[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <53869ef3-0825-55eb-fb89-f5e732b7e7ac@denx.de>
Date: Fri, 27 Mar 2020 04:23:58 +0100
From: Marek Vasut <marex@...x.de>
To: David Miller <davem@...emloft.net>
Cc: netdev@...r.kernel.org, lukas@...ner.de, ynezz@...e.cz,
yuehaibing@...wei.com
Subject: Re: [PATCH] net: ks8851-ml: Fix IO operations, again
On 3/27/20 4:21 AM, David Miller wrote:
> From: Marek Vasut <marex@...x.de>
> Date: Wed, 25 Mar 2020 15:25:47 +0100
>
>> This patch reverts 58292104832f ("net: ks8851-ml: Fix 16-bit IO operation")
>> and edacb098ea9c ("net: ks8851-ml: Fix 16-bit data access"), because it
>> turns out these were only necessary due to buggy hardware. This patch adds
>> a check for such a buggy hardware to prevent any such mistakes again.
>>
>> While working further on the KS8851 driver, it came to light that the
>> KS8851-16MLL is capable of switching bus endianness by a hardware strap,
>> EESK pin. If this strap is incorrect, the IO accesses require such endian
>> swapping as is being reverted by this patch. Such swapping also impacts
>> the performance significantly.
>>
>> Hence, in addition to removing it, detect that the hardware is broken,
>> report to user, and fail to bind with such hardware.
>>
>> Fixes: 58292104832f ("net: ks8851-ml: Fix 16-bit IO operation")
>> Fixes: edacb098ea9c ("net: ks8851-ml: Fix 16-bit data access")
>> Signed-off-by: Marek Vasut <marex@...x.de>
>
> Applied, thanks.
Sorry for this mess.
Powered by blists - more mailing lists