linux-kernel - Re: [RFC PATCH 3/5] usb: core: hcd: Add support for registering secondary RH along with primary HCD

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <06693934-28f2-d59e-b004-62cabd3f9e8e@linux.intel.com>
Date:   Tue, 24 Aug 2021 14:55:34 +0300
From:   Mathias Nyman <mathias.nyman@...ux.intel.com>
To:     Kishon Vijay Abraham I <kishon@...com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Mathias Nyman <mathias.nyman@...el.com>,
        Alan Stern <stern@...land.harvard.edu>
Cc:     linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org,
        chris.chiu@...onical.com
Subject: Re: [RFC PATCH 3/5] usb: core: hcd: Add support for registering
 secondary RH along with primary HCD

On 24.8.2021 13.53, Kishon Vijay Abraham I wrote:
> Add support for registering secondary roothub (RH) along with primary HCD.
> It has been observed with certain PCIe USB cards that as soon as the
> primary HCD is registered, port status change is handled leading to cold
> plug devices getting not detected. For such cases, registering both the
> root hubs along with the second HCD is useful.
> 
> Signed-off-by: Kishon Vijay Abraham I <kishon@...com>
> Suggested-by: Alan Stern <stern@...land.harvard.edu>
> ---
>  drivers/usb/core/hcd.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c
> index 4d7a9f0e2caa..9c8df22a7d9a 100644
> --- a/drivers/usb/core/hcd.c
> +++ b/drivers/usb/core/hcd.c
> @@ -2799,6 +2799,7 @@ int __usb_add_hcd(struct usb_hcd *hcd, unsigned int irqnum, unsigned long irqfla
>  {
>  	int retval;
>  	struct usb_device *rhdev;
> +	struct usb_hcd *shared_hcd = NULL;
>  
>  	if (!hcd->skip_phy_initialization && usb_hcd_is_primary_hcd(hcd)) {
>  		hcd->phy_roothub = usb_phy_roothub_alloc(hcd->self.sysdev);
> @@ -2961,6 +2962,15 @@ int __usb_add_hcd(struct usb_hcd *hcd, unsigned int irqnum, unsigned long irqfla
>  
>  	/* starting here, usbcore will pay attention to this root hub */
>  	if (register_hub) {
> +		shared_hcd = hcd->shared_hcd;
> +		if (shared_hcd) {
> +			retval = register_root_hub(shared_hcd);

There is a possibility we try yo register the shared roothub before it is properly set up here.

For example the mediatek driver (xhci-mtk.c) creates both hcds before adding them,
so hcd->shared_hcd exists when usb_add_hcd() is called for the primary hcd,
causing this code to register the hcd->shared_hcd roothub which is not properly added yet.

How about skipping the new __usb_hcd_pci_probe() and __usb_add_hcd() and instead add a new
flag to hcd->flags, something like HCD_FLAG_DEFER_PRI_RH_REGISTER?

The host controller driver can set this flag in the hcd->driver->start(hcd) callback called
before roothub registration here from usb_add_hcd(). If flag is set we skip the roothub registration.

So something like:
shared_hcd = hcd->share_hcd;

if (!usb_hcd_is_primary_hcd(hcd) && shared_hcd && shared_hcd->flags & HCD_FLAG_DEFER_PRI_RH_REGISTER)
        register_root_hub(shared_hcd)
if (!(hcd->flags & HCD_FLAG_DEFER_PRI_RH_REGISTER))
        register_root_hub(hcd)

-Mathias