All of lore.kernel.org
 help / color / mirror / Atom feed
From: Florian Fainelli <florian.fainelli@broadcom.com>
To: netdev-bot+sashiko@kernel.org
Cc: netdev@vger.kernel.org, nb@tipi-net.de, opendmb@gmail.com,
	bcm-kernel-feedback-list@broadcom.com, andrew@lunn.ch,
	hkallweit1@gmail.com, linux@armlinux.org.uk, davem@davemloft.net,
	edumazet@kernel.org, kuba@kernel.org, pabeni@redhat.com,
	linux-kernel@vger.kernel.org, kudzo51@gmail.com
Subject: Re: [PATCH net-next v3 1/4] net: phy: Map C45 probe read errors to -ENODEV during bus scan
Date: Thu, 19 Nov 2026 21:03:24 -0800	[thread overview]
Message-ID: <3d4bb44c-906e-4f77-9601-ab4b1f47879b@broadcom.com> (raw)
In-Reply-To: <179514208109.3136.8422011628328621271@kernel.org>



On 11/19/2026 6:34 PM, netdev-bot+sashiko@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 4 potential
> issue(s) to consider.
> 
> Critical: 0 · High: 0 · Medium: 1 · Low: 3 (1)
> 
> If you believe none of the issues are legitimate, please reply to each
> issue of Medium severity or higher and explain why the AI is wrong.
> If any of them are legitimate and you plan to respin, please let patchwork
> know by including "pw-bot: cr" as a separate line at the end of your reply
> (one such reply per series is enough).
> 
> - [Medium] On MDIO controllers that flag a missing turnaround as a bus
>    error, get_phy_c45_ids() now treats one failed Devices-In-Package read…
> - [Low] The subject says "during bus scan", but get_phy_c45_ids() is
>    reached through get_phy_device(bus, addr, true) for every C45 caller,…
> - [Low] The patch updates the kernel-doc of get_phy_c45_devs_in_pkg() but
>    leaves the return-value docs of get_phy_c45_ids()…
> - [Low] (Baseline: judged a false positive, kept only because
>    requires_validation=true.) The concern says the commit message's…
> 
> Pre-existing issues:
> - [Low] (Baseline: dismissed as a pre-existing low-severity issue, kept
>    only because requires_validation=true.) In the second loop of…
> 
> --
> 
> net: phy: Map C45 probe read errors to -ENODEV during bus scan
> 
> This commit makes get_phy_c45_devs_in_pkg() return the raw bus error. It
> maps -EIO/-ENODEV from the devices-in-package reads in get_phy_c45_ids()
> to -ENODEV, the same way get_phy_c22_id() does. The intent is to let
> mdiobus_scan_bus_c45() move past empty addresses on controllers that
> report an unanswered read as a bus error, instead of failing
> __mdiobus_register().
> 
>> net: phy: Map C45 probe read errors to -ENODEV during bus scan
> 
> [Severity: Low]
> The subject says "during bus scan". However, every C45 caller reaches
> get_phy_c45_ids() through get_phy_device(bus, addr, true). Does the new
> mapping also change behaviour outside the scan?
> 
> For DT/ACPI nodes compatible with "ethernet-phy-ieee802.3-c45",
> fwnode_mdiobus_register_phy() calls get_phy_device() directly:
> 
> drivers/net/mdio/fwnode_mdio.c:fwnode_mdiobus_register_phy() {
>      ...
> 	if (is_c45 || fwnode_get_phy_id(child, &phy_id))
> 		phy = get_phy_device(bus, addr, is_c45);
>      ...
> }
> 
> and __of_mdiobus_parse_phys() handles -ENODEV and -EIO differently:
> 
> drivers/net/mdio/of_mdio.c:__of_mdiobus_parse_phys() {
>      ...
> 		if (rc == -ENODEV)
> 			dev_err(&mdio->dev,
> 				"MDIO device at address %d is missing.\n",
> 				addr);
> 		else if (rc)
> 			goto exit;
>      ...
> }
> 
> Before this patch, a failed MDIO_DEVS1/MDIO_DEVS2 read made bus
> registration fail. Now it logs the "missing" message and the bus
> registers without that PHY.
> 
> SFP behaves the same way. sfp_sm_probe_phy() passes -ENODEV through.
> sfp_sm_main() then retries, eventually reports "no PHY detected" and
> starts the module. Before, an -EIO from mdio-i2c sent the module to
> SFP_S_FAIL.
> 
> Real controller faults reported as -EIO, such as a timed out C45 address
> phase, will also look like "no device" now.
> 
> This matches what get_phy_c22_id() already does. Could the commit message
> mention these callers as well?
> 
>> MDIO controllers that report a read from an empty address as a bus error
>> (e.g. MDIO_READ_FAIL on the UniMAC MDIO block) and that provide read_c45()
>> with unmasked addresses therefore cannot be registered: the C45 scan fails
>> on the first empty address.
> 
> [Severity: Low]
> This isn't a bug, but at this point in the series mdio-bcm-unimac.c does
> not set bus->read_c45, so the UniMAC example cannot reach
> mdiobus_scan_bus_c45() yet.
> 
> The later patch "net: mdio: mdio-bcm-unimac: Add Clause 45 support" adds:
> 
> 	bus->read_c45 = unimac_mdio_read_c45;
> 	bus->write_c45 = unimac_mdio_write_c45;
> 
> and its C45 read path returns -EIO on MDIO_READ_FAIL. So the premise
