[U-Boot] [Patch V7 04/18] net/fm/eth: Use mb() to be compatible for both ARM and PowerPC

[Hou Zhiqiang]

> -----Original Message-----
> From: Wood Scott-B07421
> Sent: 2015年10月23日 7:11
> To: Gong Qianyu-B52263
> Cc: u-boot at lists.denx.de; Hu Mingkai-B21284; Sun York-R58495; Hou
> Zhiqiang-B48286; Xie Shaohui-B21989; Song Wenbin-B53747; Wood Scott-
> B07421; Kushwaha Prabhakar-B32579; Wang Huan-B18965
> Subject: Re: [Patch V7 04/18] net/fm/eth: Use mb() to be compatible for
> both ARM and PowerPC
> 
> On Thu, 2015-10-22 at 18:46 +0800, Gong Qianyu wrote:
> > From: Shaohui Xie <Shaohui.Xie at freescale.com>
> >
> > Use mb() instead of sync() to be compatible for both ARM and PowerPC.
> >
> > Signed-off-by: Shaohui Xie <Shaohui.Xie at freescale.com>
> > Signed-off-by: Mingkai Hu <Mingkai.Hu at freescale.com>
> > Signed-off-by: Gong Qianyu <Qianyu.Gong at freescale.com>
> > ---
> > V7:
> >  - No change.
> > V6:
> >  - No change.
> > V5:
> >  - No change.
> > V4:
> >  - No change.
> > V3:
> >  - New patch. Separated from patch 'net: Move some header files to
> include/'
> >
> >  drivers/net/fm/eth.c | 14 +++++++-------
> >  1 file changed, 7 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/net/fm/eth.c b/drivers/net/fm/eth.c index
> > 368d554..ad02c66 100644
> > --- a/drivers/net/fm/eth.c
> > +++ b/drivers/net/fm/eth.c
> > @@ -371,7 +371,7 @@ static void
> > fmc_tx_port_graceful_stop_enable(struct
> > fm_eth *fm_eth)
> >       pram = fm_eth->tx_pram;
> >       /* graceful stop transmission of frames */
> >       setbits_be32(&pram->mode, PRAM_MODE_GRACEFUL_STOP);
> > -     sync();
> > +     mb();
> >  }
> >
> >  static void fmc_tx_port_graceful_stop_disable(struct fm_eth *fm_eth)
> > @@ -381,7 +381,7 @@ static void
> > fmc_tx_port_graceful_stop_disable(struct
> > fm_eth *fm_eth)
> >       pram = fm_eth->tx_pram;
> >       /* re-enable transmission of frames */
> >       clrbits_be32(&pram->mode, PRAM_MODE_GRACEFUL_STOP);
> > -     sync();
> > +     mb();
> >  }
> 
> Why is it needed at all?  The I/O accessors should include the necessary
> barriers.

The I/O accessors of powerpc does include the barrier, but it absents from ARMs'.

Thanks,
Zhiqiang