LKML Archive on
help / color / mirror / Atom feed
From: Tejun Heo <>
To: Alan <>
Subject: Re: Proposed changes for libata speed handling
Date: Mon, 15 Jan 2007 12:09:20 +0900	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <20070113100158.1d79ba9f@localhost.localdomain>

Alan wrote:
> O> Wouldn't it be better to have ->determine_xfer_mask() and
>> ->set_specific_mode() than having two somewhat overlapping callbacks?
>> Or is there some problem that can't be handled that way?
> I'm not sure I follow what you are suggesting - can you explain further.
> Right now ->set_mode does all the policy management with regards to
> picking the right modes which is sometimes done by the usual ATA rules
> and sometimes by card specific code.
> ->set_specific_mode does no policy work but merely sets up a mode.
> The default behaviour of ->set_mode() is the ATA mode selection by best
> mode available, and this function is normally not provided by a driver.
> The default behaviour of ->set_specific_mode() is to verify the mode is
> valid then issue ->set_pio|dma_mode() (for both devices in case a timing
> change on both is triggered). This function is overridable because of
> things like IT821x where the IDE mode is imaginary.

What I was thinking about was something like the following.

* ops

	unsigned int (*determine_xfer_mask)(struct ata_device *dev);
	int (*set_specific_mode)(struct ata_device *dev,
				unsigned int xfer_mode);

* during init and EH

	if (init) {
		ap->xfer_mask &= ops->determine_xfer_mask(dev);
		DETERMINE best_mode;

	if (ap->ehi.target_mode && valid)
		mode = ap->ehi.target_mode;
		mode = best_mode;

	rc = ops->set_specific_mode(dev, mode);

* when the user issues SET_XFERMODE, in the issue path

	if (command is SET_XFERMODE) {
		if (mode is invalid)
		ap->ehi.target_mode = user_specified_mode;

To sum up,

1. separate supported mode detection and mode programming such that we
can use the same programming path for both init and handling user-issued

2. group all mode programming callbacks (->set_piomode, ->set_dmamode
and ->post_set_mode into ->set_specific_mode) into ->set_specific_mode
to allow more flexibility and replace ->set_mode.

3. make sure all xfer mode programming is done in EH.  this will ease
support for weird controllers (e.g. cross-port synchronization).



  reply	other threads:[~2007-01-15  3:09 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-01-12 13:53 Alan
2007-01-13  2:02 ` Tejun Heo
2007-01-13 10:01   ` Alan
2007-01-15  3:09     ` Tejun Heo [this message]
2007-01-15 13:52       ` Jeff Garzik

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:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \ \
    --subject='Re: Proposed changes for libata speed handling' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).