> holds by the end of the series, and the sentence above is already
> conditional on read_c45() being provided.
> 
> Could the commit message say that the UniMAC case is enabled by that
> follow-up patch?
> 
>> diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
>> index b39e945cebd73..4071041e1051d 100644
>> --- a/drivers/net/phy/phy_device.c
>> +++ b/drivers/net/phy/phy_device.c
>> @@ -780,7 +780,7 @@ static int phy_c45_probe_present(struct mii_bus *bus, int prtad, int devad)
>>    * Description: reads devices in package registers of a MMD at @dev_addr
>>    * from PHY at @addr on @bus.
>>    *
>> - * Returns: 0 on success, -EIO on failure.
>> + * Returns: 0 on success, or the negative error code from the bus read.
>>    */
> 
> [Severity: Low]
> This kernel-doc is updated. Should the return value docs of
> get_phy_c45_ids() and of the exported phy_get_c45_ids() be updated too?
> The phy_get_c45_ids() doc still says:
> 
>   * Returns zero on success, %-EIO on bus access error, or %-ENODEV if
>   * the "devices in package" is invalid.
> 
> and get_phy_c45_ids() still documents "%-EIO on bus access error". With
> this patch, a bus -EIO on the devices-in-package reads is returned as
> -ENODEV.
> 
> The only user outside this file, gpy_probe() in mxl-gpy.c, returns that
> value directly from probe():
> 
> 	if (!phydev->is_c45) {
> 		ret = phy_get_c45_ids(phydev);
> 		if (ret < 0)
> 			return ret;
> 	}
> 
> A real bus -EIO there now becomes -ENODEV. The driver core treats that as
> a quiet match rejection at dev_dbg level, not a logged probe failure.
> 
> [ ... ]
> 
>> @@ -845,7 +858,8 @@ static int get_phy_c45_ids(struct mii_bus *bus, int addr,
>>   		}
>>   		phy_reg = get_phy_c45_devs_in_pkg(bus, addr, i, &devs_in_pkg);
>>   		if (phy_reg < 0)
>> -			return -EIO;
>> +			/* returning -ENODEV doesn't stop bus scanning */
>> +			return get_phy_c45_probe_err(phy_reg);
>>   	}
> 
> [Severity: Medium]
> On controllers that report a missing turnaround as a bus error, could
> this skip a real PHY whose first probed MMD doesn't answer?
> 
> The loop depends on its condition to step past unimplemented MMDs:
> 
> 	for (i = 1; i < MDIO_MMD_NUM && (devs_in_pkg == 0 ||
> 	     (devs_in_pkg & 0x1fffffff) == 0x1fffffff); i++) {
> 
> The MMD 0 fallback below exists for PHYs such as the Cortina
> CS4315/CS4340. Both of these assume that an absent MMD reads back as
> 0xffff.
> 
> With the later C45 patch, UniMAC checks TA, and unimac_mdio_do_read()
> returns -EIO for any (addr, devad) pair that doesn't answer:
> 
> 	if (!(bus->phy_ignore_ta_mask & 1 << phy_id) && (cmd & MDIO_READ_FAIL)) {
> 		ret = -EIO;
> 		goto out;
> 	}
> 
> So if the MMD 1 read fails, get_phy_c45_ids() now returns -ENODEV here.
> No later MMD and no MMD 0 fallback is tried. mdiobus_scan_bus_c45() then
> treats the address as empty:
> 
> 			if (IS_ERR(phydev) && (PTR_ERR(phydev) != -ENODEV))
> 				return PTR_ERR(phydev);
> 
> The DT path through fwnode_mdiobus_register_phy() reaches the same code.
> 
> These cases look affected:
> - a PHY with no MMD 1 (PMA/PMD)
> - a PHY whose MMD 1 reads 0 and whose next MMD doesn't answer
> - a PHY that only answers on MMD 0
> 
> I haven't confirmed an in-tree controller and PHY pair that does this.
> 
> Before this patch the same setup made bus registration fail with -EIO, so
> no configuration that worked before breaks. With the patch, the PHY is
> left off the bus without any message.
> 
> Would it be more robust to treat an -EIO/-ENODEV devices-in-package read
> as "this MMD is absent" (devs_in_pkg set to all ones) and keep walking?
> -ENODEV would then be returned only after the MMD 0 read has also failed.
> 
>> @@ -855,7 +869,7 @@ static int get_phy_c45_ids(struct mii_bus *bus, int addr,
>>   		 */
>>   		phy_reg = get_phy_c45_devs_in_pkg(bus, addr, 0, &devs_in_pkg);
>>   		if (phy_reg < 0)
>> -			return -EIO;
>> +			return get_phy_c45_probe_err(phy_reg);
>>   
>>   		/* no device there, let's get out of here */
>>   		if ((devs_in_pkg & 0x1fffffff) == 0x1fffffff)
> 
> [Severity: Low]
> This is a pre-existing issue, not one introduced by this patch. Further
> down in get_phy_c45_ids(), the second loop returns the
> phy_c45_probe_present() error unchanged:
> 
> 		if (i == MDIO_MMD_VEND1 || i == MDIO_MMD_VEND2) {
> 			...
> 			ret = phy_c45_probe_present(bus, addr, i);
> 			if (ret < 0)
> 				return ret;
> 
> The MII_PHYSID1/MII_PHYSID2 reads right after it return -EIO instead.
> This path can also return an errno outside the documented 0/-EIO/-ENODEV
> set.
> 
> The commit message says reads made after a device has answered are still
> reported as -EIO. Should this path be normalized the same way?

