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;
MIME-Version: 1.0
References: <20220905065622.1573811-1-kai.heng.feng@canonical.com>
 <YxWgGKIAvsxwSz85@black.fi.intel.com> <CAAd53p4iV=ne5bDGZ6FxE9bBUVoFh=eXF9_oMPvPzjVj=UVoog@mail.gmail.com>
 <YxWqSYDWe0NitSkL@black.fi.intel.com> <YxX3JeQLbciJ+hhc@black.fi.intel.com>
In-Reply-To: <YxX3JeQLbciJ+hhc@black.fi.intel.com>
From:   Kai-Heng Feng <kai.heng.feng@canonical.com>
Date:   Mon, 5 Sep 2022 23:24:26 +0800
Message-ID: <CAAd53p5hjmPgPNifo-QV9KU4DFbvPnrYV-YD2b6cSXNN=5GLWQ@mail.gmail.com>
Subject: Re: [PATCH] thunderbolt: Resume PCIe bridges after switch is found on
 AMD USB4 controller
To:     Mika Westerberg <mika.westerberg@linux.intel.com>
Cc:     andreas.noever@gmail.com, michael.jamet@intel.com,
        YehezkelShB@gmail.com, sanju.mehta@amd.com,
        mario.limonciello@amd.com, linux-usb@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Sep 5, 2022 at 9:18 PM Mika Westerberg
<mika.westerberg@linux.intel.com> wrote:
>
> Hi,
>
> On Mon, Sep 05, 2022 at 10:50:33AM +0300, Mika Westerberg wrote:
> > > Yes those root ports are resumed to D0 when something is plugged. They
> > > however fail to detect any externel PCIe devices.
> >
> > Hmm, so you see the actual hotplug but the tunneled PCIe link may not be
> > detected? Does the PCIe "Card Present" (or Data Link Layer Active)
> > status change at all or is it always 0?
>
> I wonder if we are simply missing the required delays here? Looking at
> the lspci dump in the bugzilla you refer the root port 03.1 supports
> active link reporting:
>
>                 LnkCap: Port #247, Speed 2.5GT/s, Width x1, ASPM L1, Exit Latency L1 <4us
>                         ClockPM- Surprise- LLActRep+ BwNot- ASPMOptComp+
>
> So when the link goes back to L0 (D3 -> D0 transition) the kernel should
> issue the 100+ ms reset delay in pci_bridge_wait_for_secondary_bus():

This was actually the first thing I tried but it doesn't work. Even a
5 seconds delay doesn't work either.

>
>   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/pci/pci.c#n5045
>
> can you check if that is happening? It should show up in the dmesg when
> CONFIG_PCI_DEBUG=y but I don't see it in yours.

That's because there isn't any child device yet, so the function bails early:
if (!dev->subordinate || list_empty(&dev->subordinate->devices))

Kai-Heng