RE: [EXT] Re: [PATCH] mtd: spi-nor: Add prep/unprep for spi_nor_resume

[Han Xu]

> -----Original Message-----
> From: Tudor.Ambarus@xxxxxxxxxxxxx <Tudor.Ambarus@xxxxxxxxxxxxx>
> Sent: Thursday, June 6, 2019 12:46 AM
> To: Han Xu <han.xu@xxxxxxx>; cyrille.pitchen@xxxxxxxxxx;
> marek.vasut@xxxxxxxxx
> Cc: boris.brezillon@xxxxxxxxxxxxxxxxxx; f.fainelli@xxxxxxxxx;
> kdasu.kdev@xxxxxxxxx; richard@xxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
> linux-mtd@xxxxxxxxxxxxxxxxxxx; dl-linux-imx <linux-imx@xxxxxxx>;
> computersforpeace@xxxxxxxxx
> Subject: [EXT] Re: [PATCH] mtd: spi-nor: Add prep/unprep for
> spi_nor_resume
> 
> Caution: EXT Email
> 
> Hi, Han,
> 
> On 04/25/2019 12:08 AM, Han Xu wrote:
> > External E-Mail
> >
> >
> > In the new implemented spi_nor_resume function, the spi_nor_init()
> > should be braced by prep/unprep functions._
> >
> 
> Would you please explain why this is needed? Have you tried a
> suspend/resume cycle while a write was in progress and it failed?

Same as all other functions that hooked up with mtd ops, such as
_read/_write/_lock/_unlock, the _resume function also need to call the
lower level controller's prep/unprep functions to properly handle runtime
pm and mutex.

I didn't try suspend/resume during writing, but found accessing registers
without enabling clock when system resumed.

> 
> Thanks,
> ta