Date: Wed, 16 Sep 2026 14:06:05 +0530
From: Mohd Ayaan Anwar <mohd.anwar@oss.qualcomm.com>
To: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>, stable@vger.kernel.org,
        patches@lists.linux.dev,
        Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>,
        Vinod Koul <vkoul@kernel.org>, Sasha Levin <sashal@kernel.org>,
        Vegard Nossum <vegard.nossum@oracle.com>
Subject: Re: [PATCH 6.12 0951/1376] phy: qcom: sgmii-eth: vote for both
 voltage rails with correct current loads
Message-ID: <aqpU9epE2YS12PRu@oss.qualcomm.com>
References: <20260912065607.535295758@linuxfoundation.org>
 <20260912065628.756028552@linuxfoundation.org>
 <a5153eb7-6047-4034-80a1-fb3c0b70c9ca@oracle.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <a5153eb7-6047-4034-80a1-fb3c0b70c9ca@oracle.com>

On Wed, Sep 16, 2026 at 01:38:54PM +0530, Harshit Mogalapalli wrote:
> 
> 
> On 12/09/26 12:26 pm, Greg Kroah-Hartman wrote:
> > 6.12-stable review patch.  If anyone has any objections, please let me know.
> > 
> > ------------------
> > 
> > From: Mohd Ayaan Anwar <mohd.anwar@oss.qualcomm.com>
> > 
> > [ Upstream commit 4f81684a1d1018d7d0e5579f08d95e7701279358 ]
<snip>
> >   #define QSERDES_COM_C_PLL_LOCKED			BIT(1)
> > +static const struct regulator_bulk_data qcom_dwmac_sgmii_phy_vregs[] = {
> > +	{ .supply = "vdda-0p9", .init_load_uA = 46000 },
> > +	{ .supply = "vdda-1p2", .init_load_uA = 15000 },
> 
> 
> Hi Greg/Sasha,
> 
> I ran an AI-assisted backport review, and I have verified the report. It
> looks valid to me.
> 
> Upstream driver code:
> 
>     { .supply = "vdda-0p9", .init_load_uA = 46000 },
>     { .supply = "vdda-1p2", .init_load_uA = 15000 },
> 
> Downstream 6.12.y board code:
> 
>     phy-supply = <&vreg_l5a>;
> 
> The downstream property matches neither required upstream supply name.
> 
> The driver now requires vdda-0p9 and vdda-1p2 supplies, but the enabled
> SA8775P SerDes nodes in 6.12 provide only phy-supply. Regulator lookup
> uses the exact named properties, so probe either fails or receives dummy
> supplies; it does not vote the physical rails which this fix requires.
> 
> Upstream companion 3719850f1b70dd adds that.
> 
> I suggest dropping the driver patch until that mapping is provided and
> validated. Thoughts?
> 
> thanks,
> Harshit> +};

Actually, phy-core.c already takes care of regulators described by
"phy-supply".

drivers/phy/phy-core.c
	/* phy-supply */
	phy->pwr = regulator_get_optional(&phy->dev, "phy");
	if (IS_ERR(phy->pwr)) {
		ret = PTR_ERR(phy->pwr);
		if (ret == -EPROBE_DEFER)
			goto put_dev;

		phy->pwr = NULL;
	}

This patch only adds support for two additional regulators
(vdda-0p9 and vdda-1p2). It does not remove or alter the existing
"phy-supply" path handled by phy-core.c.

As a result, boards that only describe "phy-supply" continue to work
as before. The new regulator lookups are optional and merely allow the
PHY driver to vote for the dedicated rails when they are available.

        Ayaan