linux-kernel - Re: [PATCH] dmaengine: tegra210-adma: fix transfer failure

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:   Fri, 21 Jun 2019 08:20:51 +0530
From:   Sameer Pujar <spujar@...dia.com>
To:     Jon Hunter <jonathanh@...dia.com>, <vkoul@...nel.org>,
        <dan.j.williams@...el.com>
CC:     <thierry.reding@...il.com>, <ldewangan@...dia.com>,
        <dmaengine@...r.kernel.org>, <linux-tegra@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] dmaengine: tegra210-adma: fix transfer failure


On 6/20/2019 10:07 PM, Jon Hunter wrote:
> On 20/06/2019 17:15, Sameer Pujar wrote:
>>  From Tegra186 onwards OUTSTANDING_REQUESTS field is added in channel
>> configuration register (bits 7:4). ADMA allows a maximum of 8 reads
>> to source and that many writes to target memory be outstanding at any
>> given point of time. If this field is not programmed, DMA transfers
>> fail to happen.
>>
>> Thus added 'ch_pending_req' member in chip data structure and the
>> same is populated with maximum allowed pending requests. Since the
>> field is not applicable to Tegra210, mentioned bit fields are unused
>> and hence the member is initialized with 0.
>>
>> Fixes: 433de642a76c ("dmaengine: tegra210-adma: add support for Tegra186/Tegra194")
>>
>> Signed-off-by: Sameer Pujar <spujar@...dia.com>
>> ---
>>   drivers/dma/tegra210-adma.c | 5 +++++
>>   1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/dma/tegra210-adma.c b/drivers/dma/tegra210-adma.c
>> index 17ea4dd99..8d291cf 100644
>> --- a/drivers/dma/tegra210-adma.c
>> +++ b/drivers/dma/tegra210-adma.c
>> @@ -96,6 +96,7 @@ struct tegra_adma;
>>    * @ch_req_tx_shift: Register offset for AHUB transmit channel select.
>>    * @ch_req_rx_shift: Register offset for AHUB receive channel select.
>>    * @ch_base_offset: Register offset of DMA channel registers.
>> + * @ch_pending_req: Outstaning DMA requests for a channel.
>>    * @ch_fifo_ctrl: Default value for channel FIFO CTRL register.
>>    * @ch_req_mask: Mask for Tx or Rx channel select.
>>    * @ch_req_max: Maximum number of Tx or Rx channels available.
>> @@ -109,6 +110,7 @@ struct tegra_adma_chip_data {
>>   	unsigned int ch_req_tx_shift;
>>   	unsigned int ch_req_rx_shift;
>>   	unsigned int ch_base_offset;
>> +	unsigned int ch_pending_req;
>>   	unsigned int ch_fifo_ctrl;
>>   	unsigned int ch_req_mask;
>>   	unsigned int ch_req_max;
>> @@ -613,6 +615,7 @@ static int tegra_adma_set_xfer_params(struct tegra_adma_chan *tdc,
>>   			 ADMA_CH_CTRL_FLOWCTRL_EN;
>>   	ch_regs->config |= cdata->adma_get_burst_config(burst_size);
>>   	ch_regs->config |= ADMA_CH_CONFIG_WEIGHT_FOR_WRR(1);
>> +	ch_regs->config |= cdata->ch_pending_req;
>>   	ch_regs->fifo_ctrl = cdata->ch_fifo_ctrl;
>>   	ch_regs->tc = desc->period_len & ADMA_CH_TC_COUNT_MASK;
>>   
>> @@ -797,6 +800,7 @@ static const struct tegra_adma_chip_data tegra210_chip_data = {
>>   	.ch_req_tx_shift	= 28,
>>   	.ch_req_rx_shift	= 24,
>>   	.ch_base_offset		= 0,
>> +	.ch_pending_req		= 0,
>>   	.ch_fifo_ctrl		= TEGRA210_FIFO_CTRL_DEFAULT,
>>   	.ch_req_mask		= 0xf,
>>   	.ch_req_max		= 10,
>> @@ -811,6 +815,7 @@ static const struct tegra_adma_chip_data tegra186_chip_data = {
>>   	.ch_req_tx_shift	= 27,
>>   	.ch_req_rx_shift	= 22,
>>   	.ch_base_offset		= 0x10000,
>> +	.ch_pending_req		= (8 << 4),
> So given that this is a value and a shift, I think that we should add a
> proper definition like we have for TEGRA186_FIFO_CTRL_DEFAULT below.
I can add.
>
>>   	.ch_fifo_ctrl		= TEGRA186_FIFO_CTRL_DEFAULT,
>>   	.ch_req_mask		= 0x1f,
>>   	.ch_req_max		= 20,
>>
> Please group your patches into a series if you have more than one for a
> given driver.
The other ADMA related change is in Kconfig and a simple change. Current 
patch can go through some review
cycles. Hence didn't want to delay the other patch.
> Cheers
> Jon
>