linux-kernel - Re: [PATCH 0/6] NXP DSPI bugfixes and support for LS1028A

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: <CA+h21hopP2XTx55iu_pG=xBx-TSPRBbdmoU7T2F0Gc9Qt=CsSQ@mail.gmail.com>
Date:   Mon, 9 Mar 2020 20:48:34 +0200
From:   Vladimir Oltean <olteanv@...il.com>
To:     Michael Walle <michael@...le.cc>
Cc:     Mark Brown <broonie@...nel.org>, linux-spi@...r.kernel.org,
        lkml <linux-kernel@...r.kernel.org>,
        Shawn Guo <shawnguo@...nel.org>,
        Rob Herring <robh+dt@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        devicetree@...r.kernel.org, Esben Haabendal <eha@...f.com>,
        angelo@...am.it, andrew.smirnov@...il.com,
        "Gustavo A. R. Silva" <gustavo@...eddedor.com>,
        Wei Chen <weic@...dia.com>, Mohamed Hosny <mhosny@...dia.com>,
        peng.ma@....com
Subject: Re: [PATCH 0/6] NXP DSPI bugfixes and support for LS1028A

On Mon, 9 Mar 2020 at 20:31, Michael Walle <michael@...le.cc> wrote:
>
> Am 2020-03-09 19:14, schrieb Vladimir Oltean:
> > On Mon, 9 Mar 2020 at 20:03, Michael Walle <michael@...le.cc> wrote:
> >> Am 2020-03-09 15:56, schrieb Vladimir Oltean:
> >> > From: Vladimir Oltean <vladimir.oltean@....com>
> >> >
> >> > This series addresses a few issues that were missed during the previous
> >> > series "[PATCH 00/12] TCFQ to XSPI migration for NXP DSPI driver", on
> >> > SoCs other than LS1021A and LS1043A. DMA mode has been completely
> >> > broken
> >> > by that series, and XSPI mode never worked on little-endian
> >> > controllers.
> >> >
> >> > Then it introduces support for the LS1028A chip, whose compatible has
> >> > recently been documented here:
> >> >
> >> > https://lore.kernel.org/linux-devicetree/20200218171418.18297-1-michael@walle.cc/
> >>
> >> If it is not compatible with the LS1021A the second compatible string
> >> should be removed. Depending on the other remark about the endianess,
> >> it might still be compatible, though.
> >>
> >
> > Please feel free to remove it. I wasn't actually planning to add it in
> > the first place, but now it that it's there it doesn't really bother
> > anybody either.
>
> But it won't work if the endianess depends on the compatible string ;)
>

True.

> >>
> >> > The device tree for the LS1028A SoC is extended with DMA channels
> >> > definition, such that even though the default operating mode is XSPI,
> >> > one can simply change DSPI_XSPI_MODE to DSPI_DMA_MODE in the
> >> > devtype_data structure of the driver and use that instead.
> >>
> >> wouldn't it make more sense, to use DMA is the dma node is present
> >> in the device tree? otherwise use XSPI mode? I don't think it is
> >> really handy to change the mode inside the driver.
> >>
> >
> > Let's keep it simple. The driver should configure the hardware in the
> > most efficient and least buggy mode available. Right now that is XSPI.
> > The hardware description (aka the device tree) is a separate topic. If
> > there ever arises any situation where there are corner cases with XSPI
> > mode, it's good to have a fallback in the form of DMA mode, and not
> > have to worry about yet another problem (which is that there are 2
> > sets of device tree blobs in deployment).
>
> Point taken. But this is not how other drivers behave, which uses the
> DMA if its given in the device node.
>

Also true.

> Btw. do other SoCs perform better with DMA?
>

Not that I know of.
My general rule of thumb for this controller is: if it supports XSPI
then use that, otherwise use DMA. Luckily there is just one controller
that supports none of those, and that would be Coldfire, which uses
the braindead EOQ mode. I don't have the hardware to do testing on
that, but in principle if I did, I would have converted that as well
to the more functional but less efficient TCFQ mode (now removed).

> -michael
>
> > TL;DR: These DMA channels don't really bother anybody but you never
> > know when they might come in handy.
> >
> >> -michael
> >>
> >> >
> >> > For testing, benchmarking and debugging, the mikroBUS connector on the
> >> > LS1028A-RDB is made available via spidev.
> >> >
> >> > Vladimir Oltean (6):
> >> >   spi: spi-fsl-dspi: Don't access reserved fields in SPI_MCR
> >> >   spi: spi-fsl-dspi: Fix little endian access to PUSHR CMD and TXDATA
> >> >   spi: spi-fsl-dspi: Fix oper_word_size of zero for DMA mode
> >> >   spi: spi-fsl-dspi: Add support for LS1028A
> >> >   arm64: dts: ls1028a: Specify the DMA channels for the DSPI
> >> > controllers
> >> >   arm64: dts: ls1028a-rdb: Add a spidev node for the mikroBUS
> >> >
> >> >  .../boot/dts/freescale/fsl-ls1028a-rdb.dts    | 14 +++++
> >> >  .../arm64/boot/dts/freescale/fsl-ls1028a.dtsi |  6 +++
> >> >  drivers/spi/spi-fsl-dspi.c                    | 54 +++++++++++++++----
> >> >  3 files changed, 64 insertions(+), 10 deletions(-)
> >
> > Thanks,
> > -Vladimir

-Vladimir