linux-kernel - Re: device struct bloat

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: <20071103231403.GB6161@kroah.com>
Date:	Sat, 3 Nov 2007 16:14:03 -0700
From:	Greg KH <greg@...ah.com>
To:	Stephen Hemminger <shemminger@...ux-foundation.org>
Cc:	linux-kernel@...r.kernel.org
Subject: Re: device struct bloat

On Sat, Nov 03, 2007 at 12:48:23PM -0700, Stephen Hemminger wrote:
> The sizeof(struct device) is way too big, especially in the network device case.
> We want to support 1000's of device's and the change from class_device to
> net_device has caused needless bloat.
> 
> sizeof(struct device) = 272
> sizeof(struct class_device) = 92
>   * not the class_id in class_device could also be removed or changed to
>      a ptr, since it is redundant for net_devices.

I agree that struct device is bigger than perhaps it should be (Kay is
working on getting rid of the bus_id field and we both just trimmed down
the base kobject by about 20 bytes) but is this really a problem that is
noticable by anyone?

I'm all for saving memory, but 1000's of struct devices is not anything
that the kernel should even notice.  s390 machines create tens of
thousands of these all the time, and they are severly memory limited,
with no apparent problem.

And I'm guessing that embedded systems would not be the ones that would
be creating 1000's of network devices, right?  Are these virtual devices
or backed by real, physical devices?

If it is an issue, we can start to work on slimming the structure down.
At first glance, I'm sure we can save memory by just rearanging the
fields to get rid of some structure padding that I'm sure is there.

After that, I'm sure we can push a lot of other fields out into a
separate structure to handle if the device is "virtual" or not, which
would let us drop a bunch of the dma and other resource-type things.

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/