RE: [PATCH v2 2/2] usb: dwc3: Add workaround for host mode VBUS glitch when boot

From: Ran Wang
Date: Tue Feb 26 2019 - 00:42:28 EST


Hi Leo,

> -----Original Message-----
> From: Yang Li <pku.leo@xxxxxxxxx>
> Sent: Tuesday, February 26, 2019 05:29
> To: Ran Wang <ran.wang_1@xxxxxxx>
> Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>; Rob Herring
> <robh+dt@xxxxxxxxxx>; Mark Rutland <mark.rutland@xxxxxxx>; Felipe Balbi
> <balbi@xxxxxxxxxx>; linux-usb@xxxxxxxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2 2/2] usb: dwc3: Add workaround for host mode VBUS
> glitch when boot
>
> On Tue, Feb 19, 2019 at 1:24 AM Ran Wang <ran.wang_1@xxxxxxx> wrote:
> >
> > When DWC3 is set to host mode by programming register DWC3_GCTL, VBUS
> > (or its control signal) will be turned on immediately on related Root
> > Hub ports. Then, the VBUS is turned off for a little while(15us) when
> > do xhci reset (conducted by xhci driver) and back to normal finally,
> > we can observe a negative glitch of related signal happen.
> >
> > This VBUS glitch might cause some USB devices enumeration fail if
> > kernel boot with them connected. Such as
> > LS1012AFWRY/LS1043ARDB/LX2160AQDS /LS1088ARDB with Kingston 16GB
> > USB2.0/Kingston USB3.0/JetFlash Transcend 4GB USB2.0 drives. The fail
> > cases include enumerated as full-speed device or report wrong device
> descriptor, etc.
> >
> > One SW workaround which can fix this is by programing all xhci
> > PORTSC[PP] to 0 to turn off VBUS immediately after setting host mode
> > in DWC3 driver (per signal measurement result, it will be too late to
> > do it in xhci-plat.c or xhci.c). Then, after xhci reset complete in
> > xhci driver, PORTSC[PP]s' value will back to 1 automatically and VBUS
> > on at that time, no glitch happen and normal enumeration process has no
> impact.
> >
> > Signed-off-by: Ran Wang <ran.wang_1@xxxxxxx>
> > ---
> > Changes in v2:
> > - Rename related property to 'snps,host-vbus-glitches'
> > - Rename related dwc member to 'host_vbus_glitches'
> > - Add member 'host_vbus_glitches' description in 'dwc3'
> >
> > drivers/usb/dwc3/core.c | 47
> +++++++++++++++++++++++++++++++++++++++++++++++
> > drivers/usb/dwc3/core.h | 12 +++++++++++-
> > 2 files changed, 58 insertions(+), 1 deletions(-)
> >
> > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c index
> > a1b126f..02d11bc 100644
> > --- a/drivers/usb/dwc3/core.c
> > +++ b/drivers/usb/dwc3/core.c
> > @@ -100,6 +100,41 @@ static int dwc3_get_dr_mode(struct dwc3 *dwc)
> > return 0;
> > }
> >
> > +/*
> > + * dwc3_power_of_all_roothub_ports - Power off all Root hub ports
> > + * @dwc3: Pointer to our controller context structure */ static void
> > +dwc3_power_off_all_roothub_ports(struct dwc3 *dwc) {
> > + int i, port_num;
> > + u32 reg, op_regs_base, offset;
> > + void __iomem *xhci_regs;
> > +
> > + /* xhci regs is not mapped yet, do it temperary here */
> > + if (dwc->xhci_resources[0].start) {
> > + xhci_regs = ioremap(dwc->xhci_resources[0].start,
> > + DWC3_XHCI_REGS_END);
> > + if (IS_ERR(xhci_regs)) {
> > + dev_err(dwc->dev, "Failed to ioremap xhci_regs\n");
> > + return;
> > + }
> > +
> > + op_regs_base = HC_LENGTH(readl(xhci_regs));
> > + reg = readl(xhci_regs + XHCI_HCSPARAMS1);
> > + port_num = HCS_MAX_PORTS(reg);
> > +
> > + for (i = 1; i <= port_num; i++) {
> > + offset = op_regs_base + XHCI_PORTSC_BASE +
> > + 0x10*(i-1);
>
> Define the 0x10 as macro? And why don't you have the initial value of i loop to
> be 0 to avoid the (i-1) here?

