linux-kernel - Re: [PATCH] [0/12] SYSFS: Sysfs attribute improvements

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: <20100105145502.GG11732@kroah.com>
Date:	Tue, 5 Jan 2010 06:55:02 -0800
From:	Greg KH <greg@...ah.com>
To:	Andi Kleen <andi@...stfloor.org>
Cc:	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] [0/12] SYSFS: Sysfs attribute improvements

On Tue, Jan 05, 2010 at 12:47:57PM +0100, Andi Kleen wrote:
> 
> sysdev_class attributes and class attributes do not pass
> their own attribute to the low level store/show functions. This means
> that itiss not possible to use shared functions for common data,
> but each data item had to have its own function. Typically this
> lead to ugly and big custom macro cascades in drivers, generating
> the store/show functions for each attribute. This leads to 
> large code and is ugly.
> 
> When the attributes are passed to the callbacks -- like they are for plain
> attributes and sysdev attributes -- it's possible to "derive"
> own attribute objects that carry some data, use that data and then call into 
> shared store/show code.
> 
> This patchkit fixes up these callback interfaces to allow to do this. This
> involves two tree sweeps for sysdev_class_attributes and class_attributes,
> changing them all to accept the new attribute argument.
> 
> I tried to convert all users, but was only able to compile test on x86.
> 
> All users were converted in a single patch to avoid unbisectable
> sections. I didn't put all the individual driver maintainers in cc,
> they probably wouldn't need to care.
> 
> Also I added some utility functions to easier deal with arrays
> of attributes. A lot of users define multiple attributes and had
> to open code array handling for this. Add new sysfs interfaces
> to handle the arrays directly. sysdev classes now can just
> directly carry their own attributes, similar to normal classes.
> 
> Then the CPU and the NODE driver in drivers/base serve as a poster
> child for the new facilities and shrink their code dramatically
> by using the new facilities.
> 
> In addition I added a new string attribute type derived 
> from class attributes. This allows various drivers who use
> class attributes to export static version strings to just
> use a single macro, no custom callbacks.
> 
> I didn't convert all drivers over to the new facilities, this
> can be done gradually.

I like these changes, thanks for doing the work.  I do have some
comments on some of the patches, they will follow.

thanks,

greg k-h
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/