linux-kernel - Re: [PATCH v2 2/2] phy: qualcomm: phy-qcom-eusb2-repeater: rework reg override handler

Open Source and information security mailing list archives

Message-ID: <n5z5atbxkwqllni6cgygcw5nn7z4w3yjznhpkaajw6tggenxb3@kxpk2e6vmvlj>
Date: Mon, 28 Apr 2025 19:58:04 +0300
From: Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com>
To: Pengyu Luo <mitltlatltl@...il.com>
Cc: abel.vesa@...aro.org, conor+dt@...nel.org, devicetree@...r.kernel.org,
        kishon@...nel.org, krzk+dt@...nel.org, linux-arm-msm@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-phy@...ts.infradead.org,
        robh@...nel.org, sfr@...b.auug.org.au, vkoul@...nel.org
Subject: Re: [PATCH v2 2/2] phy: qualcomm: phy-qcom-eusb2-repeater: rework
 reg override handler

On Sat, Apr 26, 2025 at 04:14:23PM +0800, Pengyu Luo wrote:
> On Sat, Apr 26, 2025 at 3:41 AM Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com> wrote:
> > On Wed, Apr 16, 2025 at 08:02:01PM +0800, Pengyu Luo wrote:
> > > In downstream tree, many registers need to be overridden, it varies
> > > from devices and platforms, with these registers getting more, adding
> > > a handler for this is helpful.
> >
> > It should be noted that previously all values were applied during _init
> > phase, before checking the status etc. Now the overrides are programmed
> > from the set_mode. Should you still program sane defaults at the init
> > stage too?
> >
> 
> I think programming in set_mode is ok. When we init(dwc3_core_init), we
> set_mode(dwc3_core_init_mode) later, please check
> https://elixir.bootlin.com/linux/v6.14.3/source/drivers/usb/dwc3/core.c#L2287

Yes, but that happens after reading status regs, etc.

> Actually, in the downstream, all the things are done in init, it
> overrides first, then masked write the deaults, finally it set_mode,
> you can check here
> https://github.com/OnePlusOSS/android_kernel_oneplus_sm8650/blob/oneplus/sm8650_v_15.0.0_pad_pro/drivers/usb/repeater/repeater-qti-pmic-eusb2.c#L356

I'd stick to this approach too. Program everything in init, then
program mode-dependent regs in set_mode.

> 
> > BTW, is there a real need to override those for the platform you are
> > working on? Could you please provide some details, maybe in the cover
> > letter.
> 
> I am not quite sure, recently, I expirenced mode switching failure,
> when I `echo device > /sys/kernel/debug/usb/a600000.usb/mode`, Ethernet
> Gadget wouldn't work again, my desktop can't connect to it.

Do you have at least a list of the properties / registers that
downstream programs on your platform? I mean, it's not infrequent that
vendor kernel is more versatile than necessary, as it is being used
during bringup / etc. I'd suggest to limit supported overrides to those
necessary for your platform (and add a comment that there were other
available).

> 
> BTW, as you can see in line 356, it is most likely that overrides
> related to charging feature. I have not ported charging yet, but adding
> more overrides seems harmless, and if overriding, distinguishing which
> mode seems necessary, even if some devices use the same sequence. So I
> sent the patch.
> 
> Best wishes,
> Pengyu
> 
> -- 
> linux-phy mailing list
> linux-phy@...ts.infradead.org
> https://lists.infradead.org/mailman/listinfo/linux-phy

-- 
With best wishes
Dmitry

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives