Sender: Tejun Heo <htejun@gmail.com>
Date:   Fri, 7 Jan 2022 11:30:26 -1000
From:   Tejun Heo <tj@kernel.org>
To:     Jeff Garzik <jeff@bloq.com>
Cc:     Paul Menzel <pmenzel@molgen.mpg.de>,
        Damien Le Moal <damien.lemoal@opensource.wdc.com>,
        linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: ata: For what PHY was debounce delay introduced? (was: [PATCH v3
 3/3] ahci: AMD A85 FCH (Hudson D4): Skip 200 ms debounce delay in
 `sata_link_resume()`)
Message-ID: <Ydiw8hxkNWDO2nW3@slm.duckdns.org>
References: <20211229161119.1006-3-pmenzel@molgen.mpg.de>
 <5c333718-eaa5-b41c-e8ea-59d6e01254aa@opensource.wdc.com>
 <19f6cd93-9bd7-60dc-4471-18022bcf456c@molgen.mpg.de>
 <7b64d1c3-f798-d64b-9ee3-9669d98f4e28@opensource.wdc.com>
 <fbfd865f-c88c-6ee1-6cb9-8194e170cd3a@molgen.mpg.de>
 <c6748a52-fc8f-3309-31c2-973a9e69a7e8@opensource.wdc.com>
 <33deca4a-abed-123c-9530-3f15740a3261@molgen.mpg.de>
 <27da2f5c-ca6c-1d64-3d05-5453f11e298f@opensource.wdc.com>
 <66edbfd4-063f-b995-0d15-982d365dd7d7@molgen.mpg.de>
 <CAHy7=M10exWuVJtDVo+w36YKadY533GttzwrKxPnotHf8-JQnw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CAHy7=M10exWuVJtDVo+w36YKadY533GttzwrKxPnotHf8-JQnw@mail.gmail.com>
Precedence: bulk

Hello,

On Wed, Jan 05, 2022 at 03:49:58PM -0500, Jeff Garzik wrote:
> On Tue, Jan 4, 2022 at 6:34 AM Paul Menzel <pmenzel@molgen.mpg.de> wrote:
> 
> > [cc: +jeff, +tejun]
> >
> > void sata_phy_reset(struct ata_port *ap)
> > {
> > […]
> >         /* wait for phy to become ready, if necessary */
> >         do {
> >                 msleep(200);
> >                 sstatus = scr_read(ap, SCR_STATUS);
> >                 if ((sstatus & 0xf) != 1)
> >                         break;
> >         } while (time_before(jiffies, timeout));
> > […]
> > }
> > ```
> >
> 
> The piix did not have SCRs, as I recall, so it wouldn't apply to those
> chips.   I don't recall further than that.
> 
> Presumably just give those early chips a "needs delay" quirk, and then
> start testing newer chips to make sure they survive an immediate bitbang?

I don't remember exactly but most likely the sata_sil chips, I think. But,
yeah, the way forward would be converting it to a quirk and gradually lift
them with tests.

Thanks and happy new year.

-- 
tejun