mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: John Garry <john.g.garry@oracle.com>
To: Xingui Yang <yangxingui@huawei.com>,
	yanaijie@huawei.com, jejb@linux.ibm.com,
	martin.petersen@oracle.com
Cc: linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org,
	linuxarm@huawei.com, liyihang9@h-partners.com,
	liuyonglong@huawei.com, kangfenglong@huawei.com
Subject: Re: [PATCH v4 2/2] scsi: libsas: Add linkrate and sas_addr change detection in rediscover
Date: Fri, 29 May 2026 15:09:03 +0100	[thread overview]
Message-ID: <cd70e2a4-91e1-4237-bdd0-7568b2dbdc9f@oracle.com> (raw)
In-Reply-To: <20260526015418.2022398-3-yangxingui@huawei.com>

On 26/05/2026 02:54, Xingui Yang wrote:
> In sas_rediscover_dev(), when detecting a "flutter" condition (same SAS
> address and compatible device type), the code assumes the device remains
> unchanged and only handles SATA pending state recovery. However, this
> approach misses two important scenarios:
> 
> First, the flutter detection only compares SAS address and device type,
> ignoring potential linkrate changes that may have already occurred.
> 
> Second, after sas_ex_phy_discover() re-queries the expander phy, both
> linkrate and attached SAS address may be updated. The current code does
> not validate these changes against the existing child device.
> 
> Additionally, the replace code path (different SAS address detected)
> has a sysfs duplication issue: sas_unregister_devs_sas_addr() only marks
> the device as gone, but the actual sysfs cleanup happens later in
> sas_destruct_devices(). Calling sas_discover_new() immediately after
> unregister causes sysfs_warn_dup() errors.
> 
> Introduce sas_is_flutter() to check whether it is a true flutter with
> validation for linkrate and sas_addr changes. It returns true for normal
> flutter and false when changes are detected requiring rediscovery.
> 
> Introduce sas_rediscover_phy() to handle async rediscovery for both
> flutter and replace cases. When invoked:
> - Set phy_change_count and ex_change_count to -1 to force revalidation
> - Unregister the device via sas_unregister_devs_sas_addr()
> - Queue DISCE_REVALIDATE_DOMAIN event
> 
> The old device sysfs is cleaned up by sas_destruct_devices() at the end
> of current revalidation work. The new event triggers discovery via
> sas_discover_new() since attached_sas_addr is cleared, avoiding the
> sysfs duplication issue.
> 
> Signed-off-by: Xingui Yang <yangxingui@huawei.com>
> Suggested-by: John Garry <john.g.garry@oracle.com>
> ---
>   drivers/scsi/libsas/sas_expander.c | 67 +++++++++++++++++++++++-------
>   1 file changed, 53 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/scsi/libsas/sas_expander.c b/drivers/scsi/libsas/sas_expander.c
> index f55ae9a979cd..4e8e1b339889 100644
> --- a/drivers/scsi/libsas/sas_expander.c
> +++ b/drivers/scsi/libsas/sas_expander.c
> @@ -1962,6 +1962,56 @@ static bool dev_type_flutter(enum sas_device_type new, enum sas_device_type old)
>   	return false;
>   }
>   
> +static void sas_rediscover_phy(struct domain_device *dev, int phy_id,

most of the expander phy symbols have _ex_phy ending

> +			       bool last)
> +{
> +	struct expander_device *ex = &dev->ex_dev;
> +	struct ex_phy *phy = &ex->ex_phy[phy_id];
> +
> +	phy->phy_change_count = -1;
> +	ex->ex_change_count = -1;
> +	sas_unregister_devs_sas_addr(dev, phy_id, last);
> +	sas_discover_event(dev->port, DISCE_REVALIDATE_DOMAIN);
> +}
> +
> +static bool sas_is_flutter(struct domain_device *dev, int phy_id,

sas_dev_is_flutter may be a better name

> +			   u8 *sas_addr, enum sas_device_type type)
> +{
> +	struct expander_device *ex = &dev->ex_dev;
> +	struct ex_phy *phy = &ex->ex_phy[phy_id];
> +	struct domain_device *child_dev;
> +	char *action = "";
> +
> +	if (SAS_ADDR(sas_addr) != SAS_ADDR(phy->attached_sas_addr) ||
> +	    !dev_type_flutter(type, phy->attached_dev_type))
> +		return false;
> +
> +	child_dev = sas_ex_to_dev(dev, phy_id);
> +
> +	sas_ex_phy_discover(dev, phy_id);

why not check the return code for error?

