From mboxrd@z Thu Jan 1 00:00:00 1970 From: York Sun Date: Thu, 13 Aug 2015 08:54:51 -0700 Subject: [U-Boot] [PATCH v3 11/16] net/eth.c: Add function to validate a MAC address In-Reply-To: References: <1437746136-14379-1-git-send-email-codrin.ciubotariu@freescale.com> <1437746136-14379-8-git-send-email-codrin.ciubotariu@freescale.com> <55C54DC2.2070401@freescale.com> <55C8FE55.1070302@freescale.com> <55C90389.9090804@freescale.com> <55C90D5D.4000705@freescale.com> <55CBA580.4040903@freescale.com> Message-ID: <55CCBDCB.3060908@freescale.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On 08/13/2015 08:42 AM, Ciubotariu Codrin Constantin-B43658 wrote: > Hi Joe, York, > >> -----Original Message----- >> From: Sun York-R58495 >> Sent: Wednesday, August 12, 2015 10:59 PM >> To: Ciubotariu Codrin Constantin-B43658 >> Cc: Joe Hershberger; u-boot at lists.denx.de; joe.hershberger at ni.com >> Subject: Re: [U-Boot] [PATCH v3 11/16] net/eth.c: Add function to validate a MAC >> address >> >> +Codrin >> >> Somehow I dropped Codrin in last reply. >> >> On 08/10/2015 01:45 PM, York Sun wrote: >>> >>> >>> On 08/10/2015 01:05 PM, Joe Hershberger wrote: >>>> Hi York, >>>> >>>> On Mon, Aug 10, 2015 at 3:03 PM, York Sun wrote: >>>>> >>>>> >>>>> On 08/10/2015 12:57 PM, Joe Hershberger wrote: >>>>>> Too much top-posting. >>>>>> >>>>>> On Mon, Aug 10, 2015 at 2:41 PM, York Sun wrote: >>>>>>> SPL doesn't use net/eth.c. You add a call in env_flags.c. >>>>>>> >>>>>>> I think you can put it in header file and use static inline, or >>>>>>> keep it in the same file where it is called. >>>>>> >>>>>> That is probably fine. >>>>>> >>>>>>> Another way is to undef CONFIG_CMD_NET for SPL part. It is default >>>>>>> to 'y' in Kconfig. Joe may have some good suggestion. >>>>>> >>>>>> I don't think this is the reason. The problem is that net is *not* >>>>>> build for SPL, but env is. >>>>> >>>>> Yes, env is built. The offending lines in common/env_flags.c are >>>>> gated by "#ifdef CONFIG_CMD_NET". That's why I say it could be another way. >>>> >>>> OK, sure... but that breaks intended behavior, I think. >>>> >>> >>> I see. The CONFIG_CMD_NET is not evaluated separated for SPL build. So >>> I guess the fix can be either to put the common function in header >>> file after making it really simple to reduce dependency, or to keep the >> original code in env_flag.c. >>> >> >> Codrin, >> >> Can you prepare a new patch? You don't have to send the whole set. All but one >> have been acked by Joe. >> >> York > > I can't inline eth_validate_ethaddr_str in eth.h since it depends on simple_strtoul and tolower. Also, I can't let it in common/env_flags.c because I need to access if from drivers/net/vsc9953.c . I guess it has to be in a .c file that is also used by SPL targets. Could you please suggest such a file? > > SPL targets make use of CONFIG_CMD_NET? It seems strange that ETH env is built, but net/net.c or net/eth.c not. > I was discussing with Joe about the possibility to deselect CONFIG_CMD_NET for SPL build. The issue here is Kconfig is not re-evaluated for the SPL part. If you can experiment it, you can try to gate the code in env_flags.c with CONFIG_SPL_BUILD. It sounds reasonable to me. York