Use delayed work instead of timers to run the watchdog of the e1000e
driver.
Simplify the code with one less middle function.
Signed-off-by: Detlev Casanova <[email protected]>
---
drivers/net/ethernet/intel/e1000e/e1000.h | 3 +-
drivers/net/ethernet/intel/e1000e/netdev.c | 54 ++++++++++++----------
2 files changed, 30 insertions(+), 27 deletions(-)
diff --git a/drivers/net/ethernet/intel/e1000e/e1000.h b/drivers/net/ethernet/intel/e1000e/e1000.h
index be13227f1697..942ab74030ca 100644
--- a/drivers/net/ethernet/intel/e1000e/e1000.h
+++ b/drivers/net/ethernet/intel/e1000e/e1000.h
@@ -186,12 +186,11 @@ struct e1000_phy_regs {
/* board specific private data structure */
struct e1000_adapter {
- struct timer_list watchdog_timer;
struct timer_list phy_info_timer;
struct timer_list blink_timer;
struct work_struct reset_task;
- struct work_struct watchdog_task;
+ struct delayed_work watchdog_task;
const struct e1000_info *ei;
diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
index 0e09bede42a2..22a5d594115f 100644
--- a/drivers/net/ethernet/intel/e1000e/netdev.c
+++ b/drivers/net/ethernet/intel/e1000e/netdev.c
@@ -39,6 +39,8 @@ static int debug = -1;
module_param(debug, int, 0);
MODULE_PARM_DESC(debug, "Debug level (0=none,...,16=all)");
+static struct workqueue_struct *e1000_workqueue;
+
static const struct e1000_info *e1000_info_tbl[] = {
[board_82571] = &e1000_82571_info,
[board_82572] = &e1000_82572_info,
@@ -1780,7 +1782,8 @@ static irqreturn_t e1000_intr_msi(int __always_unused irq, void *data)
}
/* guard against interrupt when we're going down */
if (!test_bit(__E1000_DOWN, &adapter->state))
- mod_timer(&adapter->watchdog_timer, jiffies + 1);
+ queue_delayed_work(e1000_workqueue,
+ &adapter->watchdog_task, 1);
}
/* Reset on uncorrectable ECC error */
@@ -1860,7 +1863,8 @@ static irqreturn_t e1000_intr(int __always_unused irq, void *data)
}
/* guard against interrupt when we're going down */
if (!test_bit(__E1000_DOWN, &adapter->state))
- mod_timer(&adapter->watchdog_timer, jiffies + 1);
+ queue_delayed_work(e1000_workqueue,
+ &adapter->watchdog_task, 1);
}
/* Reset on uncorrectable ECC error */
@@ -1905,7 +1909,8 @@ static irqreturn_t e1000_msix_other(int __always_unused irq, void *data)
hw->mac.get_link_status = true;
/* guard against interrupt when we're going down */
if (!test_bit(__E1000_DOWN, &adapter->state))
- mod_timer(&adapter->watchdog_timer, jiffies + 1);
+ queue_delayed_work(e1000_workqueue,
+ &adapter->watchdog_task, 1);
}
if (!test_bit(__E1000_DOWN, &adapter->state))
@@ -4278,7 +4283,6 @@ void e1000e_down(struct e1000_adapter *adapter, bool reset)
napi_synchronize(&adapter->napi);
- del_timer_sync(&adapter->watchdog_timer);
del_timer_sync(&adapter->phy_info_timer);
spin_lock(&adapter->stats64_lock);
@@ -5150,25 +5154,11 @@ static void e1000e_check_82574_phy_workaround(struct e1000_adapter *adapter)
}
}
-/**
- * e1000_watchdog - Timer Call-back
- * @data: pointer to adapter cast into an unsigned long
- **/
-static void e1000_watchdog(struct timer_list *t)
-{
- struct e1000_adapter *adapter = from_timer(adapter, t, watchdog_timer);
-
- /* Do the rest outside of interrupt context */
- schedule_work(&adapter->watchdog_task);
-
- /* TODO: make this use queue_delayed_work() */
-}
-
static void e1000_watchdog_task(struct work_struct *work)
{
struct e1000_adapter *adapter = container_of(work,
struct e1000_adapter,
- watchdog_task);
+ watchdog_task.work);
struct net_device *netdev = adapter->netdev;
struct e1000_mac_info *mac = &adapter->hw.mac;
struct e1000_phy_info *phy = &adapter->hw.phy;
@@ -5395,8 +5385,8 @@ static void e1000_watchdog_task(struct work_struct *work)
/* Reset the timer */
if (!test_bit(__E1000_DOWN, &adapter->state))
- mod_timer(&adapter->watchdog_timer,
- round_jiffies(jiffies + 2 * HZ));
+ queue_delayed_work(e1000_workqueue, &adapter->watchdog_task,
+ round_jiffies(2 * HZ));
}
