From mboxrd@z Thu Jan 1 00:00:00 1970 From: Lothar =?UTF-8?B?V2HDn21hbm4=?= Subject: Re: [PATCH] net: fec: fix regression on i.MX28 introduced by rx_copybreak support Date: Wed, 8 Oct 2014 10:55:54 +0200 Message-ID: <20141008105554.2af97be9@ipc1.ka-ro> References: <1412687977-11742-1-git-send-email-LW@KARO-electronics.de> <1412692034.11091.122.camel@edumazet-glaptop2.roam.corp.google.com> <063D6719AE5E284EB5DD2968C1650D6D174C6116@AcuExch.aculab.com> <20141008070147.2f3b5319@ipc1.ka-ro> <063D6719AE5E284EB5DD2968C1650D6D174C6BA0@AcuExch.aculab.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: 'Eric Dumazet' , "netdev@vger.kernel.org" , "David S. Miller" , "Russell King" , Frank Li , "Fabio Estevam" , "linux-kernel@vger.kernel.org" To: David Laight Return-path: In-Reply-To: <063D6719AE5E284EB5DD2968C1650D6D174C6BA0@AcuExch.aculab.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Hi, David Laight wrote: > From: Lothar Wa=C3=9Fmann > > David Laight wrote: > > > From: Eric Dumazet > > > > On Tue, 2014-10-07 at 15:19 +0200, Lothar Wamann wrote: > > > > > commit 1b7bde6d659d ("net: fec: implement rx_copybreak to imp= rove rx performance") > > > > > introduced a regression for i.MX28. The swap_buffer() functio= n doing > > > > > the endian conversion of the received data on i.MX28 may acce= ss memory > > > > > beyond the actual packet size in the DMA buffer. fec_enet_cop= ybreak() > > > > > does not copy those bytes, so that the last bytes of a packet= may be > > > > > filled with invalid data after swapping. > > > > > This will likely lead to checksum errors on received packets. > > > > > E.g. when trying to mount an NFS rootfs: > > > > > UDP: bad checksum. From 192.168.1.225:111 to 192.168.100.73:4= 4662 ulen 36 > > > > > > > > > > Do the byte swapping and copying to the new skb in one go if > > > > > necessary. > > > > > > > > > > Signed-off-by: Lothar Wamann > > > > > --- > > > > > drivers/net/ethernet/freescale/fec_main.c | 25 +++++++++++= ++++++++++---- > > > > > 1 file changed, 21 insertions(+), 4 deletions(-) > > > > > > > > > > diff --git a/drivers/net/ethernet/freescale/fec_main.c b/driv= ers/net/ethernet/freescale/fec_main.c > > > > > index 87975b5..eaaebad 100644 > > > > > --- a/drivers/net/ethernet/freescale/fec_main.c > > > > > +++ b/drivers/net/ethernet/freescale/fec_main.c > > > > > @@ -339,6 +339,18 @@ static void *swap_buffer(void *bufaddr, = int len) > > > > > return bufaddr; > > > > > } > > > > > > > > > > +static void *swap_buffer2(void *dst_buf, void *src_buf, int = len) > > > > > +{ > > > > > + int i; > > > > > + unsigned int *src =3D src_buf; > > > > > + unsigned int *dst =3D dst_buf; > > > > > + > > > > > + for (i =3D 0; i < DIV_ROUND_UP(len, 4); i++, src++, dst++) > > > > > + *dst =3D cpu_to_be32(*src); > > > > > > > > No need for the DIV : > > > > > > > > for (i =3D 0; i < len; i +=3D sizeof(*dst), src++, dst++) > > > > *dst =3D cpu_to_be32(*src); > > > > > > > > Also are you sure both src/dst are aligned to word boundaries, = or is > > > > this architecture OK with possible misalignment ? > > > > > > I wondered about that as well. > > > I wouldn't have expected ppc to support misaligned transfers, and= you'd also > > > want to make sure that cpu_to_be(*src) was using a byte-swapping = instruction. > > > Hmmm... cpu_to_be() doesn't sound like the right 'swap' macro nam= e. > > > > > ??? So what is cpu_to_be32() then? > > The new swap function is an exact copy of the original one already = in > > use except for the fact that it uses distinct source and destinatio= n > > buffers. >=20 > cpu_to_be32() is for converting a 'cpu' endianness value to 'big-endi= an'. > Here you are processing receive data - so you probably want be32_to_c= pu(). > (Yes, I know they are functionally identical....) >=20 > Alternatively, since these aren't actually 32bit numbers and you know > whether you want to swap, something like the __swab32p() from swab.h > - but I'm not entirely sure that one is expected to be used. >=20 OK, but that should probably be a separate patch to fix the original swap_buffer() function as well. > Clearly you are well inside the ppc 'endianness' hell. > No. i.MX28 is not PPC. It only has a "feature" in the ethernet controller that stores all data in big endian format. Lothar Wa=C3=9Fmann --=20 ___________________________________________________________ Ka-Ro electronics GmbH | Pascalstra=C3=9Fe 22 | D - 52076 Aachen Phone: +49 2408 1402-0 | Fax: +49 2408 1402-10 Gesch=C3=A4ftsf=C3=BChrer: Matthias Kaussen Handelsregistereintrag: Amtsgericht Aachen, HRB 4996 www.karo-electronics.de | info@karo-electronics.de ___________________________________________________________