Re: [PATCH] I/O space write barrier

[Jeremy Higdon <jeremy@sgi.com>]

On Thu, Sep 23, 2004 at 03:07:26PM -0400, Jesse Barnes wrote:
> On Thursday, September 23, 2004 3:03 pm, James Bottomley wrote:
> > On Thu, 2004-09-23 at 14:48, Jesse Barnes wrote:
> > > - RD_REG_WORD(&reg->ictrl); /* PCI Posted Write flush */
> > > + mmiowb(); /* make sure this write arrives before any others */
> >
> > What's going on here?  In your docs you say this mmiowb() is explicitly
> > to prevent write ordering problems and doesn't affect posting.  However
> > in this patch you're using it to replace a posting flush read.
> 
> Jeremy was the one that added these, I think the comment is misleading and 
> only ordering was intended.  If that's not the case I'll drop this bit.

No, I didn't add those.  In fact, I thought them to be unnecessary.  Jes
added them, I believe.  I added the reads after the writes of req_ring_index 
to mailbox4.  Since at the time we did not have a write ordering function,
I had to use the heavy weight flush.

Take a look at the 1.56 diffs in the scsi_misc-2.6 tree.

jeremy