Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <2880d5bb-61ce-a4d7-f870-b5549a5d88f2@linaro.org>
Date:   Tue, 9 May 2023 11:31:48 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.0
Subject: Re: [EXTERNAL] Re: [PATCH] arm64: defconfig: enable PCIe controller
 on TI platforms
Content-Language: en-US
To:     "Verma, Achal" <a-verma1@ti.com>, Arnd Bergmann <arnd@arndb.de>,
        Vignesh Raghavendra <vigneshr@ti.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>,
        Bjorn Andersson <quic_bjorande@quicinc.com>,
        Geert Uytterhoeven <geert+renesas@glider.be>,
        Konrad Dybcio <konrad.dybcio@linaro.org>,
        =?UTF-8?B?TsOtY29sYXMgRi4gUi4gQS4gUHJhZG8=?= 
        <nfraprado@collabora.com>, Mark Brown <broonie@kernel.org>,
        =?UTF-8?B?UmFmYcWCIE1pxYJlY2tp?= <rafal@milecki.pl>,
        Nishanth Menon <nm@ti.com>, Milind Parab <mparab@cadence.com>,
        Swapnil Kashinath Jakhade <sjakhade@cadence.com>
Cc:     linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
References: <20230509063431.1203367-1-a-verma1@ti.com>
 <46b0dbd4-5695-1a2d-8d06-0a60a7c3a151@linaro.org>
 <de6432db-0da0-b535-1f44-04541116be8b@ti.com>
 <63dc3d1b-ba17-48ee-a02e-83b2903e360a@app.fastmail.com>
 <b48613f5-9442-5a41-34fb-760d9edf3682@ti.com>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <b48613f5-9442-5a41-34fb-760d9edf3682@ti.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 09/05/2023 11:19, Verma, Achal wrote:
> 
> Hello,
> On 5/9/2023 1:58 PM, Arnd Bergmann wrote:
>> On Tue, May 9, 2023, at 10:08, Vignesh Raghavendra wrote:
>>
>>> Also, see [0] for history. We really want these to be
>>> modules unless its necessary for bootup.
>>>
>>> You may want to revive [1] and get it to mainline
>>>
>>> [0]
>>> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2VSBvOn1o+q1PYZaQ6LS9U4cz+DZGuDbisHkwNs2dAAw@mail.gmail.com/
>>> [1]
>>> https://lore.kernel.org/linux-arm-kernel/20230110153805.GA1505901@bhelgaas/
>>
>> Agreed, that seems simple enough. Ideally these should even
>> be removable modules, not just single-load but unremovable.
>>
>> Doing that may require changes to the cadence PCIe host
>> code if that does not support unloading yet (I have not
>> checked), but should not require any changes to the core
>> PCIe host code that supports loadable/removable modules.
>>
>>       Arnd
> So, my understanding is that following change is expected
> +CONFIG_PCIE_CADENCE=m
> +CONFIG_PCIE_CADENCE_HOST=m
> +CONFIG_PCIE_CADENCE_EP=m
> +CONFIG_PCI_J721E=m
> +CONFIG_PCI_J721E_HOST=m
> +CONFIG_PCI_J721E_EP=m
> +CONFIG_PCI_EPF_NTB=m
> 
> I also want to inform that pci_j721e.c is a single file with both host 
> and EP functionality, last attempt to build it as modules depending on 
> host or EP selected failed because of symbols dependency.
> Refactoring of pci_j721e.c into common, host and EP specific files could 
> work similar to the Cadence driver, So I will follow this way and push 
> the changes.
> 
> Please let me know if there are concerns.

Aren't HOST and EP just customizing main module?

Best regards,
Krzysztof