linux-kernel - [PATCH] USB: early: Handle AMD's spec-compliant identifiers, too

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: <20200401074619.8024-1-jannh@google.com>
Date:   Wed,  1 Apr 2020 09:46:19 +0200
From:   Jann Horn <jannh@...gle.com>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc:     Lu Baolu <baolu.lu@...ux.intel.com>, linux-usb@...r.kernel.org,
        linux-kernel@...r.kernel.org, x86@...nel.org
Subject: [PATCH] USB: early: Handle AMD's spec-compliant identifiers, too

This fixes a bug that causes the USB3 early console to freeze after
printing a single line on AMD machines because it can't parse the
Transfer TRB properly.

The spec at
https://www.intel.com/content/dam/www/public/us/en/documents/technical-specifications/extensible-host-controler-interface-usb-xhci.pdf
says in section "4.5.1 Device Context Index" that the Context Index,
also known as Endpoint ID according to
section "1.6 Terms and Abbreviations", is normally computed as
`DCI = (Endpoint Number * 2) + Direction`, which matches the current
definitions of XDBC_EPID_OUT and XDBC_EPID_IN.

However, the numbering in a Debug Capability Context data structure is
supposed to be different:
Section "7.6.3.2 Endpoint Contexts and Transfer Rings" explains that a
Debug Capability Context data structure has the endpoints mapped to indices
0 and 1.

Change XDBC_EPID_OUT/XDBC_EPID_IN to the spec-compliant values, add
XDBC_EPID_OUT_INTEL/XDBC_EPID_IN_INTEL with Intel's incorrect values, and
let xdbc_handle_tx_event() handle both.

I have verified that with this patch applied, the USB3 early console works
on both an Intel and an AMD machine.

Fixes: aeb9dd1de98c ("usb/early: Add driver for xhci debug capability")
Cc: stable@...r.kernel.org
Signed-off-by: Jann Horn <jannh@...gle.com>
---
 drivers/usb/early/xhci-dbc.c |  8 ++++----
 drivers/usb/early/xhci-dbc.h | 18 ++++++++++++++++--
 2 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/drivers/usb/early/xhci-dbc.c b/drivers/usb/early/xhci-dbc.c
index cac991173ac0..5a462a1d1896 100644
--- a/drivers/usb/early/xhci-dbc.c
+++ b/drivers/usb/early/xhci-dbc.c
@@ -728,19 +728,19 @@ static void xdbc_handle_tx_event(struct xdbc_trb *evt_trb)
 	case COMP_USB_TRANSACTION_ERROR:
 	case COMP_STALL_ERROR:
 	default:
-		if (ep_id == XDBC_EPID_OUT)
+		if (ep_id == XDBC_EPID_OUT || ep_id == XDBC_EPID_OUT_INTEL)
 			xdbc.flags |= XDBC_FLAGS_OUT_STALL;
-		if (ep_id == XDBC_EPID_IN)
+		if (ep_id == XDBC_EPID_IN || ep_id == XDBC_EPID_IN_INTEL)
 			xdbc.flags |= XDBC_FLAGS_IN_STALL;
 
 		xdbc_trace("endpoint %d stalled\n", ep_id);
 		break;
 	}
 
-	if (ep_id == XDBC_EPID_IN) {
+	if (ep_id == XDBC_EPID_IN || ep_id == XDBC_EPID_IN_INTEL) {
 		xdbc.flags &= ~XDBC_FLAGS_IN_PROCESS;
 		xdbc_bulk_transfer(NULL, XDBC_MAX_PACKET, true);
-	} else if (ep_id == XDBC_EPID_OUT) {
+	} else if (ep_id == XDBC_EPID_OUT || ep_id == XDBC_EPID_OUT_INTEL) {
 		xdbc.flags &= ~XDBC_FLAGS_OUT_PROCESS;
 	} else {
 		xdbc_trace("invalid endpoint id %d\n", ep_id);
diff --git a/drivers/usb/early/xhci-dbc.h b/drivers/usb/early/xhci-dbc.h
index 673686eeddd7..6e2b7266a695 100644
--- a/drivers/usb/early/xhci-dbc.h
+++ b/drivers/usb/early/xhci-dbc.h
@@ -120,8 +120,22 @@ struct xdbc_ring {
 	u32			cycle_state;
 };
 
-#define XDBC_EPID_OUT		2
-#define XDBC_EPID_IN		3
+/*
+ * These are the "Endpoint ID" (also known as "Context Index") values for the
+ * OUT Transfer Ring and the IN Transfer Ring of a Debug Capability Context data
+ * structure.
+ * According to the "eXtensible Host Controller Interface for Universal Serial
+ * Bus (xHCI)" specification, section "7.6.3.2 Endpoint Contexts and Transfer
+ * Rings", these should be 0 and 1, and those are the values AMD machines give
+ * you; but Intel machines seem to use the formula from section "4.5.1 Device
+ * Context Index", which is supposed to be used for the Device Context only.
+ * Luckily the values from Intel don't overlap with those from AMD, so we can
+ * just test for both.
+ */
+#define XDBC_EPID_OUT		0
+#define XDBC_EPID_IN		1
+#define XDBC_EPID_OUT_INTEL	2
+#define XDBC_EPID_IN_INTEL	3
 
 struct xdbc_state {
 	u16			vendor;
-- 
2.26.0.rc2.310.g2932bb562d-goog