From mboxrd@z Thu Jan  1 00:00:00 1970
From: Gabbasov, Andrew <Andrew_Gabbasov@mentor.com>
Date: Wed, 8 May 2013 17:38:39 +0000
Subject: [U-Boot] [PATCH] ARM: cfi_flash: Fix unaligned accesses to
 cfi_qry structure
In-Reply-To: <201305081926.01516.marex@denx.de>
References: <1367228070-6351-1-git-send-email-andrew_gabbasov@mentor.com>,
	<201305081926.01516.marex@denx.de>
Message-ID: <6C5EA58090A5ED459815C4D04C2B466FD932EEF6@EU-MBX-01.mgc.mentorg.com>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Hi Marek,

> From: Marek Vasut [marex at denx.de]
> Sent: Wednesday, May 08, 2013 21:26
> To: u-boot at lists.denx.de
> Cc: Gabbasov, Andrew; Tom Rini; Albert Aribaud
> Subject: Re: [U-Boot] [PATCH] ARM: cfi_flash: Fix unaligned accesses to cfi_qry structure
> 
> Dear Andrew Gabbasov,
> 
> > Packed structure cfi_qry contains unaligned 16- and 32-bits members,
> > accessing which causes problems when cfi_flash driver is compiled with
> > -munaligned-access option: flash initialization hangs, probably
> > due to data error.
> >
> > This fix converts 16- and 32-bit members to byte arrays and uses special
> > macros to access such fields. It removes possible unaligned accesses
> > in cfi_flash driver.
> >
> > Signed-off-by: Andrew Gabbasov <andrew_gabbasov@mentor.com>
> 
> It seems OK. I just wonder if the le16_to_cpu you removed can have no impact now
> on obscure-endian architectures, that's my only concern.
> 
> Best regards,
> Marek Vasut

Thank you for your feedback.

Using le16_to_cpu macros assumed that the data, read from flash, are always in little-endian
order, that is the LSB byte comes first in those 2-bytes unaligned fields. This is exactly how
the data is read by new macros. So, it looks like new macros should work on any host
endianness.

Thanks.

Best regards,
Andrew