EHCI software retries break Supermicro IPKVM
From: Grzegorz Nosek
Date: Thu Apr 19 2012 - 11:35:09 EST
Hi,
(Supermicro support: cc'd as you might be interested)
Commit a2c2706e1043c17139c2dafd171c4a5cf008ef7e introduced software retries for transient USB errors. Unfortunately, it turns out that this change breaks the Supermicro AOC SIMSO IPKVM board plugged into a SYS-6015B-3R server, X7DBR-3 board. It renders unusable both the keyboard and media redirection as soon as the kernel boots (display works fine). An excerpt from dmesg:
[ 76.587521] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 1
[ 76.587646] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 2
[ 76.587772] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 3
[ 76.587896] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 4
[ 76.588020] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 5
[ 76.588144] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 6
[ 76.588271] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 7
[ 76.588395] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 8
[ 76.588518] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 9
[ 76.588644] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 10
[ 76.588770] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 11
[ 76.588893] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 12
[ 76.589017] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 13
[ 76.589142] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 14
[ 76.589268] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 15
[ 76.589391] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 16
[ 76.589516] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 17
[ 76.589641] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 18
[ 76.589767] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 19
[ 76.589891] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 20
[ 76.590015] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 21
[ 76.590140] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 22
[ 76.590266] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 23
[ 76.590390] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 24
[ 76.590514] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 25
[ 76.590639] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 26
[ 76.590765] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 27
[ 76.590889] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 28
[ 76.591013] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 29
[ 76.591138] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 30
[ 76.591264] ehci_hcd 0000:00:1d.7: detected XactErr len 0/8 retry 31
[ 76.591388] ehci_hcd 0000:00:1d.7: devpath 7 ep0out 3strikes
[ 76.591402] usb 1-7: can't set config #1, error -71
[ 76.596519] drivers/usb/core/inode.c: creating file '007'
[ 76.596542] hub 1-0:1.0: state 7 ports 8 chg 0000 evt fe80
After disarming the retry logic with the patch below, the KVM works fine again. Now, this is certainly a crude hack but apparently some better checks are needed to determine whether to retry the transaction.
Best regards,
Grzegorz Nosek
---
drivers/usb/host/ehci.h | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/usb/host/ehci.h b/drivers/usb/host/ehci.h
index 0a5fda7..aa3a09e 100644
--- a/drivers/usb/host/ehci.h
+++ b/drivers/usb/host/ehci.h
@@ -373,7 +373,7 @@ struct ehci_qh {
#define QH_STATE_COMPLETING 5 /* don't touch token.HALT */
u8 xacterrs; /* XactErr retry counter */
-#define QH_XACTERR_MAX 32 /* XactErr retry limit */
+#define QH_XACTERR_MAX 1 /* XactErr retry limit */
/* periodic schedule info */
u8 usecs; /* intr bandwidth */
--
1.7.2.3
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/