igc: fix BUG() during the driver probe
Subsystems (4)
cover igc: fix BUG() during the driver probe
Hi,
I noticed that with the v6.10-rc1 kernel I get BUG() because the spinlock used by the PTP code is not properly initialized. This is due the ordering change done recently and I suspect there might be other issues the change causes. These two patches try to address this by reverting the commit in question and then fixing the original issue by using PCI device pointer instead for logging.
Mika Westerberg (2):
Revert "igc: fix a log entry using uninitialized netdev"
igc: Use PCI device pointer in logging in igc_ptp_init()
drivers/net/ethernet/intel/igc/igc_main.c | 5 ++--- drivers/net/ethernet/intel/igc/igc_ptp.c | 4 ++-- 2 files changed, 4 insertions(+), 5 deletions(-)
Signature · 1 line
2.43.0
PATCH 1/2 Revert "igc: fix a log entry using uninitialized netdev"
This reverts commit 86167183a17e03ec77198897975e9fdfbd53cb0b.
The commit in question moved call to igc_ptp_init() to happen after igc_reset() which calls igc_ptp_reset() with the spinlock not yet initialized so we get following splat:
BUG: spinlock bad magic on CPU#4, udevd/249 lock: 0xffff98f317e6d4b0, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0 CPU: 4 PID: 249 Comm: udevd Not tainted 6.10.0-rc1+ #1312 Call Trace: <TASK> dump_stack_lvl+0x6c/0x90 dump_stack+0x10/0x20 spin_bug+0x8c/0xc0 do_raw_spin_lock+0x6c/0xc0 _raw_spin_lock_irqsave+0x30/0x40 igc_ptp_clear_tx_tstamp+0x2e/0xc0 [igc] igc_ptp_set_timestamp_mode+0x191/0x280 [igc] igc_ptp_reset+0x33/0x230 [igc] igc_reset+0xba/0x100 [igc] igc_probe+0x7d1/0xa10 [igc]
It is likely that there are other things igc_ptp_init() does that are required by igc_ptp_reset(), so for this reason revert the commit in question.
--- drivers/net/ethernet/intel/igc/igc_main.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/igc/igc_main.c +2 -3
@@ -7028,6 +7028,8 @@ static int igc_probe(struct pci_dev *pdev,device_set_wakeup_enable (&adapter->pdev->dev, adapter->flags &IGC_FLAG_WOL_SUPPORTED ); +igc_ptp_init (adapter); +igc_tsn_clear_schedule (adapter);/* reset the hardware with the new settings */
@@ -7049,9 +7051,6 @@ static int igc_probe(struct pci_dev *pdev,/* Check if Media Autosense is enabled */ adapter->ei = *ei; -/* do hw tstamp init after resetting */ -igc_ptp_init (adapter); -/* print pcie link status and MAC address */ pcie_print_link_status (pdev);netdev_info (netdev,"MAC: %pM\n" , netdev->dev_addr);
Signature · 1 line
2.43.0
Re: [PATCH 1/2] Revert "igc: fix a log entry using uninitialized netdev"
Quoted 1 line · Mika Westerberg
This reverts commit
86167183a17e03ec77198897975e9fdfbd53cb0b.
TWIMC, Sasha Neftin already submitted a revert for that commit on Wednesday:
https://lore.kernel.org/all/[email protected]/
Regression reports for this problem I'm tracking:
https://lore.kernel.org/lkml/CABXGCsOkiGxAfA9tPKjYX7wqjBZQxqK2PzTcW-RgLfgo8G74EQ@mail.gmail.com/
https://lore.kernel.org/intel-gfx/[email protected]/
Ciao, Thorsten
PATCH 2/2 igc: Use PCI device pointer in logging in igc_ptp_init()
In reply to Mika WesterbergAs described in the commit 86167183a17e ("igc: fix a log entry using uninitialized netdev"), the netdev has not yet been registered so we get weird log entry like:
[ 5.133667] igc 0000:01:00.0 (unnamed net_device) (uninitialized): PHC added
Fix this by using the PCI device pointer instead, that's valid and available at this point.
--- drivers/net/ethernet/intel/igc/igc_ptp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/igc/igc_ptp.c +2 -2
@@ -1169,9 +1169,9 @@ void igc_ptp_init(struct igc_adapter *adapter) &adapter->pdev->dev);if (IS_ERR (adapter->ptp_clock)) { adapter->ptp_clock =NULL ; -netdev_err (netdev,"ptp_clock_register failed\n" ); +dev_err (&adapter->pdev->dev,"ptp_clock_register failed\n" ); }else if (adapter->ptp_clock) { -netdev_info (netdev,"PHC added\n" ); +dev_info (&adapter->pdev->dev,"PHC added\n" ); adapter->ptp_flags |=IGC_PTP_ENABLED ; } }
Signature · 1 line
2.43.0