From: Niklas Cassel <cassel@kernel.org>
To: yangxingui <yangxingui@huawei.com>
Cc: Damien Le Moal <dlemoal@kernel.org>,
linux-kernel@vger.kernel.org, liuyonglong@huawei.com,
kangfenglong@huawei.com, linux-ide@vger.kernel.org
Subject: Re: [PATCH] ata: libata-sata: retry hardreset when device detected but PHY not established
Date: Thu, 30 Apr 2026 10:46:18 +0200 [thread overview]
Message-ID: <afMW2scJrBSI_jLo@ryzen> (raw)
In-Reply-To: <c78f6dad-b0fe-a2fb-8458-9fe182cd6b54@huawei.com>
On Wed, Apr 29, 2026 at 03:01:48PM +0800, yangxingui wrote:
> > > > > > This is preceeded by a call to sata_link_resume(), which calls
> > > > > > sata_link_debounce() and that function makes sure that DET is stable. So if
> > > > > > after that DET still shows that their is no PHY, there is likely a big problem
> > > > > > with it and it is super slow to be established.
I agree with Damien, sata_link_debounce() is supposed to make sure that
DET is stable.
sata_link_debounce() will not explicitly wait for SStatus.DET to turn 0x3.
If value is stable, and SStatus.DET == 1, and time is before "deadline",
sata_link_debounce() will continue looping.
Else, if value is stable, and has been stable for "duration" amount of time,
it will return.
Since your print shows that SStatus == 1, that most likely means that the
deadline expired in sata_link_debounce().
I suggest that you try to increase the deadline, perhaps start off by simply
multiplying it by some factor in sata_link_debounce().
It would also be helpful if your commit message explained why returning
-EAGAIN makes a difference, because from what I can see, if the deadline
expires, sata_link_debounce() returns 0, which should cause sata_link_resume()
to return 0, which should cause sata_link_hardreset() to
return 0, with online == false.
If that is the case ata_do_reset() would return 0, and
ata_eh_followup_srst_needed() (returns true only if -EAGAIN) would return false.
Which should eventually cause us to retry another hard reset, as long as
tries <= max_tries.
By making sata_link_hardreset() return -EAGAIN, the difference I see is that
we will for a software reset followed by the hardreset, but you commit message
did not mention that.
So, my question is, why is it not sufficient to retry another
hardreset/COMRESET?
Does it work to only do a hardreset (without if a follow up softreset) if you
increase the deadline?
Kind regards,
Niklas
next prev parent reply other threads:[~2026-04-30 8:46 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-25 6:04 [PATCH] ata: libata-sata: retry hardreset when device detected but PHY not established Xingui Yang
2026-04-25 22:53 ` Damien Le Moal
2026-04-27 1:51 ` yangxingui
2026-04-27 4:45 ` Damien Le Moal
2026-04-29 1:14 ` yangxingui
2026-04-29 1:36 ` Damien Le Moal
2026-04-29 7:01 ` yangxingui
2026-04-30 8:46 ` Niklas Cassel [this message]
2026-04-30 9:28 ` Niklas Cassel
2026-05-06 2:23 ` yangxingui
2026-04-27 13:17 ` Niklas Cassel
2026-04-29 1:06 ` yangxingui
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=afMW2scJrBSI_jLo@ryzen \
--to=cassel@kernel.org \
--cc=dlemoal@kernel.org \
--cc=kangfenglong@huawei.com \
--cc=linux-ide@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=liuyonglong@huawei.com \
--cc=yangxingui@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.