linux-kernel - [PATCH] usb: musb: gadget: fix enumeration on heavy-loaded systems

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: <1366126904-31605-1-git-send-email-ruslan.bilovol@ti.com>
Date:	Tue, 16 Apr 2013 18:41:44 +0300
From:	Ruslan Bilovol <ruslan.bilovol@...com>
To:	balbi@...com, gregkh@...uxfoundation.org,
	linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: [PATCH] usb: musb: gadget: fix enumeration on heavy-loaded systems

>From musb point of view, the Address Assignment sequence during
device enumeration is next:
 - first ep0 interrupt:
	* read the address from USB_REQ_SET_ADDRESS request
	* set up CSR0L.DataEnd bit (that is ACK
	  signalization for the host)
 - second ep0 interrupt:
	* indicates that the request completed successfully
	* set up musb device address
	  Now musb device should answer to this address

>From the host perspective, if peripheral device acquires
SET_ADDRESS request, it now may be accessed only using that address.
However, on heavy loaded systems, time between first and
second musb ep0 interrupts may be too long and musb controller
misses requests between. As result, device enumeration may be
unsuccessful. This can be checked on USB3.0 Host and
using USB3.0 test suite (from usb.org) running ch9 tests
for USB2.0 devices.
Usually 'Addressed state/TD9.1: Device Descriptor Test' will fail

The fix consists in checking CSR0L.DataEnd state and assigning
the device address in the first ep0 interrupt handling, so
delay is as minimal as possible

Signed-off-by: Ruslan Bilovol <ruslan.bilovol@...com>
---
 drivers/usb/musb/musb_gadget_ep0.c |   31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

diff --git a/drivers/usb/musb/musb_gadget_ep0.c b/drivers/usb/musb/musb_gadget_ep0.c
index c9c1ac4..59bc5a5 100644
--- a/drivers/usb/musb/musb_gadget_ep0.c
+++ b/drivers/usb/musb/musb_gadget_ep0.c
@@ -885,6 +885,37 @@ stall:
 finish:
 				musb_writew(regs, MUSB_CSR0,
 						musb->ackpend);
+
+				/*
+				 * If we are at end of SET_ADDRESS sequence,
+				 * update the address immediately if possible,
+				 * otherwise we may miss packets between
+				 * sending ACK from musb side and musb's next
+				 * interrupt handler firing (in which we update
+				 * the address). At least this fixes next
+				 * USB2.0 ch9 test of USB30CV utility:
+				 * "Addressed state - Device Descriptor test"
+				 */
+				if (musb->set_address && (musb->ackpend &
+							MUSB_CSR0_P_DATAEND) &&
+						(musb->ep0_state ==
+						MUSB_EP0_STAGE_STATUSIN)) {
+					u16 ack_delay = 500;
+
+					while ((musb_readw(regs, MUSB_CSR0) &
+							MUSB_CSR0_P_DATAEND) &&
+							--ack_delay) {
+						cpu_relax();
+						udelay(1);
+					}
+
+					if (ack_delay) {
+						musb->set_address = false;
+						musb_writeb(mbase, MUSB_FADDR,
+								musb->address);
+					}
+				}
+
 				musb->ackpend = 0;
 			}
 		}
-- 
1.7.9.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/