[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <433418e889347784bc74f3c22c23e644@walle.cc>
Date: Sat, 07 Mar 2020 14:10:19 +0100
From: Michael Walle <michael@...le.cc>
To: Peng Ma <peng.ma@....com>
Cc: dmaengine@...r.kernel.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
Vinod Koul <vkoul@...nel.org>,
Rob Herring <robh+dt@...nel.org>,
Mark Rutland <mark.rutland@....com>,
Shawn Guo <shawnguo@...nel.org>, Leo Li <leoyang.li@....com>,
Vladimir Oltean <olteanv@...il.com>
Subject: Re: [EXT] [PATCH 2/2] arm64: dts: ls1028a: add "fsl,vf610-edma"
compatible
Hi Peng,
Am 2020-03-07 11:32, schrieb Peng Ma:
>> -----Original Message-----
>> From: Michael Walle <michael@...le.cc>
>> Sent: 2020年3月7日 17:26
>> To: Peng Ma <peng.ma@....com>
>> Cc: dmaengine@...r.kernel.org; devicetree@...r.kernel.org;
>> linux-kernel@...r.kernel.org; linux-arm-kernel@...ts.infradead.org;
>> Vinod Koul
>> <vkoul@...nel.org>; Rob Herring <robh+dt@...nel.org>; Mark Rutland
>> <mark.rutland@....com>; Shawn Guo <shawnguo@...nel.org>; Leo Li
>> <leoyang.li@....com>
>> Subject: Re: [EXT] [PATCH 2/2] arm64: dts: ls1028a: add
>> "fsl,vf610-edma"
>> compatible
>>
>> Caution: EXT Email
>>
>> Hi Peng,
>>
>> Am 2020-03-07 03:09, schrieb Peng Ma:
>>>> -----Original Message-----
>>>> From: Michael Walle <michael@...le.cc>
>>>> Sent: 2020年3月7日 4:54
>>>> To: dmaengine@...r.kernel.org; devicetree@...r.kernel.org;
>>>> linux-kernel@...r.kernel.org; linux-arm-kernel@...ts.infradead.org
>>>> Cc: Vinod Koul <vkoul@...nel.org>; Rob Herring <robh+dt@...nel.org>;
>>>> Mark Rutland <mark.rutland@....com>; Shawn Guo
>> <shawnguo@...nel.org>;
>>>> Leo Li <leoyang.li@....com>; Peng Ma <peng.ma@....com>; Michael
>>>> Walle
>>>> <michael@...le.cc>
>>>> Subject: [EXT] [PATCH 2/2] arm64: dts: ls1028a: add "fsl,vf610-edma"
>>>> compatible
>>>>
>>>> Caution: EXT Email
>>>>
>>>> The bootloader does the IOMMU fixup and dynamically adds the
>>>> "iommus"
>>>> property to devices according to its compatible string. In case of
>>>> the eDMA controller this property is missing. Add it. After that the
>>>> IOMMU will work with the eDMA core.
>>>>
>>>> Signed-off-by: Michael Walle <michael@...le.cc>
>>>> ---
>>>> arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi | 2 +-
>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi
>>>> b/arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi
>>>> index b152fa90cf5c..aa467bff2209 100644
>>>> --- a/arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi
>>>> +++ b/arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi
>>>> @@ -447,7 +447,7 @@
>>>>
>>>> edma0: dma-controller@...0000 {
>>>> #dma-cells = <2>;
>>>> - compatible = "fsl,ls1028a-edma";
>>>> + compatible = "fsl,ls1028a-edma",
>>>> + "fsl,vf610-edma";
>>> Hi Michael,
>>>
>>> You should change it on bootloader instead of kernel, Some Reg of
>>> LS1028a is different from others, So we used compatible
>>> "fsl,ls1028a-edm" to distinguish "
>>> fsl,vf610-edma".
>>
>> Yes this might be the right thing to do. So since it is NXPs
>> bootloader feel free to
>> fix that ;) Looking at the u-boot code right now, I don't even know it
>> that is the
>> right fix at all. The fixup code in u-boot is SoC independent (its in
>> fsl_icid.h and is
>> enabled with CONFIG_LSCH3, ie your chassis version). For example, the
>> sdhc
>> fixup will scan the nodes for "compatible = fsl,esdhc", which is also
>> the
>> secondary compatible for the "ls1028a-esdhc" compatible.
>>
>> And here is another reason to have it this way: we need backwards
>> compatibility,
>> the are already boards out there whose bootloader will fix-up the
>> "old" node.
>> Thus I don't see any other possibilty.
>>
> [Peng Ma] OK, There is non fixed on uboot.
> I will fix it on uboot, if you want to use now, please change the
> uboot as below:
As I told you, I cannot be changed for shipped bootloaders. While it can
be
changed for newer ones, I would really like to retain backwards
compatibility.
And so should you.
That being said, I don't see a problem in having both compatibles. Linux
will use the ls1028a-emda one and u-boot will fix up the "older"
vf610-edma one.
Unfortunately, this patch will not only affect eDMA but all other
drivers
which uses eDMA, eg. sound, lpuart, i2c and maybe DSPI.
-michael
>
> diff --git a/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> b/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> index d9d125e8ba..db9dd69548 100644
> --- a/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> +++ b/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> @@ -14,7 +14,7 @@ struct icid_id_table icid_tbl[] = {
> SET_SDHC_ICID(1, FSL_SDMMC_STREAM_ID),
> SET_SDHC_ICID(2, FSL_SDMMC2_STREAM_ID),
> SET_SATA_ICID(1, "fsl,ls1028a-ahci", FSL_SATA1_STREAM_ID),
> - SET_EDMA_ICID(FSL_EDMA_STREAM_ID),
> + SET_EDMA_ICID_LS1028(FSL_EDMA_STREAM_ID),
> SET_QDMA_ICID("fsl,ls1028a-qdma", FSL_DMA_STREAM_ID),
> SET_GPU_ICID("fsl,ls1028a-gpu", FSL_GPU_STREAM_ID),
> SET_DISPLAY_ICID(FSL_DISPLAY_STREAM_ID),
> diff --git a/arch/arm/include/asm/arch-fsl-layerscape/fsl_icid.h
> b/arch/arm/include/asm/arch-fsl-layerscape/fsl_icid.h
> index 37e2fe4e66..15d0b60dbe 100644
> --- a/arch/arm/include/asm/arch-fsl-layerscape/fsl_icid.h
> +++ b/arch/arm/include/asm/arch-fsl-layerscape/fsl_icid.h
> @@ -144,6 +144,10 @@ extern int fman_icid_tbl_sz;
> SET_GUR_ICID("fsl,vf610-edma", streamid, spare3_amqr,\
> EDMA_BASE_ADDR)
>
> +#define SET_EDMA_ICID_LS1028(streamid) \
> + SET_GUR_ICID("fsl,ls1028a-edma", streamid, spare3_amqr,\
> + EDMA_BASE_ADDR)
> +
> #define SET_GPU_ICID(compat, streamid) \
> SET_GUR_ICID(compat, streamid, misc1_amqr,\
> GPU_BASE_ADDR)
>
> BR,
> Peng
>> -michael
>>
>>>
>>> Thanks,
>>> Peng
>>>> reg = <0x0 0x22c0000 0x0 0x10000>,
>>>> <0x0 0x22d0000 0x0 0x10000>,
>>>> <0x0 0x22e0000 0x0 0x10000>;
>>>> --
>>>> 2.20.1
Powered by blists - more mailing lists