> +
> +	if (child_dev && dev_is_sata(child_dev) &&
> +	    phy->attached_dev_type == SAS_SATA_PENDING) {
> +		action = ", needs recovery";
> +	} else if (child_dev && child_dev->linkrate != phy->linkrate) {
> +		pr_info("ex %016llx phy%02d linkrate changed from %d to %d\n",
> +			SAS_ADDR(dev->sas_addr), phy_id,
> +			child_dev->linkrate, phy->linkrate);
> +		return false;
> +	} else if (child_dev &&

Can you factor out the child_dev checks for all if/else legs?

> +		   SAS_ADDR(child_dev->sas_addr) != SAS_ADDR(phy->attached_sas_addr)) {
> +		pr_info("ex %016llx phy%02d sas_addr changed from %016llx to %016llx\n",
> +			SAS_ADDR(dev->sas_addr), phy_id,
> +			SAS_ADDR(child_dev->sas_addr),
> +			SAS_ADDR(phy->attached_sas_addr));
> +		return false;
> +	}
> +
> +	pr_debug("ex %016llx phy%02d broadcast flutter%s\n",
> +		 SAS_ADDR(dev->sas_addr), phy_id, action);
> +	return true;
> +}
> +
>   static int sas_rediscover_dev(struct domain_device *dev, int phy_id,
>   			      bool last, int sibling)
>   {
> @@ -2015,27 +2065,16 @@ static int sas_rediscover_dev(struct domain_device *dev, int phy_id,
>   		if (res == 0)
>   			sas_set_ex_phy(dev, phy_id, disc_resp);
>   		goto out_free_resp;
> -	} else if (SAS_ADDR(sas_addr) == SAS_ADDR(phy->attached_sas_addr) &&
> -		   dev_type_flutter(type, phy->attached_dev_type)) {
> -		struct domain_device *ata_dev = sas_ex_to_ata(dev, phy_id);
> -		char *action = "";
> -
> -		sas_ex_phy_discover(dev, phy_id);
> +	}
>   
> -		if (ata_dev && phy->attached_dev_type == SAS_SATA_PENDING)
> -			action = ", needs recovery";
> -		pr_debug("ex %016llx phy%02d broadcast flutter%s\n",
> -			 SAS_ADDR(dev->sas_addr), phy_id, action);
> +	if (sas_is_flutter(dev, phy_id, sas_addr, type))
>   		goto out_free_resp;
> -	}
>   
>   	/* we always have to delete the old device when we went here */
>   	pr_info("ex %016llx phy%02d replace %016llx\n",
>   		SAS_ADDR(dev->sas_addr), phy_id,
>   		SAS_ADDR(phy->attached_sas_addr));
> -	sas_unregister_devs_sas_addr(dev, phy_id, last);
> -
> -	res = sas_discover_new(dev, phy_id);
> +	sas_rediscover_phy(dev, phy_id, last);
>   out_free_resp:
>   	kfree(disc_resp);
>   	return res;

Can res still hold non-zero value from earlier?


  reply	other threads:[~2026-05-29 14:09 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-26  1:54 [PATCH v4 0/2] libsas: rediscover improvements for linkrate/sas_addr changes Xingui Yang
2026-05-26  1:54 ` [PATCH v4 1/2] scsi: libsas: refactor sas_ex_to_ata() using new helper sas_ex_to_dev() Xingui Yang
2026-05-26  1:54 ` [PATCH v4 2/2] scsi: libsas: Add linkrate and sas_addr change detection in rediscover Xingui Yang
2026-05-29 14:09   ` John Garry [this message]
2026-05-30  2:04     ` 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=cd70e2a4-91e1-4237-bdd0-7568b2dbdc9f@oracle.com \
    --to=john.g.garry@oracle.com \
    --cc=jejb@linux.ibm.com \
    --cc=kangfenglong@huawei.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=linuxarm@huawei.com \
    --cc=liuyonglong@huawei.com \
    --cc=liyihang9@h-partners.com \
    --cc=martin.petersen@oracle.com \
    --cc=yanaijie@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®