OK, thanks for pointing out.

> > + reg = readl(xhci_regs + offset);
> > + reg &= ~PORT_POWER;
> > + writel(reg, xhci_regs + offset);
> > + }
> > +
> > + iounmap(xhci_regs);
> > + } else
> > + dev_err(dwc->dev, "xhci base reg invalid\n"); }
> > +
> > void dwc3_set_prtcap(struct dwc3 *dwc, u32 mode) {
> > u32 reg;
> > @@ -109,6 +144,15 @@ void dwc3_set_prtcap(struct dwc3 *dwc, u32 mode)
> > reg |= DWC3_GCTL_PRTCAPDIR(mode);
> > dwc3_writel(dwc->regs, DWC3_GCTL, reg);
> >
> > + /*
> > + * We have to power off all Root hub ports immediately after DWC3 set
> > + * to host mode to avoid VBUS glitch happen when xhci get reset later.
> > + */
> > + if (dwc->host_vbus_glitches) {
> > + if (mode == DWC3_GCTL_PRTCAP_HOST)
> > + dwc3_power_off_all_roothub_ports(dwc);
> > + }
> > +
> > dwc->current_dr_role = mode;
> > }
> >
> > @@ -1306,6 +1350,9 @@ static void dwc3_get_properties(struct dwc3 *dwc)
> > dwc->dis_metastability_quirk = device_property_read_bool(dev,
> > "snps,dis_metastability_quirk");
> >
> > + dwc->host_vbus_glitches = device_property_read_bool(dev,
> > + "snps,host-vbus-glitches");
> > +
> > dwc->lpm_nyet_threshold = lpm_nyet_threshold;
> > dwc->tx_de_emphasis = tx_de_emphasis;
> >
> > diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h index
> > df87641..9b3a7ed 100644
> > --- a/drivers/usb/dwc3/core.h
> > +++ b/drivers/usb/dwc3/core.h
> > @@ -606,6 +606,14 @@
> > #define DWC3_OSTS_VBUSVLD BIT(1)
> > #define DWC3_OSTS_CONIDSTS BIT(0)
> >
> > +/* Partial XHCI Register and Bit fields for quirk */
> > +#define XHCI_HCSPARAMS1 0x4
> > +#define XHCI_PORTSC_BASE 0x400
> > +#define PORT_POWER (1 << 9)
> > +#define HCS_MAX_PORTS(p) (((p) >> 24) & 0x7f)
> > +#define XHCI_HC_LENGTH(p) (((p)>>00)&0x00ff)
> > +#define HC_LENGTH(p) XHCI_HC_LENGTH(p)
> > +
> > /* Structures */
> >
> > struct dwc3_trb;
> > @@ -1024,6 +1032,8 @@ struct dwc3_scratchpad_array {
> > * 2 - No de-emphasis
> > * 3 - Reserved
> > * @dis_metastability_quirk: set to disable metastability quirk.
> > + * @host-vbus-glitches: set to avoid vbus glitch during
> > + * xhci reset.
> > * @imod_interval: set the interrupt moderation interval in 250ns
> > * increments or 0 to disable.
> > */
> > @@ -1209,6 +1219,7 @@ struct dwc3 {
> > unsigned tx_de_emphasis:2;
> >
> > unsigned dis_metastability_quirk:1;
> > + unsigned host_vbus_glitches:1;
> >
> > u16 imod_interval;
> > };
> > @@ -1217,7 +1228,6 @@ struct dwc3 {
> > #define INCRX_UNDEF_LENGTH_BURST_MODE 1
> >
> > #define work_to_dwc(w) (container_of((w), struct dwc3, drd_work))
> > -
>
> No need to remove the new line here.

OK, will fix it in next version patch

Regards,
Ran
> > /*
> > ----------------------------------------------------------------------
> > ---- */
> >
> > struct dwc3_event_type {
> > --
> > 1.7.1
> >
>
>