#define E1000_TX_FLAGS_CSUM 0x00000001
@@ -7251,11 +7241,20 @@ static int e1000_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
goto err_eeprom;
}
- timer_setup(&adapter->watchdog_timer, e1000_watchdog, 0);
+ e1000_workqueue = alloc_workqueue("%s", WQ_MEM_RECLAIM, 0,
+ e1000e_driver_name);
+
+ if (!e1000_workqueue) {
+ err = -ENOMEM;
+ goto err_workqueue;
+ }
+
+ INIT_DELAYED_WORK(&adapter->watchdog_task, e1000_watchdog_task);
+ queue_delayed_work(e1000_workqueue, &adapter->watchdog_task, 0);
+
timer_setup(&adapter->phy_info_timer, e1000_update_phy_info, 0);
INIT_WORK(&adapter->reset_task, e1000_reset_task);
- INIT_WORK(&adapter->watchdog_task, e1000_watchdog_task);
INIT_WORK(&adapter->downshift_task, e1000e_downshift_workaround);
INIT_WORK(&adapter->update_phy_task, e1000e_update_phy_task);
INIT_WORK(&adapter->print_hang_task, e1000_print_hw_hang);
@@ -7349,6 +7348,9 @@ static int e1000_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
return 0;
err_register:
+ flush_workqueue(e1000_workqueue);
+ destroy_workqueue(e1000_workqueue);
+err_workqueue:
if (!(adapter->flags & FLAG_HAS_AMT))
e1000e_release_hw_control(adapter);
err_eeprom:
@@ -7395,15 +7397,17 @@ static void e1000_remove(struct pci_dev *pdev)
*/
if (!down)
set_bit(__E1000_DOWN, &adapter->state);
- del_timer_sync(&adapter->watchdog_timer);
del_timer_sync(&adapter->phy_info_timer);
cancel_work_sync(&adapter->reset_task);
- cancel_work_sync(&adapter->watchdog_task);
cancel_work_sync(&adapter->downshift_task);
cancel_work_sync(&adapter->update_phy_task);
cancel_work_sync(&adapter->print_hang_task);
+ cancel_delayed_work(&adapter->watchdog_task);
+ flush_workqueue(e1000_workqueue);
+ destroy_workqueue(e1000_workqueue);
+
if (adapter->flags & FLAG_HAS_HW_TIMESTAMP) {
cancel_work_sync(&adapter->tx_hwtstamp_work);
if (adapter->tx_hwtstamp_skb) {
--
2.22.0
On Tue, 2019-06-18 at 18:08 -0400, Detlev Casanova wrote:
> Use delayed work instead of timers to run the watchdog of the e1000e
> driver.
>
> Simplify the code with one less middle function.
>
> Signed-off-by: Detlev Casanova <[email protected]>
> ---
> drivers/net/ethernet/intel/e1000e/e1000.h | 3 +-
> drivers/net/ethernet/intel/e1000e/netdev.c | 54 ++++++++++++----------
> 2 files changed, 30 insertions(+), 27 deletions(-)
>
This patch is causing a NULL pointer dereference trace (or just a freeze) on a number of my regression systems when I unload the e1000e driver. The crashing test systems each have multiple e1000e
based ports, I can pull the systems from the rack and change cards out if necessary, though it's not a trivial task getting to them.
Systems affected had the following e1000e chipsets:
- 82579LM Gigabit Network Connection rev 05) LOM
- 82574L LOM
- I217-LM LOM
- 82572EI (Rimon)
- 80003ES2LAN x2 LOMs
- 82572EI Add in card
- 80003ES2LAN x2 LOMs
- 82571EB (Ophir) copper (dual port card)
- 82571EB (Ophir) fibre (dual port card)
Here is the trace caputrued from the last system listed, the 80003ES2LAN LOMs and 2 dual port Ophir cards:
-------------------------------------------------------------
u1462:[ttyS1]/root> rmmod e1000e
[ 131.799739] BUG: kernel NULL pointer dereference, address: 0000000000000000
[ 131.806908] #PF: supervisor write access in kernel mode
[ 131.812285] #PF: error_code(0x0002) - not-present page
[ 131.817565] PGD 8000000072327067 P4D 8000000072327067 PUD 3050f067 PMD 0
[ 131.824540] Oops: 0002 [#1] SMP PTI
[ 131.828133] CPU: 3 PID: 4306 Comm: rmmod Not tainted 5.2.0-rc5_next-queue_dev-queue_15bf1c7+ #9
[ 131.837073] Hardware name: Supermicro X7DBX/X7DBX, BIOS 2.1 06/23/2008
[ 131.843791] RIP: 0010:_raw_spin_lock_irq+0x13/0x30
[ 131.848722] Code: 00 75 02 f3 c3 e9 3d 16 8a ff 0f 1f 00 66 2e 0f 1f 84 00 00 00 00 00 66 66 66 66 90 fa 66 66 90 66 66 90 31 c0 ba 01 00 00 00 <f0> 0f b1 17 0f 94 c2 84 d2 74 02 f3 c3 89 c6 e9 b2
fd 89 ff 66 0f
[ 131.867998] RSP: 0018:ffffb85601f87d58 EFLAGS: 00010046
