linux-kernel - Re: [PATCH v2 1/2] platform_device: add devres function region-reqs

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] [day] [month] [year] [list]

Message-ID: <da747c7ca3424cfa5a0d7ecf1733dc1a7303a9ee.camel@redhat.com>
Date: Tue, 09 Jan 2024 10:37:56 +0100
From: Philipp Stanner <pstanner@...hat.com>
To: Uwe Kleine-König <u.kleine-koenig@...gutronix.de>
Cc: dri-devel@...ts.freedesktop.org, Fabio Estevam <festevam@...il.com>, 
 Daniel Vetter <daniel@...ll.ch>, "Rafael J. Wysocki" <rafael@...nel.org>,
 Takashi Iwai <tiwai@...e.de>, Greg Kroah-Hartman
 <gregkh@...uxfoundation.org>, Sascha Hauer <s.hauer@...gutronix.de>,
 Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
 linux-kernel@...r.kernel.org, Maxime Ripard <mripard@...nel.org>, Mark
 Brown <broonie@...nel.org>, NXP Linux Team <linux-imx@....com>,  Thomas
 Zimmermann <tzimmermann@...e.de>, Laurentiu Palcu
 <laurentiu.palcu@....nxp.com>, David Gow <davidgow@...gle.com>,  Shawn Guo
 <shawnguo@...nel.org>, David Airlie <airlied@...il.com>, Pengutronix Kernel
 Team <kernel@...gutronix.de>, linux-arm-kernel@...ts.infradead.org, Lucas
 Stach <l.stach@...gutronix.de>
Subject: Re: [PATCH v2 1/2] platform_device: add devres function region-reqs

Yo!

On Mon, 2024-01-08 at 12:46 +0100, Uwe Kleine-König wrote:
> On Mon, Jan 08, 2024 at 10:45:31AM +0100, Philipp Stanner wrote:
> > On Mon, 2024-01-08 at 10:37 +0100, Uwe Kleine-König wrote:
> > > Other than that I indifferent if this is a good idea. There are
> > > so many
> > > helpers around these functions ...
> > 
> > Around which, the devres functions in general? There are, but
> > that's
> > kind of the point, unless we'd want everyone to call into the
> > lowest
> > level region-request functions with their own devres callbacks.
> > 
> > In any case: What would your suggestion be, should parties who
> > can't
> > (without restructuring very large parts of their code) ioremap()
> > and
> > request() simultaneously just not use devres? See my patch #2
> > Or is there another way to reach that goal that I'm not aware of?
> 
> This wasn't a constructive feedback unfortunately and more a feeling
> than a measurable criticism. To actually improve the state, maybe
> first
> check what helpers are actually there, how they are used and if they
> are
> suitable to what they are used for.
> 
> Having many helpers is a hint that the usage is complicated. Is that
> because the situation is complicated, or is this just a big pile of
> inconsistency that can be simplified and consolidated?

I thought about that and tend to believe that you are right in this
case. The reason being that there'd be very few callers to such a
wrapper.
We have the functions for doing pure requests and pure ioremaps, so
that should be sufficient.

I think we can do sth like this in the rare cases where someone needs
to request without (immediately) mapping:

struct dcss_dev *dcss_dev_create(struct device *dev, bool hdmi_output)
{
	struct platform_device *pdev = to_platform_device(dev);
	int ret;
	struct resource *res;
	struct dcss_dev *dcss;
	const struct dcss_type_data *devtype;
	resource_size_t res_len;

	devtype = of_device_get_match_data(dev);
	if (!devtype) {
		dev_err(dev, "no device match found\n");
		return ERR_PTR(-ENODEV);
	}

	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
	if (!res) {
		dev_err(dev, "cannot get memory resource\n");
		return ERR_PTR(-EINVAL);
	}

	res_len = res->end - res->start;
	if (!devm_request_mem_region(pdev->dev, res->start, res_len, "dcss")) {
		dev_err(dev, "cannot request memory region\n");
		return ERR_PTR(-EBUSY);
	}

And then do the associated devm_ioremap()s where they're needed.

So I'd 'close' this patch series and handle it entirely through my dcss
patch-series.

Thx for the feedback

P.

> 
> Also I think there are helpers that take a resource type parameter
> (as
> your function) and others hard code it in the function name. Maybe
> unifying that would be nice, too.
> 
> Best regards
> Uwe
>