linux-kernel - Re: [PATCH v4 10/15] dmaengine: ti: New driver for K3 UDMA - split#2: probe/remove, xlate and filter

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:   Tue, 12 Nov 2019 11:04:40 +0530
From:   Vinod Koul <vkoul@...nel.org>
To:     Peter Ujfalusi <peter.ujfalusi@...com>
Cc:     robh+dt@...nel.org, nm@...com, ssantosh@...nel.org,
        dan.j.williams@...el.com, dmaengine@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, grygorii.strashko@...com,
        lokeshvutla@...com, t-kristo@...com, tony@...mide.com,
        j-keerthy@...com
Subject: Re: [PATCH v4 10/15] dmaengine: ti: New driver for K3 UDMA -
 split#2: probe/remove, xlate and filter_fn

On 11-11-19, 11:16, Peter Ujfalusi wrote:
> 
> 
> On 11/11/2019 7.33, Vinod Koul wrote:
> > On 01-11-19, 10:41, Peter Ujfalusi wrote:
> > 
> >> +static bool udma_dma_filter_fn(struct dma_chan *chan, void *param)
> >> +{
> >> +	struct psil_endpoint_config *ep_config;
> >> +	struct udma_chan *uc;
> >> +	struct udma_dev *ud;
> >> +	u32 *args;
> >> +
> >> +	if (chan->device->dev->driver != &udma_driver.driver)
> >> +		return false;
> >> +
> >> +	uc = to_udma_chan(chan);
> >> +	ud = uc->ud;
> >> +	args = param;
> >> +	uc->remote_thread_id = args[0];
> >> +
> >> +	if (uc->remote_thread_id & K3_PSIL_DST_THREAD_ID_OFFSET)
> >> +		uc->dir = DMA_MEM_TO_DEV;
> >> +	else
> >> +		uc->dir = DMA_DEV_TO_MEM;
> > 
> > Can you explain this a bit?
> 
> The UDMAP in K3 works between two PSI-L endpoint. The source and
> destination needs to be paired to allow data flow.
> Source thread IDs are in range of 0x0000 - 0x7fff, while destination
> thread IDs are 0x8000 - 0xffff.
> 
> If the remote thread ID have the bit 31 set (0x8000) then the transfer
> is MEM_TO_DEV and I need to pick one unused tchan for it. If the remote
> is the source then it can be handled by rchan.
> 
> dmas = <&main_udmap 0xc400>, <&main_udmap 0x4400>;
> dma-names = "tx", "rx";
> 
> 0xc400 is a destination thread ID, so it is MEM_TO_DEV
> 0x4400 is a source thread ID, so it is DEV_TO_MEM
> 
> Even in MEM_TO_MEM case I need to pair two UDMAP channels:
> UDMAP source threads are starting at offset 0x1000, UDMAP destination
> threads are 0x9000+

Okay so a channel is set for a direction until teardown. Also this and
other patch comments are quite useful, can we add them here?

> Changing direction runtime is hardly possible as it would involve
> tearing down the channel, removing interrupts, destroying rings,
> removing the PSI-L pairing and redoing everything.

okay I would expect the prep_ to check for direction and reject the call
if direction is different.

-- 
~Vinod