linux-kernel - Re: [RESEND PATCH v7 2/5] leds: sun50i-a100: New driver for the A100 LED controller

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20231031074807.GQ8909@google.com>
Date:   Tue, 31 Oct 2023 07:48:07 +0000
From:   Lee Jones <lee@...nel.org>
To:     Samuel Holland <samuel@...lland.org>
Cc:     Pavel Machek <pavel@....cz>, linux-leds@...r.kernel.org,
        Chen-Yu Tsai <wens@...e.org>,
        Jernej Skrabec <jernej.skrabec@...il.com>,
        linux-kernel@...r.kernel.org, linux-sunxi@...ts.linux.dev
Subject: Re: [RESEND PATCH v7 2/5] leds: sun50i-a100: New driver for the A100
 LED controller

On Sun, 29 Oct 2023, Samuel Holland wrote:

> Hi Lee,
> 
> On 10/17/23 19:38, Samuel Holland wrote:
> > On 3/16/23 08:34, Lee Jones wrote:
> >> On Sat, 31 Dec 2022, Samuel Holland wrote:
> >>> +	for_each_available_child_of_node(np, child) {
> >>> +		struct sun50i_a100_ledc_led *led;
> >>> +		struct led_classdev *cdev;
> >>> +		u32 addr, color;
> >>> +
> >>> +		ret = of_property_read_u32(child, "reg", &addr);
> >>> +		if (ret || addr >= count) {
> >>> +			dev_err(dev, "LED 'reg' values must be from 0 to %d\n",
> >>
> >> Doesn't sounds like an address.
> > 
> > The one-wire protocol involves the first LED responding to the first 24
> > bits of the transfer, then forwarding the rest to the next LED. The
> > address is the ordinal position in the chain. So I don't think there is
> > any reason to have gaps in the addresses--the LEDs would still have to
> > physically be there, but you would not be able to control them. That
> > said, the driver doesn't need to enforce this, so I can remove the check.
> 
> There's actually a reason for the driver enforcing that LED addresses
> are contiguous. Removing this check decouples the largest address from
> the number of LED class devices. Unfortunately, there's a register field
> (LEDC_RESET_TIMING_CTRL_REG_LED_NUM) that must be set to the largest
> address for transfers to work correctly.
> 
> This means the driver would need to iterate through the child nodes in
> two passes inside the probe function: first to find the largest reg
> value, and second to actually register the LED class devices.
> 
> Since I don't think having gaps in the addresses is a realistic use
> case, I'd like to keep this restriction for now (with a comment). We
> could always pay the complexity cost later if someone really does want
> to relax this check.

You're replying to a review conducted 7 months ago which in turn was for
a patch submitted nearly a year ago!

Also the review comment was simple - if it's not an address, rename the
variable.

-- 
Lee Jones [李琼斯]