linux-kernel - Re: [PATCH v3 15/15] xhci: tegra: Enable ELPG for runtime/system PM

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] [day] [month] [year] [list]

Message-ID: <d9447d2e-d82d-57cb-b813-c883a502c2d1@nvidia.com>
Date:   Thu, 15 Oct 2020 16:12:34 +0800
From:   JC Kuo <jckuo@...dia.com>
To:     Thierry Reding <thierry.reding@...il.com>
CC:     <gregkh@...uxfoundation.org>, <robh@...nel.org>,
        <jonathanh@...dia.com>, <kishon@...com>,
        <linux-tegra@...r.kernel.org>, <linux-usb@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <devicetree@...r.kernel.org>,
        <nkristam@...dia.com>
Subject: Re: [PATCH v3 15/15] xhci: tegra: Enable ELPG for runtime/system PM

I will amend accordingly and submit new patch.

Thanks for review.
JC

On 9/28/20 10:06 PM, Thierry Reding wrote:
> On Wed, Sep 09, 2020 at 04:10:41PM +0800, JC Kuo wrote:
>> This commit implements the complete programming sequence for ELPG
>> entry and exit.
>>
>>  1. At ELPG entry, invokes tegra_xusb_padctl_enable_phy_sleepwalk()
>>     and tegra_xusb_padctl_enable_phy_wake() to configure XUSB PADCTL
>>     sleepwalk and wake detection circuits to maintain USB lines level
>>     and respond to wake events (wake-on-connect, wake-on-disconnect,
>>     device-initiated-wake).
>>
>>  2. At ELPG exit, invokes tegra_xusb_padctl_disable_phy_sleepwalk()
>>     and tegra_xusb_padctl_disable_phy_wake() to disarm sleepwalk and
>>     wake detection circuits.
>>
>> At runtime suspend, XUSB host controller can enter ELPG to reduce
>> power consumption. When XUSB PADCTL wake detection circuit detects
>> a wake event, an interrupt will be raised. xhci-tegra driver then
>> will invoke pm_runtime_resume() for xhci-tegra.
>>
>> Runtime resume could also be triggered by protocol drivers, this is
>> the host-initiated-wake event. At runtime resume, xhci-tegra driver
>> brings XUSB host controller out of ELPG to handle the wake events.
>>
>> The same ELPG enter/exit procedure will be performed for system
>> suspend/resume path so USB devices can remain connected across SC7.
>>
>> Signed-off-by: JC Kuo <jckuo@...dia.com>
>> ---
>> v3:
>>    use 'unsigned int' for PHY index
>>    remove unnecessary 'else'
>>    drop IRQF_TRIGGER_HIGH when invokes devm_request_threaded_irq()
>>    
>>  drivers/usb/host/xhci-tegra.c | 389 +++++++++++++++++++++++++++++++---
>>  1 file changed, 360 insertions(+), 29 deletions(-)
>>
>> diff --git a/drivers/usb/host/xhci-tegra.c b/drivers/usb/host/xhci-tegra.c
>> index aabff8ee0bb3..ba3f40e78171 100644
>> --- a/drivers/usb/host/xhci-tegra.c
>> +++ b/drivers/usb/host/xhci-tegra.c
>> @@ -15,9 +15,11 @@
>>  #include <linux/kernel.h>
>>  #include <linux/module.h>
>>  #include <linux/of_device.h>
>> +#include <linux/of_irq.h>
>>  #include <linux/phy/phy.h>
>>  #include <linux/phy/tegra/xusb.h>
>>  #include <linux/platform_device.h>
>> +#include <linux/usb/ch9.h>
>>  #include <linux/pm.h>
>>  #include <linux/pm_domain.h>
>>  #include <linux/pm_runtime.h>
>> @@ -224,6 +226,7 @@ struct tegra_xusb {
>>  
>>  	int xhci_irq;
>>  	int mbox_irq;
>> +	int padctl_irq;
>>  
>>  	void __iomem *ipfs_base;
>>  	void __iomem *fpci_base;
>> @@ -269,10 +272,13 @@ struct tegra_xusb {
>>  		dma_addr_t phys;
>>  	} fw;
>>  
>> +	bool suspended;
>>  	struct tegra_xusb_context context;
>>  };
>>  
>>  static struct hc_driver __read_mostly tegra_xhci_hc_driver;
>> +static int tegra_xusb_exit_elpg(struct tegra_xusb *tegra, bool runtime);
>> +static int tegra_xusb_enter_elpg(struct tegra_xusb *tegra, bool runtime);
> 
> Can we reshuffle the code to avoid these predeclarations? Looks like
> they're only used in tegra_xusb_runtime_{suspend,resume}(), so perhaps
> move the implementations right before those?
> 
> Thierry
>