linux-kernel - Re: [PATCH] dma: ti: k3-udma: Workaround errata i2234

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] [day] [month] [year] [list]

Message-ID: <2ddcbea8-d3fa-ed73-ead1-834a8f304f88@ti.com>
Date:   Thu, 23 Mar 2023 17:25:34 +0530
From:   Jayesh Choudhary <j-choudhary@...com>
To:     Péter Ujfalusi <peter.ujfalusi@...il.com>,
        Vignesh Raghavendra <vigneshr@...com>,
        Vinod Koul <vkoul@...nel.org>
CC:     <dmaengine@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        Linux ARM Mailing List <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH] dma: ti: k3-udma: Workaround errata i2234

Hello Peter,

On 10/12/21 20:52, Péter Ujfalusi wrote:
> Hi Vignesh,
> 
> On 09/12/2021 20:07, Vignesh Raghavendra wrote:
>> Per [1], UDMA TR15 transactions may hang if ICNT0 is less than 64B
>> Work around is to set EOL flag is to 1 for ICNT0.
>>
>> Since, there is no performance penalty / side effects of setting EOL
>> flag event ICNTO > 64B, just set the flag for all UDMAP TR15
>> descriptors.
> 
> PDMAs and CSI does not send EOL? If you set it the EOL to one then when
> it arrives the remaining icnt0 is skipped...

I am planning to respin v2 for this.
Will fix this in v2.

> 
>>
>> [1] https://www.ti.com/lit/er/sprz455a/sprz455a.pdf
>> Errata doc for J721E DRA829/TDA4VM Processors Silicon Revision 1.1/1.0 (Rev. A)
>>
>> Signed-off-by: Vignesh Raghavendra <vigneshr@...com>
>> ---
>>   drivers/dma/ti/k3-udma.c     | 48 +++++++++++++++++++-----------------
>>   include/linux/dma/ti-cppi5.h |  1 +
>>   2 files changed, 27 insertions(+), 22 deletions(-)
>>

[...]

>> diff --git a/include/linux/dma/ti-cppi5.h b/include/linux/dma/ti-cppi5.h
>> index efa2f0309f00..c53c0f6e3b1a 100644
>> --- a/include/linux/dma/ti-cppi5.h
>> +++ b/include/linux/dma/ti-cppi5.h
>> @@ -616,6 +616,7 @@ static inline void *cppi5_hdesc_get_swdata(struct cppi5_host_desc_t *desc)
>>   #define   CPPI5_TR_CSF_SUPR_EVT			BIT(2)
>>   #define   CPPI5_TR_CSF_EOL_ADV_SHIFT		(4U)
>>   #define   CPPI5_TR_CSF_EOL_ADV_MASK		GENMASK(6, 4)
>> +#define   CPPI5_TR_CSF_EOL_ICNT0		BIT(4)
> 
> the correct expression is: (1 << CPPI5_TR_CSF_EOL_ADV_SHIFT)

Both these expressions expands to the same value
(CPPI5_TR_CSF_EOL_ADV_SHIFT = 4U)

And according to the linux checkpatch, the usage of BIT macro
is preferred so I will keep it the same.

Warm regards,
-Jayesh

> as EOL = 1 is what you want to set.
> EOL = 2 will clear icnt0 and 1 on EOL.
> 3 will do the same for icnt 0, 1 and 2
> 4 will skip the remainin tr.
> 
>>   #define   CPPI5_TR_CSF_EOP			BIT(7)
>>   
>>   /**
>>
>