Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Vineet Gupta <vineet.gupta1@synopsys.com>
To:     Vitor Soares <vitor.soares@synopsys.com>,
        Alexey Brodkin <alexey.brodkin@synopsys.com>,
        Joao Pinto <joao.pinto@synopsys.com>,
        Jose Abreu <jose.abreu@synopsys.com>
CC:     arcml <linux-snps-arc@lists.infradead.org>,
        lkml <linux-kernel@vger.kernel.org>,
        Arnd Bergmann <arnd@arndb.de>
Subject: Re: Misaligned Access
Thread-Topic: Misaligned Access
Thread-Index: AQHUa8BEl692JJiolkiUiQERHFjjDw==
Date:   Wed, 21 Nov 2018 19:41:56 +0000
Message-ID: <C2D7FE5348E1B147BCA15975FBA23075014641580A@US01WEMBX2.internal.synopsys.com>
References: <13D59CF9CEBAF94592A12E8AE55501350113CD10@DE02WEMBXB.internal.synopsys.com>
 <C2D7FE5348E1B147BCA15975FBA23075012B0B4CD5@US01WEMBX2.internal.synopsys.com>
 <9ffd1270-ad92-88b7-eb53-85c65fe971dd@synopsys.com>
 <f8c469e7-9e1d-18d6-4e56-ba4338462048@synopsys.com>
 <ebe7f165-2901-2885-0583-c7be53329372@synopsys.com>
 <4881796E12491D4BB15146FE0209CE6468198AC6@DE02WEMBXB.internal.synopsys.com>
 <C2D7FE5348E1B147BCA15975FBA230750146415496@US01WEMBX2.internal.synopsys.com>
 <661a4323-5768-4977-5434-ab644c5418aa@synopsys.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

+CC lkml, Arnd : subject matter expert=0A=
=0A=
On 11/21/18 10:06 AM, Vitor Soares wrote:=0A=
> I use the follow function to get data from a RX Fifo.=0A=
>=0A=
>=0A=
> static void dw_i3c_master_read_rx_fifo(struct dw_i3c_master *master,=0A=
>                         u8 *bytes, int nbytes)=0A=
> {=0A=
>      readsl(master->regs + RX_TX_DATA_PORT, bytes, nbytes / 4);=0A=
=0A=
So the semantics are reading the same fifo register N times, to get the N w=
ords,=0A=
hence read*s*l is appropriate. That however expects the buffer to be 4 byte=
s=0A=
aligned, hence your issue. You can't possibly use the reads*b* as we want t=
he=0A=
=0A=
The obvious but crude hack is to use a temp array for readsl and then copy =
over=0A=
using memcpy, but I'm sure there are better ways, @Arnd ? To summarize is i=
ssue is=0A=
a driver triggering unaligned access due to the misinteraction of API (driv=
er get=0A=
an unaligned u8 *) which goes against expectations of io accessor readl  (n=
eeded=0A=
since the register contents are 4 bytes)=0A=
=0A=
=0A=
>      if (nbytes & 3) {=0A=
>          u32 tmp;=0A=
>=0A=
>          readsl(master->regs + RX_TX_DATA_PORT, &tmp, 1);=0A=
>          memcpy(bytes + (nbytes & ~3), &tmp, nbytes & 3);=0A=
>      }=0A=
> }=0A=
>=0A=
>=0A=
> and the pointer u8 *bytes is what is unaligned and breaks when inside of =
=0A=
> realdsl() it does:=0A=
>=0A=
>      *buf++ =3D x;=0A=
>=0A=
>=0A=
> Note that the u8 *bytes pointer is the __u8 *buf of the i2c_msg struct.=
=0A=
=0A=