linux-kernel - RE: [PATCH v6 1/2] spi: Add spi driver for Sunplus SP7021

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: <638055e6c4dd4f66bab98c12bd5bd74a@sphcmbx02.sunplus.com.tw>
Date:   Mon, 24 Jan 2022 03:27:45 +0000
From:   Lh Kuo 郭力豪 <lh.Kuo@...plus.com>
To:     Andy Shevchenko <andy.shevchenko@...il.com>
CC:     Li-hao Kuo <lhjeff911@...il.com>,
        Philipp Zabel <p.zabel@...gutronix.de>,
        Mark Brown <broonie@...nel.org>,
        Rob Herring <robh+dt@...nel.org>,
        linux-spi <linux-spi@...r.kernel.org>,
        devicetree <devicetree@...r.kernel.org>,
        "Linux Kernel Mailing List" <linux-kernel@...r.kernel.org>,
        Wells Lu 呂芳騰 <wells.lu@...plus.com>
Subject: RE: [PATCH v6 1/2] spi: Add spi driver for Sunplus SP7021

> > > > > > +       if (xfer->tx_buf)
> > > > > > +               dma_unmap_single(dev, xfer->tx_dma, xfer->len, DMA_TO_DEVICE);
> > > > > > +       if (xfer->rx_buf)
> > > > > > +               dma_unmap_single(dev, xfer->rx_dma, xfer->len,
> > > > > > + DMA_FROM_DEVICE);
> > > > >
> > > > > Why can't you use SPI core DMA mapping code?
> > > >
> > > > I didn't find the SPI core DMA mapping code for single maping.
> > > > The method currently used is the general DMA single-map code usage method.
> > >
> > > Why do you need single page mapping?
> > > What's wrong with SG mapping that SPI core provides?
> >
> > SP7021 SPI slave dma only supports single dma in one trigger.
> > It is not suitable for using SG mapping.
> > If the length of the transfer is larger than the length of the
> > SG-mapping, Slave-mode will get error in the transfer.
> 
> Same story for SPI DesignWare on Intel Medfield, where no SG transfers are supported by hardware.
> Nevertheless, the DMA driver takes care of this and on each interrupt recharges a channel to continue.
> Why can't the same be implemented here?
> 
> 
I think it should work in master. spi master must actively send clk and date to slave device.
And yes, in the "master" mode it can handle SG-DMA on each interrupt.
But if working in "slave" mode, the master will not know the state of the slave. Slaves work on interrupt and recharge channels.
When master send clk and date in the same time. It may lose data and errors occur