linux-kernel - RE: [PATCH v17 2/4] dmaengine: tegra: Add tegra gpcdma driver

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]

Date:   Sun, 30 Jan 2022 16:43:46 +0000
From:   Akhil R <akhilrajeev@...dia.com>
To:     Dmitry Osipenko <digetx@...il.com>,
        "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
        "dmaengine@...r.kernel.org" <dmaengine@...r.kernel.org>,
        Jonathan Hunter <jonathanh@...dia.com>,
        Krishna Yarlagadda <kyarlagadda@...dia.com>,
        Laxman Dewangan <ldewangan@...dia.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-tegra@...r.kernel.org" <linux-tegra@...r.kernel.org>,
        "p.zabel@...gutronix.de" <p.zabel@...gutronix.de>,
        Rajesh Gumasta <rgumasta@...dia.com>,
        "robh+dt@...nel.org" <robh+dt@...nel.org>,
        "thierry.reding@...il.com" <thierry.reding@...il.com>,
        "vkoul@...nel.org" <vkoul@...nel.org>
CC:     Pavan Kunapuli <pkunapuli@...dia.com>
Subject: RE: [PATCH v17 2/4] dmaengine: tegra: Add tegra gpcdma driver

> 30.01.2022 13:26, Dmitry Osipenko пишет:
> > 30.01.2022 13:05, Dmitry Osipenko пишет:
> >> Still nothing prevents interrupt handler to fire during the pause.
> >>
> >> What you actually need to do is to disable/enable interrupt. This
> >> will prevent the interrupt racing and then pause/resume may look like this:
> >
> > Although, seems this won't work, unfortunately. I see now that
> > device_pause() doesn't have might_sleep().
> >
> 
> Ah, I see now that the pause/unpause is actually a separate control and doesn't
> conflict with "start next transfer".
> 
> So you just need to set/unset the pause under lock. And don't touch
> tdc->dma_desc. That's it.
> 
> static int tegra_dma_device_pause(struct dma_chan *dc) {
>         struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>         unsigned long flags;
>         u32 val;
> 
>         if (!tdc->tdma->chip_data->hw_support_pause)
>                 return -ENOSYS;
> 
>         spin_lock_irqsave(&tdc->vc.lock, flags);
> 
>         val = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE);
>         val |= TEGRA_GPCDMA_CHAN_CSRE_PAUSE;
>         tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSRE, val);
> 
>         spin_unlock_irqrestore(&tdc->vc.lock, flags);
> 
>         return 0;
> }
> 
> static int tegra_dma_device_resume(struct dma_chan *dc) {
>         struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>         unsigned long flags;
>         u32 val;
> 
>         if (!tdc->tdma->chip_data->hw_support_pause)
>                 return -ENOSYS;
> 
>         spin_lock_irqsave(&tdc->vc.lock, flags);
> 
>         val = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE);
>         val &= ~TEGRA_GPCDMA_CHAN_CSRE_PAUSE;
>         tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSRE, val);
> 
>         spin_unlock_irqrestore(&tdc->vc.lock, flags);
> 
>         return 0;
> }

The reason I separated out register writes was to conveniently call those
in dma_start() and terminate_all(). Do you see any issue there?
The recommended way of terminating a transfer in between is to pause
it before disabling the channel.

dma_desc could be NULL while these functions are called. pause() or
resume() is unneeded if there isn't any transfer going on. Moreover,
if we are to calculate the xfer_size, the check would be mandatory.

Agree with the other comments.

Thanks,
Akhil