Re: [PATCH v9 2/2] PCI: rockchip: Add Rockchip RK356X host controller driver

From: xxm <xxm@rock-chips.com>
To: "Pali Rohár" <pali@kernel.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
	Bjorn Helgaas <bhelgaas@google.com>,
	linux-pci@vger.kernel.org, linux-rockchip@lists.infradead.org,
	devicetree@vger.kernel.org, robh+dt@kernel.org,
	Johan Jonker <jbx6244@gmail.com>,
	Heiko Stuebner <heiko@sntech.de>,
	Peter Geis <pgwipeout@gmail.com>,
	Kever Yang <kever.yang@rock-chips.com>,
	kernel test robot <lkp@intel.com>,
	Shawn Lin <shawn.lin@rock-chips.com>
Subject: Re: [PATCH v9 2/2] PCI: rockchip: Add Rockchip RK356X host controller driver
Date: Thu, 24 Jun 2021 16:58:22 +0800	[thread overview]
Message-ID: <64de9d9c-1cc9-bf6f-00c5-2360ddfa4b24@rock-chips.com> (raw)
In-Reply-To: <20210624082314.mw3ilcufswmb635m@pali>

Hi,

在 2021/6/24 16:23, Pali Rohár 写道:
> On Thursday 24 June 2021 10:08:54 xxm wrote:
>> 在 2021/6/23 22:33, Lorenzo Pieralisi 写道:
>>> On Thu, May 06, 2021 at 10:35:44AM +0800, Simon Xue wrote:
>>>> +static int rockchip_pcie_start_link(struct dw_pcie *pci)
>>>> +{
>>>> +	struct rockchip_pcie *rockchip = to_rockchip_pcie(pci);
>>>> +
>>>> +	/* Reset device */
>>>> +	gpiod_set_value_cansleep(rockchip->rst_gpio, 0);
>>>> +
>>>> +	rockchip_pcie_enable_ltssm(rockchip);
>>>> +
>>>> +	/*
>>>> +	 * PCIe requires the refclk to be stable for 100µs prior to releasing
>>>> +	 * PERST. See table 2-4 in section 2.6.2 AC Specifications of the PCI
>>>> +	 * Express Card Electromechanical Specification, 1.1. However, we don't
>>>> +	 * know if the refclk is coming from RC's PHY or external OSC. If it's
>>>> +	 * from RC, so enabling LTSSM is the just right place to release #PERST.
>>>> +	 * We need more extra time as before, rather than setting just
>>>> +	 * 100us as we don't know how long should the device need to reset.
>>>> +	 */
>>>> +	msleep(100);
>>> Any rationale behind the time chosen ?
>> We found some device need about 30ms, so 100ms here just leave more room for
>> other devices.
> Can you share information which PCIe card needs 30ms?
>
> Last year I did tests with more WiFi AC cards and "the slowest" one was
> Compex WLE1216 which needed about 11ms (more than 10ms). All other cards
> were happy with just 1-2ms.

Sorry, it was about 5 years ago, I can't find the specific card now.

By the way, we also take other's upstream code "msleep(100)" as a reference.

>
>