linux-kernel - Re: [PATCH v3 1/3] PCI: apple: Follow the PCIe specifications when resetting the port

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]

Date:   Wed, 24 Nov 2021 12:56:06 +0000
From:   Robin Murphy <robin.murphy@....com>
To:     Marc Zyngier <maz@...nel.org>, linux-kernel@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-pci@...r.kernel.org
Cc:     Pali Rohár <pali@...nel.org>,
        Alyssa Rosenzweig <alyssa@...enzweig.io>,
        Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Mark Kettenis <mark.kettenis@...all.nl>,
        Luca Ceresoli <luca@...aceresoli.net>, kernel-team@...roid.com
Subject: Re: [PATCH v3 1/3] PCI: apple: Follow the PCIe specifications when
 resetting the port

On 2021-11-23 18:06, Marc Zyngier wrote:
> While the Apple PCIe driver works correctly when directly booted
> from the firmware, it fails to initialise when the kernel is booted
> from a bootloader using PCIe such as u-boot.
> 
> That's beacuse we're missing a proper reset of the port (we only
> clear the reset, but never assert it).
> 
> The PCIe spec requirements are two-fold:
> 
> - #PERST must be asserted before setting up the clocks, and
>    stay asserted for at least 100us (Tperst-clk).
> 
> - Once #PERST is deasserted, the OS must wait for at least 100ms
>    "from the end of a Conventional Reset" before we can start talking
>    to the devices
> 
> Implementing this results in a booting system.
> 
> Fixes: 1e33888fbe44 ("PCI: apple: Add initial hardware bring-up")
> Acked-by: Pali Rohár <pali@...nel.org>
> Signed-off-by: Marc Zyngier <maz@...nel.org>
> Cc: Alyssa Rosenzweig <alyssa@...enzweig.io>
> Cc: Lorenzo Pieralisi <lorenzo.pieralisi@....com>
> Cc: Bjorn Helgaas <bhelgaas@...gle.com>
> ---
>   drivers/pci/controller/pcie-apple.c | 10 ++++++++++
>   1 file changed, 10 insertions(+)
> 
> diff --git a/drivers/pci/controller/pcie-apple.c b/drivers/pci/controller/pcie-apple.c
> index 1bf4d75b61be..957960a733c4 100644
> --- a/drivers/pci/controller/pcie-apple.c
> +++ b/drivers/pci/controller/pcie-apple.c
> @@ -539,13 +539,23 @@ static int apple_pcie_setup_port(struct apple_pcie *pcie,
>   
>   	rmw_set(PORT_APPCLK_EN, port->base + PORT_APPCLK);
>   
> +	/* Engage #PERST before setting up the clock */
> +	gpiod_set_value(reset, 0);

FWIW, given that getting the GPIO with GPIOD_OUT_LOW should have had 
this effect in the first place, if this isn't a no-op at this point then 
it would hint at something being more significantly wrong down at the 
GPIO/pinctrl end :/

Once you fix the polarity in the later patch, though, adding the 
explicit reset assertion here does seem a far nicer option than fiddling 
the flags to preserve the implicit assertion earlier.

Cheers,
Robin.

> +
>   	ret = apple_pcie_setup_refclk(pcie, port);
>   	if (ret < 0)
>   		return ret;
>   
> +	/* The minimal Tperst-clk value is 100us (PCIe CMS r2.0, 2.6.2) */
> +	usleep_range(100, 200);
> +
> +	/* Deassert #PERST */
>   	rmw_set(PORT_PERST_OFF, port->base + PORT_PERST);
>   	gpiod_set_value(reset, 1);
>   
> +	/* Wait for 100ms after #PERST deassertion (PCIe r2.0, 6.6.1) */
> +	msleep(100);
> +
>   	ret = readl_relaxed_poll_timeout(port->base + PORT_STATUS, stat,
>   					 stat & PORT_STATUS_READY, 100, 250000);
>   	if (ret < 0) {
>