HPSA related kernel panic on boot in 3.15 rc1 on Proliant with P420i

DISCUSSION [email protected] 2 participants 2 messages Apr 14, 2014

Subsystems (2)

SCSI Hewlett-packard smart array RAID driver (hpsa)

lore

Share & tools

lore

[email protected]

author

2014-04-14 13:04 UTC #0 lore raw

Quoted 14 lines · Darius D.

Hi,
on P420i (2GB FBWC) with latest(5.22?) FW, and 2 SSD smart path enabled RAID0 arrays (1 and 3 SSD), i get panic on initialization. What is sad, i can't capture complete stack trace and it is deep in kernel worker, but top items are:
SA5_performant_intr_mask+0x30/0x30
spin_unlock_irqrestore
hpsa_unit_one
local_pci_probe
... kthread worker stuff

Quoted 5 lines

RIP is calc_bucket_map+0x30/0x3a.
Anyone got idea what is going on? 3.14 was working fine with exact config, so i suspect it could be HPSA smart path code related.

Try the patch below, and see also: http://marc.info/?t=139701135500001&r=1&w=2

-- steve

Quoted 1 line

From f6b44f25c451d32cca5cf2e9825e0636132e69cf Mon Sep 17 00:00:00 2001

From: Stephen M. Cameron <[email protected]>
Date: Thu, 10 Apr 2014 16:44:59 -0500
Subject: [PATCH] hpsa: fix uninitialized variable in hpsa_put_ctlr_into_performant_mode()

Without this, you'll see a null pointer dereference in hpsa_enter_performant_mode().

Signed-off-by Stephen M. Cameron <[email protected]>

---
 drivers/scsi/hpsa.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

drivers/scsi/hpsa.c +4 -0

index 8cf4a0c..ef4dfdd 100644
@@ -7463,6 +7463,10 @@ static void hpsa_put_ctlr_into_performant_mode(struct ctlr_info *h)
 	if (hpsa_simple_mode)
 		return;
 
+	trans_support = readl(&(h->cfgtable->TransportSupport));
+	if (!(trans_support & PERFORMANT_MODE))
+		return;
+
 	/* Check for I/O accelerator mode support */
 	if (trans_support & CFGTBL_Trans_io_accel1) {
 		transMethod |= CFGTBL_Trans_io_accel1 |

Signature · 1 line

1.7.1

Darius D.

2014-04-14 13:29 UTC #1 lore raw

Hi,

i've found that patch, but somehow relevant thread was showing only on
http://www.spinics.net/lists/linux-scsi/msg73770.html but not on marc.info

Anyway, after applying patch, my system passes HPSA initialization, ...but goes boom with unknown FS on block 0,0 error, but it is probably not related, as somehow i see initrd generated, even if config was copied from 3.14 where there was non.

http://tinypic.com/r/311qvc3/8

Looking into this problem, but probably not HPSA related, sorry for repeat bug report.

Thanks for help,

Darius.

Quoted 57 lines · On Mon, Apr 14, 2014 at 4:03 PM, <[email protected]> wrote:

Quoted 14 lines · Darius D.
Hi,
on P420i (2GB FBWC) with latest(5.22?) FW, and 2 SSD smart path enabled RAID0 arrays (1 and 3 SSD), i get panic on initialization. What is sad, i can't capture complete stack trace and it is deep in kernel worker, but top items are:
SA5_performant_intr_mask+0x30/0x30
spin_unlock_irqrestore
hpsa_unit_one
local_pci_probe
... kthread worker stuff
Quoted 5 lines
RIP is calc_bucket_map+0x30/0x3a.
Anyone got idea what is going on? 3.14 was working fine with exact config, so i suspect it could be HPSA smart path code related.
Try the patch below, and see also: http://marc.info/?t=139701135500001&r=1&w=2
-- steve
From f6b44f25c451d32cca5cf2e9825e0636132e69cf Mon Sep 17 00:00:00 2001
From: Stephen M. Cameron <[email protected]>
Date: Thu, 10 Apr 2014 16:44:59 -0500
Subject: [PATCH] hpsa: fix uninitialized variable in hpsa_put_ctlr_into_performant_mode()
Without this, you'll see a null pointer dereference in hpsa_enter_performant_mode().
Signed-off-by Stephen M. Cameron <[email protected]>
---
 drivers/scsi/hpsa.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)
drivers/scsi/hpsa.c +4 -1
index 8cf4a0c..ef4dfdd 100644
@@ -7463,6 +7463,10 @@ static void hpsa_put_ctlr_into_performant_mode(struct ctlr_info *h)
        if (hpsa_simple_mode)
                return;
 
+       trans_support = readl(&(h->cfgtable->TransportSupport));
+       if (!(trans_support & PERFORMANT_MODE))
+               return;
+
        /* Check for I/O accelerator mode support */
        if (trans_support & CFGTBL_Trans_io_accel1) {
                transMethod |= CFGTBL_Trans_io_accel1 |
--
1.7.1