lkml.org 
[lkml]   [2008]   [Sep]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [Bug #11382] e1000e: 2.6.27-rc1 corrupts EEPROM/NVM
On Wed, Sep 24, 2008 at 12:36:38AM -0700, David Miller wrote:
> The e1000e side here is reproducable way too easily for it to be the
> same case, as far as I see it.
>

I've been working on a patch to detect (using a timer and checking at
up/down) whether or not the flash has been corrupted, and, if it is
rewrite it with the saved good copy (which obviously only helps if
it's the same boot.)

Unfortunately, I don't have enough time to finish it before I go away
for the weekend, so I'll toss it over the wall and see if it sticks to
anything.

At a glance, one would need to add support for rewriting
adapter->hw.flash from ethtool if someone reprograms the good firmware
back, and writing the good flash back on down/remove if it detects
a change.

Bear in mind, super quick hack, and I haven't even run-tested it yet.

If nobody decides to run with it, I'll probably give it another poke
late tonight.

Definitely-not-signed-off-by-or-tested-by: Kyle

At the very least, if someone pokes in a hexdump of the firmware, at
least we might be able to see some of the method to the madness of the
corruption pattern.

diff --git a/drivers/net/e1000e/e1000.h b/drivers/net/e1000e/e1000.h
index ac4e506..08cce8c 100644
--- a/drivers/net/e1000e/e1000.h
+++ b/drivers/net/e1000e/e1000.h
@@ -168,6 +168,7 @@ struct e1000_adapter {
struct timer_list watchdog_timer;
struct timer_list phy_info_timer;
struct timer_list blink_timer;
+ struct timer_list flash_timer;

struct work_struct reset_task;
struct work_struct watchdog_task;
diff --git a/drivers/net/e1000e/hw.h b/drivers/net/e1000e/hw.h
index 74f263a..ca3f645 100644
--- a/drivers/net/e1000e/hw.h
+++ b/drivers/net/e1000e/hw.h
@@ -863,6 +863,11 @@ struct e1000_hw {

u8 __iomem *hw_addr;
u8 __iomem *flash_address;
+ int flash_len;
+
+ u8 *flash;
+ u8 *flash_backup;
+ spinlock_t flashlock;

struct e1000_mac_info mac;
struct e1000_fc_info fc;
diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index d266510..13f05f8 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -2535,6 +2535,7 @@ void e1000e_down(struct e1000_adapter *adapter)

del_timer_sync(&adapter->watchdog_timer);
del_timer_sync(&adapter->phy_info_timer);
+ del_timer_sync(&adapter->flash_timer);

netdev->tx_queue_len = adapter->tx_queue_len;
netif_carrier_off(netdev);
@@ -2922,6 +2923,33 @@ static void e1000_update_phy_info(unsigned long data)
e1000_get_phy_info(&adapter->hw);
}

+static inline int e1000_test_flash(struct e1000_adapter *adapter)
+{
+ int ret = 0;
+
+ if (adapter->hw.flash && adapter->hw.flash_backup) {
+ spin_lock(&adapter->hw.flashlock);
+ memcpy(adapter->hw.flash_backup, adapter->hw.flash_address,
+ adapter->hw.flash_len);
+ ret = memcmp(adapter->hw.flash, adapter->hw.flash_backup,
+ adapter->hw.flash_len);
+ spin_unlock(&adapter->hw.flashlock);
+ if (ret) {
+ /* dump_eeprom(adapter); */
+ printk(KERN_ERR "AWOOOGA AWOOOGA flash changed\n");
+ }
+ }
+
+ return ret;
+}
+
+static void e1000_flash_test(unsigned long data)
+{
+ struct e1000_adapter *adapter = (struct e1000_adapter *) data;
+ e1000_test_flash(adapter);
+ mod_timer(&adapter->flash_timer, jiffies+(20*HZ));
+}
+
/**
* e1000e_update_stats - Update the board statistics counters
* @adapter: board private structure
@@ -4439,6 +4467,22 @@ static int __devinit e1000_probe(struct pci_dev *pdev,
adapter->hw.flash_address = ioremap(flash_start, flash_len);
if (!adapter->hw.flash_address)
goto err_flashmap;
+
+ adapter->hw.flash_len = (int)flash_len;
+ /* stash away a copy of the flash, and allocate
+ space for a second copy... */
+ if (!adapter->hw.flash) {
+ u8 *flash = kmalloc(flash_len, GFP_KERNEL);
+ u8 *flash_backup = kmalloc(flash_len, GFP_KERNEL);
+ if (flash && flash_backup) {
+ memcpy(flash, adapter->hw.flash_address,
+ adapter->hw.flash_len);
+ adapter->hw.flash = flash;
+ adapter->hw.flash_backup = flash_backup;
+ spin_lock_init(&adapter->hw.flashlock);
+ }
+ }
+
}

/* construct the net_device struct */
@@ -4570,6 +4614,10 @@ static int __devinit e1000_probe(struct pci_dev *pdev,
adapter->phy_info_timer.function = &e1000_update_phy_info;
adapter->phy_info_timer.data = (unsigned long) adapter;

+ init_timer(&adapter->flash_timer);
+ adapter->flash_timer.function = &e1000_flash_test;
+ adapter->flash_timer.data = (unsigned long) adapter;
+
INIT_WORK(&adapter->reset_task, e1000_reset_task);
INIT_WORK(&adapter->watchdog_task, e1000_watchdog_task);

@@ -4641,6 +4689,9 @@ static int __devinit e1000_probe(struct pci_dev *pdev,

e1000_print_device_info(adapter);

+ /* every twenty seconds, test the flash */
+ mod_timer(&adapter->flash_timer, jiffies+(HZ*20));
+
return 0;

err_register:
@@ -4690,6 +4741,7 @@ static void __devexit e1000_remove(struct pci_dev *pdev)
set_bit(__E1000_DOWN, &adapter->state);
del_timer_sync(&adapter->watchdog_timer);
del_timer_sync(&adapter->phy_info_timer);
+ del_timer_sync(&adapter->flash_timer);

flush_scheduled_work();


\
 
 \ /
  Last update: 2008-09-24 21:13    [W:0.228 / U:0.780 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site