[ 131.873377] RAX: 0000000000000000 RBX: ffff9a5a9faa5b90 RCX: 0000000000000003
[ 131.880702] RDX: 0000000000000001 RSI: 0000000000000002 RDI: 0000000000000000
[ 131.888032] RBP: 0000000000000002 R08: 0000000000000000 R09: 0000000000000000
[ 131.895356] R10: ffff9a5ab340a550 R11: ffffe13500cd0280 R12: 0000000000000003
[ 131.902679] R13: 0000000000000002 R14: ffff9a5a9faa4400 R15: 0000000000000000
[ 131.910003] FS: 00007f37948a9740(0000) GS:ffff9a5afcac0000(0000) knlGS:0000000000000000
[ 131.918325] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 131.924229] CR2: 0000000000000000 CR3: 0000000020602000 CR4: 00000000000006e0
[ 131.931557] Call Trace:
[ 131.934093] flush_workqueue_prep_pwqs+0x55/0x120
[ 131.938929] flush_workqueue+0x1b6/0x460
[ 131.942987] e1000_remove+0x93/0x190 [e1000e]
[ 131.947466] pci_device_remove+0x3b/0xc0
[ 131.951511] device_release_driver_internal+0xdf/0x1a0
[ 131.956797] driver_detach+0x43/0x80
[ 131.960487] bus_remove_driver+0x55/0xd0
[ 131.964535] pci_unregister_driver+0x26/0xa0
[ 131.968938] __x64_sys_delete_module+0x16c/0x250
[ 131.973698] ? exit_to_usermode_loop+0xaa/0xc6
[ 131.978270] do_syscall_64+0x5b/0x1b0
[ 131.982041] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 131.987237] RIP: 0033:0x7f3793d82397
[ 131.990923] Code: 73 01 c3 48 8b 0d f9 7a 2c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 b8 b0 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d c9 7a 2c 00 f7
d8 64 89 01 48
[ 132.010190] RSP: 002b:00007ffc6b3dea18 EFLAGS: 00000206 ORIG_RAX: 00000000000000b0
[ 132.017972] RAX: ffffffffffffffda RBX: 0000000001a23190 RCX: 00007f3793d82397
[ 132.034233] RDX: 00007f3793df6b20 RSI: 0000000000000800 RDI: 0000000001a231f8
[ 132.050479] RBP: 0000000000000000 R08: 00007f379404b060 R09: 00007f3793df6b20
[ 132.066683] R10: 00007ffc6b3de5e0 R11: 0000000000000206 R12: 00007ffc6b3e0d4a
[ 132.082874] R13: 0000000000000000 R14: 0000000001a23190 R15: 0000000001a23010
[ 132.098961] Modules linked in: xt_CHECKSUM iptable_mangle xt_MASQUERADE iptable_nat nf_nat xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ipt_REJECT nf_reject_ipv4 tun bridge stp llc
ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter dm_mirror dm_region_hash dm_log dm_mod coretemp kvm_intel kvm iTCO_wdt i2c_i801 gpio_ich iTCO_vendor_support lpc_ich pcspkr i5000_edac
sg irqbypass acpi_cpufreq i5k_amb nfsd auth_rpcgss nfs_acl lockd grace sunrpc ip_tables xfs libcrc32c radeon sr_mod sd_mod cdrom ata_generic pata_acpi i2c_algo_bit drm_kms_helper syscopyarea
sysfillrect sysimgblt fb_sys_fops ttm drm ata_piix libata serio_raw e1000e(-) e1000
[ 132.215961] CR2: 0000000000000000
[ 132.229473] ---[ end trace 31a92a65bb543b98 ]---
[ 132.244260] RIP: 0010:_raw_spin_lock_irq+0x13/0x30
[ 132.259217] Code: 00 75 02 f3 c3 e9 3d 16 8a ff 0f 1f 00 66 2e 0f 1f 84 00 00 00 00 00 66 66 66 66 90 fa 66 66 90 66 66 90 31 c0 ba 01 00 00 00 <f0> 0f b1 17 0f 94 c2 84 d2 74 02 f3 c3 89 c6 e9 b2
fd 89 ff 66 0f
[ 132.299254] RSP: 0018:ffffb85601f87d58 EFLAGS: 00010046
[ 132.315166] RAX: 0000000000000000 RBX: ffff9a5a9faa5b90 RCX: 0000000000000003
[ 132.333000] RDX: 0000000000000001 RSI: 0000000000000002 RDI: 0000000000000000
[ 132.350749] RBP: 0000000000000002 R08: 0000000000000000 R09: 0000000000000000
[ 132.368150] R10: ffff9a5ab340a550 R11: ffffe13500cd0280 R12: 0000000000000003
[ 132.385257] R13: 0000000000000002 R14: ffff9a5a9faa4400 R15: 0000000000000000
[ 132.402202] FS: 00007f37948a9740(0000) GS:ffff9a5afcac0000(0000) knlGS:0000000000000000
[ 132.420165] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 132.435818] CR2: 0000000000000000 CR3: 0000000020602000 CR4: 00000000000006e0
[ 132.453030] Kernel panic - not syncing: Fatal exception
[ 132.468323] Kernel Offset: 0x7e00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
[ 132.489386] ---[ end Kernel panic - not syncing: Fatal exception ]---