[PATCH v2] net: fsl_mdio: Fix busy flag polling register

[Camelia Alexandra Groza (OSS)]

> -----Original Message-----
> From: Markus Koch <markus at notsyncing.net>
> Sent: Friday, January 7, 2022 19:49
> To: Camelia Alexandra Groza (OSS) <camelia.groza at oss.nxp.com>;
> joe.hershberger at ni.com; rfried.dev at gmail.com; Ioana Ciornei
> <ioana.ciornei at nxp.com>
> Cc: Madalin Bucur (OSS) <madalin.bucur at oss.nxp.com>; u-
> boot at lists.denx.de
> Subject: Re: [PATCH v2] net: fsl_mdio: Fix busy flag polling register
> 
> On 1/7/22 17:23, Camelia Alexandra Groza (OSS) wrote:
> >> -----Original Message-----
> >> From: U-Boot <u-boot-bounces at lists.denx.de> On Behalf Of Markus
> Koch
> >> Sent: Tuesday, January 4, 2022 17:42
> >> To: Ioana Ciornei <ioana.ciornei at nxp.com>; joe.hershberger at ni.com;
> >> rfried.dev at gmail.com
> >> Cc: Madalin Bucur (OSS) <madalin.bucur at oss.nxp.com>; Camelia
> Alexandra
> >> Groza <camelia.groza at nxp.com>; u-boot at lists.denx.de; Markus Koch
> >> <markus at notsyncing.net>
> >> Subject: [PATCH v2] net: fsl_mdio: Fix busy flag polling register
> >>
> >> NXP's mEMAC reference manual, Chapter 6.5.5 "MDIO Ethernet
> >> Management
> >> Interface usage", specifies to poll the BSY (0) bit in the CFG/STAT
> >> register to wait until a transaction has finished, not bit 31 in the
> >> data register.
> >>
> >> In the Linux kernel, this has already been fixed in commit 26eee0210ad7
> >> ("net/fsl: fix a bug in xgmac_mdio").
> >>
> >> Signed-off-by: Markus Koch <markus at notsyncing.net>
> >
> > I am ok with the change for the mEMAC driver but MDIO_DATA_BSY is still
> > used by the fsl_ls_mdio driver. The MDIO driver suffers from the same
> issue.
> >
> > Please send a v3 and either patch both drivers or don't remove the define.
> > Also, if removing the define, please mention it explicitly in the patch
> > description.
> 
> I'll change the other driver as well.
> 
> There's also the fm/tgec_phy driver that uses the same construct, but it has
> its own define for MDIO_DATA_BSY. Should I also change that? I think it
> corresponds to Linux' xgmac_mdio.c which also uses the STAT_BSY register,
> but
> I don't have a datasheet to verify the register definition.

The TGEC controller has BUSY bits mirroring each other in both the 
MDIO_DATA and MDIO_CFG_STAT registers. The tgec_phy driver doesn't need
changing.

Regards,
Camelia 
 
> Thanks,
> Markus
> 
> > Thanks,
> > Camelia
> >
> >> ---
> >>
> >> Changed to use the mdio_stat register. Thanks, Ioana!
> >>
> >>   drivers/net/fm/memac_phy.c | 2 +-
> >>   include/fsl_memac.h        | 1 -
> >>   2 files changed, 1 insertion(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/net/fm/memac_phy.c b/drivers/net/fm/memac_phy.c
> >> index 72b500a6d1..3ddae97e09 100644
> >> --- a/drivers/net/fm/memac_phy.c
> >> +++ b/drivers/net/fm/memac_phy.c
> >> @@ -64,7 +64,7 @@ static int memac_wait_until_done(struct
> >> memac_mdio_controller *regs)
> >>   {
> >>   	unsigned int timeout = MAX_NUM_RETRIES;
> >>
> >> -	while ((memac_in_32(&regs->mdio_data) & MDIO_DATA_BSY) &&
> >> timeout--)
> >> +	while ((memac_in_32(&regs->mdio_stat) & MDIO_STAT_BSY) &&
> >> timeout--)
> >>   		;
> >>
> >>   	if (!timeout) {
> >> diff --git a/include/fsl_memac.h b/include/fsl_memac.h
> >> index d067f1511c..6ac1e558b9 100644
> >> --- a/include/fsl_memac.h
> >> +++ b/include/fsl_memac.h
> >> @@ -254,7 +254,6 @@ struct memac_mdio_controller {
> >>   #define MDIO_CTL_READ		(1 << 15)
> >>
> >>   #define MDIO_DATA(x)		(x & 0xffff)
> >> -#define MDIO_DATA_BSY		(1 << 31)
> >>
> >>   struct fsl_enet_mac;
> >>
> >> --
> >> 2.34.1
> >