Re: [RFC PATCH 3/3] mwifiex: pcie: add reset_wsid quirk for Surface 3

From: Andy Shevchenko <andy.shevchenko@gmail.com>
To: Bjorn Helgaas <helgaas@kernel.org>
Cc: "Jonas Dreßler" <verdre@v0yd.nl>,
	"Amitkumar Karwar" <amitkarwar@gmail.com>,
	"Ganapathi Bhat" <ganapathi017@gmail.com>,
	"Sharvari Harisangam" <sharvari.harisangam@nxp.com>,
	"Xinming Hu" <huxinming820@gmail.com>,
	"Kalle Valo" <kvalo@codeaurora.org>,
	"David S. Miller" <davem@davemloft.net>,
	"Jakub Kicinski" <kuba@kernel.org>,
	"Tsuchiya Yuto" <kitakar@gmail.com>,
	linux-wireless@vger.kernel.org, netdev@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org,
	"Maximilian Luz" <luzmaximilian@gmail.com>,
	"Andy Shevchenko" <andriy.shevchenko@linux.intel.com>,
	"Bjorn Helgaas" <bhelgaas@google.com>,
	"Pali Rohár" <pali@kernel.org>
Subject: Re: [RFC PATCH 3/3] mwifiex: pcie: add reset_wsid quirk for Surface 3
Date: Sat, 5 Jun 2021 00:26:52 +0300	[thread overview]
Message-ID: <CAHp75Vcab8COGmQdY7M0ag5qS0mTXpOmvbyP426aLJ=wNfM+0A@mail.gmail.com> (raw)
In-Reply-To: <20210604211447.GA2240001@bjorn-Precision-5520>

On Sat, Jun 5, 2021 at 12:15 AM Bjorn Helgaas <helgaas@kernel.org> wrote:
> On Sat, May 22, 2021 at 03:18:27PM +0200, Jonas Dreßler wrote:
> > From: Tsuchiya Yuto <kitakar@gmail.com>
> >
> > This commit adds reset_wsid quirk and uses this quirk for Surface 3 on
> > card reset.
> >
> > To reset mwifiex on Surface 3, it seems that calling the _DSM method
> > exists in \_SB.WSID [1] device is required.
> >
> > On Surface 3, calling the _DSM method removes/re-probes the card by
> > itself. So, need to place the reset function before performing FLR and
> > skip performing any other reset-related works.
>
> Maybe this is a nit-pick, but I understand "probing" to be something
> the OS does, namely what we normally call "enumeration," i.e.,
> discovering a device.
>
> So it sounds like the _DSM causes a logical hot-removal of the card,
> which the PCI hotplug driver should notice and it should remove the
> driver and remove the pci_dev.
>
> And the _DSM also causes a hot-add (reading the code below, it looks
> like this is actually a second _DSM),

_DSM can be only one (single) per device node in ACPI.
But _DSM may have "functions", that's what we see here.

> which the PCI hotplug driver
> should also notice and enumerate the bus (i.e., it reads config space
> looking for a device).  This all would cause a new pci_dev to be
> allocated, resources assigned for its BARs, and the driver .probe()
> method to be called again?
>
> That seems like a lot, so maybe I didn't understand what's actually
> happening.

-- 
With Best Regards,
Andy Shevchenko