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-next>] [day] [month] [year] [list]
Message-ID: <20240823055642.27638-1-quic_uaggarwa@quicinc.com>
Date: Fri, 23 Aug 2024 11:26:42 +0530
From: Uttkarsh Aggarwal <quic_uaggarwa@...cinc.com>
To: Thinh Nguyen <Thinh.Nguyen@...opsys.com>,
        Greg Kroah-Hartman
	<gregkh@...uxfoundation.org>
CC: <linux-kernel@...r.kernel.org>, <linux-usb@...r.kernel.org>,
        "Uttkarsh
 Aggarwal" <quic_uaggarwa@...cinc.com>
Subject: [PATCH] usb: dwc3: core: Set GUCTL1[29] to ignore single SE0 glitch for FS/LS operation

Currently in few of Qualcomm chips USB LS mouse not detected
showing following errors:

[ 81.585361] usb 1-1: Device not responding to setup address.
[ 81.798676] usb 1-1: device not accepting address 2, error -71
[ 81.928378] usb 1-1: new low-speed USB device number 3 using xhci-hcd
[ 82.076248] usb 1-1: Device not responding to setup address.
[ 82.290630] usb 1-1: Device not responding to setup address.
[ 82.503877] usb 1-1: device not accepting address 3, error -71
[ 82.512548] usb usb1-port1: attempt power cycle

Based on the Logic analyzer waveforms, It has been identified that
there is skew of about 8nS b/w DP & DM linestate signals
(o/p of PHY & i/p to controller) at the UTMI interface,
Due to this controller is seeing SE0 glitch condition,
this is causing controller to pre-maturely assume that PHY
has sent all the data & is initiating next packet much early,
though in reality PHY is still busy sending previous packets.

Enabling the GUCTL1.FILTER_SE0_FSLS_EOP bit29 allows the controller
to ignore single SE0 glitches on the linestate during transmission.
Only two or more SE0 signals are recognized as a valid EOP.

When this feature is activated, SE0 signals on the linestate
are validated over two consecutive UTMI/ULPI clock edges for EOP detection.

Device mode (FS): If GUCTL1.FILTER_SE0_FSLS_EOP is set,
then for device LPM handshake, the controller ignores single
SE0 glitch on the linestate during transmit. Only two or more
SE0 is considered as a valid EOP on FS port.

Host mode (FS/LS): If GUCTL1.FILTER_SE0_FSLS_EOP is set, then
the controller ignores single SE0 glitch on the linestate during transmit.
Only two or more SE0 is considered as a valid EOP on FS/LS port.

Signed-off-by: Uttkarsh Aggarwal <quic_uaggarwa@...cinc.com>
---
 drivers/usb/dwc3/core.c | 10 ++++++++++
 drivers/usb/dwc3/core.h |  1 +
 2 files changed, 11 insertions(+)

diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
index 3e55838c0001..1edb16d2940c 100644
--- a/drivers/usb/dwc3/core.c
+++ b/drivers/usb/dwc3/core.c
@@ -194,6 +194,16 @@ static void __dwc3_set_mode(struct work_struct *work)
 
 	switch (desired_dr_role) {
 	case DWC3_GCTL_PRTCAP_HOST:
+
+		/*
+		 * Setting GUCTL1 bit 29 so that controller
+		 * will ignore single SE0 glitch on the linestate
+		 * during transmit.
+		 */
+		reg = dwc3_readl(dwc->regs, DWC3_GUCTL1);
+		reg |= DWC3_GUCTL1_FILTER_SE0_FSLS_EOP;
+		dwc3_writel(dwc->regs, DWC3_GUCTL1, reg);
+
 		ret = dwc3_host_init(dwc);
 		if (ret) {
 			dev_err(dwc->dev, "failed to initialize host\n");
diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h
index e3eea965e57b..73e90daafecf 100644
--- a/drivers/usb/dwc3/core.h
+++ b/drivers/usb/dwc3/core.h
@@ -271,6 +271,7 @@
 #define DWC3_GUCTL1_PARKMODE_DISABLE_SS		BIT(17)
 #define DWC3_GUCTL1_PARKMODE_DISABLE_HS		BIT(16)
 #define DWC3_GUCTL1_RESUME_OPMODE_HS_HOST	BIT(10)
+#define DWC3_GUCTL1_FILTER_SE0_FSLS_EOP		BIT(29)
 
 /* Global Status Register */
 #define DWC3_GSTS_OTG_IP	BIT(10)
-- 
2.17.1


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