The question is:
why was "memset(eth_da, 0, sizeof(*eth_da));" removed ?
This is not explained in the commit log.

	Best regards

	Thierry

-- 
Thierry Herbelot
CSE - Titulaire au collège Cadres
Senior Software Engineer
http://www.6wind.com/

Follow us:
https://www.linkedin.com/company/6wind/
https://twitter.com/6WINDsoftware
https://www.youtube.com/user/6windsoftware

Re: Question on a patch in lib/ethdev/ethdev_driver.c

[Ferruh Yigit]

On 3/11/2024 9:07 AM, Thierry Herbelot wrote:
> Hello,
> 
> Commit 9a9eb104ed ("ethdev: parse multiple representor devargs") removes
> the following variable initialization in rte_eth_devargs_parse():
> 
>     memset(eth_da, 0, sizeof(*eth_da));
> 
> This in turn causes a failure in ixgbe PCI probe on Arm server:
> 
>     eth_ixgbe_pci_probe(): unsupported representor type:
> 
> Indeed, nb_representor_ports was supposed to be correctly defined when
> calling rte_eth_devargs_parse().
> 
> The question is:
> why was "memset(eth_da, 0, sizeof(*eth_da));" removed ?
> This is not explained in the commit log.
> 

Hi Thierry,

I think it is an oversight, thanks for reporting it.

+1 to preserve the existing behavior and function memset the input
parameter.

@Thierry, do you want to submit a fix? Else me or Harman can send one.

Thomas/David, can we hold the -rc3 for this fix please?

Thanks,
ferruh