[U-Boot] [PATCH 1/5] fsl: serdes: ensure accessing the initialized maps of serdes protocol

[Prabhakar Kushwaha]

Hi Zhiqiang,

Sorry for late queries. 

As per description of patch " Up to now, the function is_serdes_configed() doesn't check if the map
of serdes protocol is initialized before accessing it. The function is_serdes_configed() will get wrong result when it was called before
the serdes protocol maps initialized. As the first eliment of the map isn't used for any device, so use it as the flag to indicate if the
map has been initialized."

I am just wondering the use-case/situation where this can happen.
Can you please help me with understanding. 

fsl_serdes_init is called from arch_early_init_r in board_r.c.  
As per my understanding all the driver calling is_serdes_configed (SATA, PCIe, SGMII) etc requires DDR.  
So are we talking about moving any driver in board_f.c. 

Regards,
Prabhakar

> -----Original Message-----
> From: york sun
> Sent: Thursday, July 21, 2016 2:45 AM
> To: Zhiqiang Hou <zhiqiang.hou at nxp.com>; u-boot at lists.denx.de;
> albert.u.boot at aribaud.net; wd at denx.de; Prabhakar Kushwaha
> <prabhakar.kushwaha at nxp.com>; alison.wang at freescale.com;
> Mingkai.Hu at freescale.com
> Cc: yao.yuan at freescale.com; Qianyu.Gong at freescale.com;
> bmeng.cn at gmail.com; Shengzhou Liu <shengzhou.liu at nxp.com>
> Subject: Re: [PATCH 1/5] fsl: serdes: ensure accessing the initialized maps of
> serdes protocol
> 
> On 07/20/2016 02:38 AM, Zhiqiang Hou wrote:
> > Hi York,
> >
> > Thanks for your comments!
> >
> >> -----Original Message-----
> >> From: york sun
> >> Sent: 2016年7月19日 23:46
> >> To: Zhiqiang Hou <zhiqiang.hou at nxp.com>; u-boot at lists.denx.de;
> >> albert.u.boot at aribaud.net; wd at denx.de; Prabhakar Kushwaha
> >> <prabhakar.kushwaha at nxp.com>; alison.wang at freescale.com;
> >> Mingkai.Hu at freescale.com
> >> Cc: yao.yuan at freescale.com; Qianyu.Gong at freescale.com;
> >> bmeng.cn at gmail.com; Shengzhou Liu <shengzhou.liu at nxp.com>
> >> Subject: Re: [PATCH 1/5] fsl: serdes: ensure accessing the initialized maps of
> serdes
> >> protocol
> >>
> >> On 07/03/2016 11:39 PM, Zhiqiang Hou wrote:
> >>> From: Hou Zhiqiang <Zhiqiang.Hou at nxp.com>
> >>>
> >>> Up to now, the function is_serdes_configed() doesn't check if the map
> >>> of serdes protocol is initialized before accessing it. The function
> >>> is_serdes_configed() will get wrong result when it was called before
> >>> the serdes protocol maps initialized. As the first eliment of the map
> >>
> >> s/eliment/element
> >
> > Will fix my spelling mistakes next version, thanks a lot!
> >
> >>> isn't used for any device, so use it as the flag to indicate if the
> >>> map has been initialized.
> >>
> >> I am not sure it is safe to presume the first element is always not used. Take
> >> LS2080A as an example, the serdes map array is initialized by serdes_init(),
> which
> >> calls serdes_get_prtcl() to get the index of the array. With normal condition,
> the
> >> index shouldn't be zero. Zero is used as an error but it is not checked before
> setting
> >>
> >> serdes_prtcl_map[lane_prtcl] = 1;
> >>
> >
> > The first element of enum srds_prtcl 'NONE' is aim to describe a lane isn't
> assigned to
> > any device, and the array serdesn_prtcl_map is used to check if the specified
> device
> > selected by the current serdes protocol, right? Nobody will check if the device
> 'NONE'
> > has been configured, right? So it can be used to indicate if the
> serdes_prtcl_map has
> > been initialized.
> > Don't care whether the function serdes_get_prtcl() will return zero, just focus
> if the
> > serdes protocol map has been filled. For example, if the serdes protocol value
> read from
> > RCW doesn't match with any entry of the serdes_cfg_tbl, then all lane will be
> assigned to
> > 'NONE'. It still means the serdes protocol map has been filled, even if there is
> only the
> > serdesn_prtcl_map[NONE] was set to 1.
> >
> >> If you presumption is correct, and you want to use the first one as a flag, you
> >> probably need to check all of them to make sure errors are handled correctly,
> >> instead of setting the flag unexpected. Also it is important to document this
> idea
> >> so future platform code follows the same.
> >>
> >
> > Is it necessary to add it to a document, if add a comment to the
> serdesn_prtcl_map make it?
> >
> 
> If you don't want to add a document, please at least put some comments,
> in case we need to change some code in the future.
> 
> York