linux-kernel - Re: [PATCH v4] irqchip/ls-extirq: fix invalid wait context by avoiding to use regmap

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]

Date:   Thu, 28 Jul 2022 16:19:50 +0000
From:   Vladimir Oltean <vladimir.oltean@....com>
To:     Sean Anderson <sean.anderson@...o.com>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Lee Jones <lee.jones@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Marc Zyngier <maz@...nel.org>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        Arnd Bergmann <arnd@...db.de>,
        "Z.Q. Hou" <zhiqiang.hou@....com>, Biwen Li <biwen.li@....com>
Subject: Re: [PATCH v4] irqchip/ls-extirq: fix invalid wait context by
 avoiding to use regmap

On Thu, Jul 28, 2022 at 11:43:40AM -0400, Sean Anderson wrote:
> > All of this should be rewritten to use of_device_is_big_endian(), and
> > reduce the whole thing to two cases (I don't think native endian makes
> > much sense anyway). I also wonder what the result is if none of these
> > properties is present...
> 
> I think regmap_get_val_endian would be better here.

It needs a struct device.

> >> +
> >> +	priv->regmap = regmap_init_mmio(NULL, base, &extirq_regmap_config);
> 
> It could also be done automatically if we pass the syscon dev instead of
> NULL. The only downside is that some regmap error messages will use the
> syscon device

How do you get the struct device of the syscon?

> > Finally, what is the actual benefit of using a regmap here? It seems
> > like a very roundabout way of performing a RMW on a register whilst
> > holding a lock... Passing NULL for a device to regmap_init_mmio() also
> > seems to be an extremely rare idiom (only 5 cases in the tree), and
> > this doesn't seem completely right to me.
> 
> The benefit is that you don't have to write (yet another) set of
> endian-converting read/write functions. The above (non-NULL) usage of
> regmap_init would also address your criticism here.

I don't have a particular attraction towards using regmap for a single
register either, to be honest.