From: Andiry Xu <andiry.xu@xxxxxxx>
On AMD SB700/SB800/Hudson-2/3 platforms, USB EHCI controller may read/write
to memory space not allocated to USB controller if there is longer than
normal latency on DMA read encountered. In this condition the exposure will
be encountered only if the driver has following format of Periodic Frame
List link pointer structure:
For any idle periodic schedule, the Frame List link pointers that have the
T-bit set to 1 intending to terminate the use of frame list link pointer
as a physical memory pointer.
Idle periodic schedule Frame List Link pointer shoule be in the following
format to avoid the issue:
Frame list link pointer should be always contains a valid pointer to a
inactive QHead with T-bit set to 0.
Signed-off-by: Andiry Xu <andiry.xu@xxxxxxx>
Acked-by: David Brownell <dbrownell@xxxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxx>
---
drivers/usb/host/ehci-mem.c | 26 ++++++++++++++++++++++++--
drivers/usb/host/ehci-pci.c | 13 +++++++++++++
drivers/usb/host/ehci-sched.c | 21 ++++++++++++++++++---
drivers/usb/host/ehci.h | 2 ++
4 files changed, 57 insertions(+), 5 deletions(-)
diff --git a/drivers/usb/host/ehci-mem.c b/drivers/usb/host/ehci-mem.c
index d36e4e7..12f70c3 100644
--- a/drivers/usb/host/ehci-mem.c
+++ b/drivers/usb/host/ehci-mem.c
@@ -141,6 +141,10 @@ static void ehci_mem_cleanup (struct ehci_hcd *ehci)
qh_put (ehci->async);
ehci->async = NULL;
+ if (ehci->dummy)
+ qh_put(ehci->dummy);
+ ehci->dummy = NULL;
+
/* DMA consistent memory and pools */
if (ehci->qtd_pool)
dma_pool_destroy (ehci->qtd_pool);
@@ -227,8 +231,26 @@ static int ehci_mem_init (struct ehci_hcd *ehci, gfp_t flags)
if (ehci->periodic == NULL) {
goto fail;
}
- for (i = 0; i < ehci->periodic_size; i++)
- ehci->periodic [i] = EHCI_LIST_END(ehci);
+
+ if (ehci->use_dummy_qh) {
+ struct ehci_qh_hw *hw;
+ ehci->dummy = ehci_qh_alloc(ehci, flags);
+ if (!ehci->dummy)
+ goto fail;
+
+ hw = ehci->dummy->hw;
+ hw->hw_next = EHCI_LIST_END(ehci);
+ hw->hw_qtd_next = EHCI_LIST_END(ehci);
+ hw->hw_alt_next = EHCI_LIST_END(ehci);
+ hw->hw_token &= ~QTD_STS_ACTIVE;
+ ehci->dummy->hw = hw;
+
+ for (i = 0; i < ehci->periodic_size; i++)
+ ehci->periodic[i] = ehci->dummy->qh_dma;
+ } else {
+ for (i = 0; i < ehci->periodic_size; i++)
+ ehci->periodic[i] = EHCI_LIST_END(ehci);
+ }
/* software shadow of hardware table */
ehci->pshadow = kcalloc(ehci->periodic_size, sizeof(void *), flags);
diff --git a/drivers/usb/host/ehci-pci.c b/drivers/usb/host/ehci-pci.c
index a1e8d27..01bb72b 100644
--- a/drivers/usb/host/ehci-pci.c
+++ b/drivers/usb/host/ehci-pci.c
@@ -103,6 +103,19 @@ static int ehci_pci_setup(struct usb_hcd *hcd)
if (retval)
return retval;
+ if ((pdev->vendor == PCI_VENDOR_ID_AMD && pdev->device == 0x7808) ||
+ (pdev->vendor == PCI_VENDOR_ID_ATI && pdev->device == 0x4396)) {
+ /* EHCI controller on AMD SB700/SB800/Hudson-2/3 platforms may
+ * read/write memory space which does not belong to it when
+ * there is NULL pointer with T-bit set to 1 in the frame list
+ * table. To avoid the issue, the frame list link pointer
+ * should always contain a valid pointer to a inactive qh.
+ */
+ ehci->use_dummy_qh = 1;
+ ehci_info(ehci, "applying AMD SB700/SB800/Hudson-2/3 EHCI "
+ "dummy qh workaround\n");
+ }
+
/* data structure init */
retval = ehci_init(hcd);
if (retval)
diff --git a/drivers/usb/host/ehci-sched.c b/drivers/usb/host/ehci-sched.c
index a92526d..d9f78eb 100644
--- a/drivers/usb/host/ehci-sched.c
+++ b/drivers/usb/host/ehci-sched.c
@@ -98,7 +98,14 @@ static void periodic_unlink (struct ehci_hcd *ehci, unsigned frame, void *ptr)
*/
*prev_p = *periodic_next_shadow(ehci, &here,
Q_NEXT_TYPE(ehci, *hw_p));
- *hw_p = *shadow_next_periodic(ehci, &here, Q_NEXT_TYPE(ehci, *hw_p));
+
+ if (!ehci->use_dummy_qh ||
+ *shadow_next_periodic(ehci, &here, Q_NEXT_TYPE(ehci, *hw_p))
+ != EHCI_LIST_END(ehci))
+ *hw_p = *shadow_next_periodic(ehci, &here,
+ Q_NEXT_TYPE(ehci, *hw_p));
+ else
+ *hw_p = ehci->dummy->qh_dma;
}
/* how many of the uframe's 125 usecs are allocated? */
@@ -2335,7 +2342,11 @@ restart:
* pointer for much longer, if at all.
*/
*q_p = q.itd->itd_next;
- *hw_p = q.itd->hw_next;
+ if (!ehci->use_dummy_qh ||
+ q.itd->hw_next != EHCI_LIST_END(ehci))
+ *hw_p = q.itd->hw_next;
+ else
+ *hw_p = ehci->dummy->qh_dma;
type = Q_NEXT_TYPE(ehci, q.itd->hw_next);
wmb();
modified = itd_complete (ehci, q.itd);
@@ -2368,7 +2379,11 @@ restart:
* URB completion.
*/
*q_p = q.sitd->sitd_next;
- *hw_p = q.sitd->hw_next;
+ if (!ehci->use_dummy_qh ||
+ q.sitd->hw_next != EHCI_LIST_END(ehci))
+ *hw_p = q.sitd->hw_next;
+ else
+ *hw_p = ehci->dummy->qh_dma;
type = Q_NEXT_TYPE(ehci, q.sitd->hw_next);
wmb();
modified = sitd_complete (ehci, q.sitd);
diff --git a/drivers/usb/host/ehci.h b/drivers/usb/host/ehci.h
index bde823f..ba8eab3 100644
--- a/drivers/usb/host/ehci.h
+++ b/drivers/usb/host/ehci.h
@@ -73,6 +73,7 @@ struct ehci_hcd { /* one per controller */
/* async schedule support */
struct ehci_qh *async;
+ struct ehci_qh *dummy; /* For AMD quirk use */
struct ehci_qh *reclaim;
unsigned scanning : 1;
@@ -131,6 +132,7 @@ struct ehci_hcd { /* one per controller */
unsigned need_io_watchdog:1;
unsigned broken_periodic:1;
unsigned fs_i_thresh:1; /* Intel iso scheduling */
+ unsigned use_dummy_qh:1; /* AMD Frame List table quirk*/
/* required for usb32 quirk */
#define OHCI_CTRL_HCFS (3 << 6)
--
1.7.1
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