Because this one is a pre-existing issue, it won't be addressed. All of 
the other comments incorporated into v4.

pw-bot: cr
-- 
Florian


  reply	other threads:[~2026-11-20  5:03 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 23:31 [PATCH net-next v3 0/4] net: mdio: Clause 45 support for mdio-bcm-unimac Florian Fainelli
2026-11-17 23:31 ` [PATCH net-next v3 1/4] net: phy: Map C45 probe read errors to -ENODEV during bus scan Florian Fainelli
2026-11-20  2:34   ` netdev-bot+sashiko
2026-11-20  5:03     ` Florian Fainelli [this message]
2026-11-17 23:31 ` [PATCH net-next v3 2/4] net: mdio: mdio-bcm-unimac: Define MDIO_OP_SHIFT Florian Fainelli
2026-11-17 23:31 ` [PATCH net-next v3 3/4] net: mdio: mdio-bcm-unimac: Poll for completion at 1us granularity Florian Fainelli
2026-11-17 23:31 ` [PATCH net-next v3 4/4] net: mdio: mdio-bcm-unimac: Add Clause 45 support Florian Fainelli
2026-11-18  8:49   ` Nicolai Buchwitz
2026-11-20  2:34   ` netdev-bot+sashiko
2026-11-20  5:15     ` Florian Fainelli

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=3d4bb44c-906e-4f77-9601-ab4b1f47879b@broadcom.com \
    --to=florian.fainelli@broadcom.com \
    --cc=andrew@lunn.ch \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@kernel.org \
    --cc=hkallweit1@gmail.com \
    --cc=kuba@kernel.org \
    --cc=kudzo51@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@armlinux.org.uk \
    --cc=nb@tipi-net.de \
    --cc=netdev-bot+sashiko@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=opendmb@gmail.com \
    --cc=pabeni@redhat.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.