Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:6 as permitted sender) client-ip=2620:137:e000::3:6;
Message-ID: <6842cdf9-d73b-4895-891f-993eaf5bee6e@ti.com>
Date:   Thu, 19 Oct 2023 10:14:24 +0530
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
CC:     <lpieralisi@kernel.org>, <robh@kernel.org>, <kw@linux.com>,
        <bhelgaas@google.com>, <linux-pci@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>,
        <linux-arm-kernel@lists.infradead.org>, <r-gunasekaran@ti.com>,
        <srk@ti.com>, Serge Semin <fancer.lancer@gmail.com>,
        <s-vadapalli@ti.com>
Subject: Re: [PATCH v2] PCI: keystone: Fix ks_pcie_v3_65_add_bus() for AM654x
 SoC
Content-Language: en-US
To:     Bjorn Helgaas <helgaas@kernel.org>
References: <20231018163632.GA1364574@bhelgaas>
From:   Siddharth Vadapalli <s-vadapalli@ti.com>
In-Reply-To: <20231018163632.GA1364574@bhelgaas>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hello Bjorn,

On 18/10/23 22:06, Bjorn Helgaas wrote:
> [+cc Serge (please cc people who have commented on previous revisions)]

Sure, I will do so.

> 
> On Wed, Oct 18, 2023 at 01:20:38PM +0530, Siddharth Vadapalli wrote:

...

>>
>> Changes since v1:
>> - Updated patch subject and commit message.
>> - Determined that issue is not with the absence of Link as mentioned in
>>   v1 patch. Even with Link up and endpoint device connected, if
>>   ks_pcie_v3_65_add_bus() is invoked and executed, all reads to the
>>   MSI-X offsets return 0xffffffff when pcieport driver attempts to setup
>>   AER and PME services. The all Fs return value indicates that the MSI-X
>>   configuration is failing even if Endpoint device is connected. This is
>>   because the ks_pcie_v3_65_add_bus() function is not applicable to the
>>   AM654x SoC which uses DW PCIe IP-core version 4.90a.
> 
> Thanks for verifying that this doesn't actually depend on whether the
> link is up.
> 
> I think that means we should be able to get rid of the
> ks_pcie_v3_65_add_bus() callback altogether and instead do this along
> with the rest of the Root Port init.

Yes, I will follow Serge's suggestion of adding a new pci_ops structure for the
AM654x SoC which uses the new 4.90a controller. I have described it at:
https://lore.kernel.org/r/ba217723-1501-4e72-b143-e0047266ea9a@ti.com/
and am summarizing it below:

I will add the following:
static struct pci_ops ks_pcie_am6_ops = {
	.map_bus = dw_pcie_own_conf_map_bus,
	.read = pci_generic_config_read,
	.write = pci_generic_config_write,
};
which shall be used for AM654x SoC

I will also modify the contents of ks_pcie_host_init() as follows:
if(ks_pcie->is_am6)
	pp->bridge->ops = &ks_pcie_am6_ops;
else
	pp->bridge->ops = &ks_pcie_ops;

which will ensure that the .add_bus() method is no longer applicable to the
AM654x SoC, which was the case prior to commit 6ab15b5e7057.

I shall post the v3 patch with the above changes and also Cc Serge.

-- 
Regards,
Siddharth.