linux-kernel - Re: drivers/net/wireless/rt2x00/: struct data

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: <20071023194344.GX30533@stusta.de>
Date:	Tue, 23 Oct 2007 21:43:44 +0200
From:	Adrian Bunk <bunk@...nel.org>
To:	Ivo van Doorn <ivdoorn@...il.com>
Cc:	"John W. Linville" <linville@...driver.com>,
	linux-wireless@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: drivers/net/wireless/rt2x00/: struct data_desc strangeness

On Sun, Oct 14, 2007 at 09:06:20PM +0200, Ivo van Doorn wrote:
> On Sunday 14 October 2007, Adrian Bunk wrote:
> > drivers/net/wireless/rt2x00/rt2x00ring.h contains the following:
> > 
> > <--  snip  -->
> > 
> > ...
> > /*
> >  * data_desc
> >  * Each data entry also contains a descriptor which is used by the
> >  * device to determine what should be done with the packet and
> >  * what the current status is.
> >  * This structure is greatly simplified, but the descriptors
> >  * are basically a list of little endian 32 bit values.
> >  * Make the array by default 1 word big, this will allow us
> >  * to use sizeof() correctly.
> >  */
> > struct data_desc {
> >         __le32 word[1];
> > };
> > ...
> > /*
> >  * TX/RX Descriptor access functions.
> >  */
> > static inline void rt2x00_desc_read(struct data_desc *desc,
> >                                     const u8 word, u32 *value)
> > {
> >         *value = le32_to_cpu(desc->word[word]);
> > }
> > 
> > static inline void rt2x00_desc_write(struct data_desc *desc,
> >                                      const u8 word, const u32 value)
> > {
> >         desc->word[word] = cpu_to_le32(value);
> > }
> > ...
> > 
> > <--  snip  -->
> > 
> > I haven't checked whether it might work in all cases, but passing 
> > non-zero values as second parameter to rt2x00_desc_{read,write}()
> > (as is done in many cases) is even in the best case bad coding style.
> 
> Access to the array is correct, even with non-zero values the code that is
> reading/writing to the array knows the exact size of the descriptor. Within
> rt2x00 are however 5 drivers who have different descriptor sizes. That means
> I can't create a structure which has the correct array length.
> 
> The structure itself is just a simple map over preallocated memory
> (skb->data in case of USB or dma in case of PCI).
> 
> So possible solutions would be:
>  - remove struct data_desc and make it a void* or __le32*
>    This is at the cost of code readibility since the above pointers
>    have less meaning then a pointer to a structure which can be nicely
>    documented.
>...

The worst is a wrong meaning.
__le32 word[1] is an array with _one_ element.

And an __le32* can be used as an array.

> Ivo

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

-
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/