linux-kernel - Re: [PATCH 1/2] staging: android: ashmem: fix ashmem pin/unpin interface

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: <20130204014114.GA14058@kroah.com>
Date:	Sun, 3 Feb 2013 17:41:14 -0800
From:	Greg KH <gregkh@...uxfoundation.org>
To:	Serban Constantinescu <Serban.Constantinescu@....com>
Cc:	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"kernel-team@...roid.com" <kernel-team@...roid.com>,
	"arve@...roid.com" <arve@...roid.com>,
	"john.stultz@...aro.org" <john.stultz@...aro.org>,
	Dave Butcher <Dave.Butcher@....com>
Subject: Re: [PATCH 1/2] staging: android: ashmem: fix ashmem pin/unpin
 interface

On Fri, Feb 01, 2013 at 04:55:01PM +0000, Serban Constantinescu wrote:
> Hi Greg,
> 
> On 01/02/13 16:18, Greg KH wrote:
> >On Fri, Feb 01, 2013 at 04:08:00PM +0000, Serban Constantinescu wrote:
> >>The values exchanged between kernel and userspace through struct
> >>ashmem_pin should be of type size_t. This change won't affect the
> >>existing interface but will stand as the basis of 64bit compat layer.
> >
> >How do you define size_t with a 64bit kernel and a 32bit userspace
> >properly?  Doesn't this change open up a bunch of problems?
> 
> The current ashmem pin/unpin kernel interface uses __u32 to specify
> the memory region and length in bytes. However these values  should
> be of type size_t so that they are able to represent the whole range
> of possible values when compiled for a 64bit platform.

Yes, the issue is, what size is size_t on the system if you have a 32bit
userspace and a 64bit kernel?  :)

That's why we have specific types for when we cross the user/kernel
boundry.  Why not use them instead here so that you know it will work
properly in the future?

> Android API uses ashmem driver through libcutils, from where I
> attach the following snippet:
> 
> <aosp>/system/core/libcutils/ashmem-dev.c
> 
>   75 int ashmem_pin_region(int fd, size_t offset, size_t len)
>   76 {
>   77         struct ashmem_pin pin = { offset, len };
>   78         return ioctl(fd, ASHMEM_PIN, &pin);
>   79 }

Again, the 32/64 bit issue is to blame.

> The kernel changes inline with the userspace usage and do not affect
> existing 32bit Android (we have exported the new kernel header,
> rebuilt and tested the interface with success).
> 
> However this change will affect any 64bit userspace using the
> current faulty interface, but there is none that we know of.

I'd like the Android developers to give some feedback on this, before
I'll do anything.  I still think you need to change this to use the
proper kernel types.

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/