[EXT] Re: [PATCH 1/2] ethdev: parsing multiple representor devargs string

[Harman Kalra]

Hi Andrew

Thanks for review
Please find response inline

> -----Original Message-----
> From: Andrew Rybchenko <andrew.rybchenko at oktetlabs.ru>
> Sent: Friday, January 12, 2024 12:56 PM
> To: Harman Kalra <hkalra at marvell.com>; Ajit Khaparde
> <ajit.khaparde at broadcom.com>; Somnath Kotur
> <somnath.kotur at broadcom.com>; John Daley <johndale at cisco.com>;
> Hyong Youb Kim <hyonkim at cisco.com>; Yuying Zhang
> <Yuying.Zhang at intel.com>; Beilei Xing <beilei.xing at intel.com>; Qiming Yang
> <qiming.yang at intel.com>; Qi Zhang <qi.z.zhang at intel.com>; Wenjun Wu
> <wenjun1.wu at intel.com>; Dariusz Sosnowski <dsosnowski at nvidia.com>;
> Viacheslav Ovsiienko <viacheslavo at nvidia.com>; Ori Kam
> <orika at nvidia.com>; Suanming Mou <suanmingm at nvidia.com>; Matan
> Azrad <matan at nvidia.com>; Chaoyong He <chaoyong.he at corigine.com>;
> Thomas Monjalon <thomas at monjalon.net>; Ferruh Yigit
> <ferruh.yigit at amd.com>
> Cc: dev at dpdk.org
> Subject: [EXT] Re: [PATCH 1/2] ethdev: parsing multiple representor devargs
> string
> 
> External Email
> 
> ----------------------------------------------------------------------
> On 1/11/24 09:44, Harman Kalra wrote:
> > Adding support for parsing multiple representor devargs strings passed
> > to a PCI BDF. There may be scenario where port representors for
> > various PFs or VFs under PFs are required and all these are
> > representor ports shall be backed by single pci device. In such case
> > port representors can be created using devargs string:
> > <PCI
> > BDF>,representor=pf[0-1],representor=pf2vf[1,2-3],representor=[4-5]
> >
> > Signed-off-by: Harman Kalra <hkalra at marvell.com>
> 
> [snip]
> 
> > diff --git a/lib/ethdev/ethdev_driver.c b/lib/ethdev/ethdev_driver.c
> > index bd917a15fc..62a06b75a2 100644
> > --- a/lib/ethdev/ethdev_driver.c
> > +++ b/lib/ethdev/ethdev_driver.c
> > @@ -470,15 +470,14 @@ eth_dev_devargs_tokenise(struct rte_kvargs
> *arglist, const char *str_in)
> >   }
> >
> >   int
> > -rte_eth_devargs_parse(const char *dargs, struct rte_eth_devargs
> > *eth_da)
> > +rte_eth_devargs_parse(const char *dargs, struct rte_eth_devargs
> > +*eth_devargs)
> >   {
> > -	struct rte_kvargs args;
> > +	struct rte_eth_devargs *eth_da;
> >   	struct rte_kvargs_pair *pair;
> > -	unsigned int i;
> > +	struct rte_kvargs args;
> > +	unsigned int i, j = 0;
> 
> Please, avoid multiple variable in one line declaration with initialization. It is
> too error prone. Also j is a bad name here. It is not a loop counter or
> seomthing like this.
> Maybe 'parsed' or 'devargs_parsed'?

Sure, will replace 'j' with a better name.

> 
> >   	int result = 0;
> >
> > -	memset(eth_da, 0, sizeof(*eth_da));
> > -
> >   	result = eth_dev_devargs_tokenise(&args, dargs);
> >   	if (result < 0)
> >   		goto parse_cleanup;
> > @@ -486,18 +485,16 @@ rte_eth_devargs_parse(const char *dargs, struct
> rte_eth_devargs *eth_da)
> >   	for (i = 0; i < args.count; i++) {
> >   		pair = &args.pairs[i];
> >   		if (strcmp("representor", pair->key) == 0) {
> > -			if (eth_da->type != RTE_ETH_REPRESENTOR_NONE) {
> > -				RTE_ETHDEV_LOG_LINE(ERR, "duplicated
> representor key: %s",
> > -					dargs);
> > -				result = -1;
> > -				goto parse_cleanup;
> > -			}
> > +			eth_da = &eth_devargs[j];
> > +			memset(eth_da, 0, sizeof(*eth_da));
> >   			result = rte_eth_devargs_parse_representor_ports(
> >   					pair->value, eth_da);
> >   			if (result < 0)
> >   				goto parse_cleanup;
> > +			j++;
> >   		}
> >   	}
> > +	result = (int)j;
> >
> >   parse_cleanup:
> >   	free(args.str);
> > diff --git a/lib/ethdev/ethdev_driver.h b/lib/ethdev/ethdev_driver.h
> > index b482cd12bb..6f4f1a1606 100644
> > --- a/lib/ethdev/ethdev_driver.h
> > +++ b/lib/ethdev/ethdev_driver.h
> > @@ -1800,10 +1800,10 @@ rte_eth_representor_id_get(uint16_t port_id,
> >    * @param devargs
> >    *  device arguments
> >    * @param eth_devargs
> > - *  parsed ethdev specific arguments.
> > + *  contiguous memory populated with parsed ethdev specific arguments.
> 
> Caller must provide information about array size. Right now you simply
> introduce out-of-bound array access.
> All drivers just provide one entry and if I pass many-many devargs I can write
> far beyond that location and corrupt stack.

We thought of adding an additional third argument to the API:
int rte_eth_devargs_parse(const char *dargs, struct rte_eth_devargs *eth_devargs, uint8_t nb_da);

Later skipped just with an intention to have minimal changes to existing driver calls.
But yes, stack corruption case may happen without passing array size.

Will update the implementation as per above protype with 3 args.

@All, any thoughts on this??

Thanks
Harman


> 
> >    *
> >    * @return
> > - *   Negative errno value on error, 0 on success.
> > + *   Negative errno value on error, no of devargs parsed on success.
> >    */
> >   __rte_internal
> >   int