From 766da58378855d4216ae47cfe63517acf005af0b Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Wed, 3 May 2017 10:28:50 -0700 Subject: e1000e: fix race condition around skb_tstamp_tx() [ Upstream commit 5012863b7347866764c4a4e58b62fb05346b0d06 ] The e1000e driver and related hardware has a limitation on Tx PTP packets which requires we limit to timestamping a single packet at once. We do this by verifying that we never request a new Tx timestamp while we still have a tx_hwtstamp_skb pointer. Unfortunately the driver suffers from a race condition around this. The tx_hwtstamp_skb pointer is not set to NULL until after skb_tstamp_tx() is called. This function notifies the stack and applications of a new timestamp. Even a well behaved application that only sends a new request when the first one is finished might be woken up and possibly send a packet before we can free the timestamp in the driver again. The result is that we needlessly ignore some Tx timestamp requests in this corner case. Fix this by assigning the tx_hwtstamp_skb pointer prior to calling skb_tstamp_tx() and use a temporary pointer to hold the timestamped skb until that function finishes. This ensures that the application is not woken up until the driver is ready to begin timestamping a new packet. This ensures that well behaved applications do not accidentally race with condition to skip Tx timestamps. Obviously an application which sends multiple Tx timestamp requests at once will still only timestamp one packet at a time. Unfortunately there is nothing we can do about this. Reported-by: David Mirabito Signed-off-by: Jacob Keller Tested-by: Aaron Brown Signed-off-by: Jeff Kirsher Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman --- drivers/net/ethernet/intel/e1000e/netdev.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'drivers/net/ethernet/intel/e1000e/netdev.c') diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c index e356e9187e84..2577a47fe65b 100644 --- a/drivers/net/ethernet/intel/e1000e/netdev.c +++ b/drivers/net/ethernet/intel/e1000e/netdev.c @@ -1182,6 +1182,7 @@ static void e1000e_tx_hwtstamp_work(struct work_struct *work) struct e1000_hw *hw = &adapter->hw; if (er32(TSYNCTXCTL) & E1000_TSYNCTXCTL_VALID) { + struct sk_buff *skb = adapter->tx_hwtstamp_skb; struct skb_shared_hwtstamps shhwtstamps; u64 txstmp; @@ -1190,9 +1191,14 @@ static void e1000e_tx_hwtstamp_work(struct work_struct *work) e1000e_systim_to_hwtstamp(adapter, &shhwtstamps, txstmp); - skb_tstamp_tx(adapter->tx_hwtstamp_skb, &shhwtstamps); - dev_kfree_skb_any(adapter->tx_hwtstamp_skb); + /* Clear the global tx_hwtstamp_skb pointer and force writes + * prior to notifying the stack of a Tx timestamp. + */ adapter->tx_hwtstamp_skb = NULL; + wmb(); /* force write prior to skb_tstamp_tx */ + + skb_tstamp_tx(skb, &shhwtstamps); + dev_kfree_skb_any(skb); } else if (time_after(jiffies, adapter->tx_hwtstamp_start + adapter->tx_timeout_factor * HZ)) { dev_kfree_skb_any(adapter->tx_hwtstamp_skb); -- cgit v1.2.3 From aec0f7859616d2f260deefe034461077684a9a8e Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 31 May 2017 18:50:43 +0300 Subject: e1000e: Undo e1000e_pm_freeze if __e1000_shutdown fails [ Upstream commit 833521ebc65b1c3092e5c0d8a97092f98eec595d ] An error during suspend (e100e_pm_suspend), [ 429.994338] ACPI : EC: event blocked [ 429.994633] e1000e: EEE TX LPI TIMER: 00000011 [ 430.955451] pci_pm_suspend(): e1000e_pm_suspend+0x0/0x30 [e1000e] returns -2 [ 430.955454] dpm_run_callback(): pci_pm_suspend+0x0/0x140 returns -2 [ 430.955458] PM: Device 0000:00:19.0 failed to suspend async: error -2 [ 430.955581] PM: Some devices failed to suspend, or early wake event detected [ 430.957709] ACPI : EC: event unblocked lead to complete failure: [ 432.585002] ------------[ cut here ]------------ [ 432.585013] WARNING: CPU: 3 PID: 8372 at kernel/irq/manage.c:1478 __free_irq+0x9f/0x280 [ 432.585015] Trying to free already-free IRQ 20 [ 432.585016] Modules linked in: cdc_ncm usbnet x86_pkg_temp_thermal intel_powerclamp coretemp mii crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep lpc_ich snd_hda_core snd_pcm mei_me mei sdhci_pci sdhci i915 mmc_core e1000e ptp pps_core prime_numbers [ 432.585042] CPU: 3 PID: 8372 Comm: kworker/u16:40 Tainted: G U 4.10.0-rc8-CI-Patchwork_3870+ #1 [ 432.585044] Hardware name: LENOVO 2356GCG/2356GCG, BIOS G7ET31WW (1.13 ) 07/02/2012 [ 432.585050] Workqueue: events_unbound async_run_entry_fn [ 432.585051] Call Trace: [ 432.585058] dump_stack+0x67/0x92 [ 432.585062] __warn+0xc6/0xe0 [ 432.585065] warn_slowpath_fmt+0x4a/0x50 [ 432.585070] ? _raw_spin_lock_irqsave+0x49/0x60 [ 432.585072] __free_irq+0x9f/0x280 [ 432.585075] free_irq+0x34/0x80 [ 432.585089] e1000_free_irq+0x65/0x70 [e1000e] [ 432.585098] e1000e_pm_freeze+0x7a/0xb0 [e1000e] [ 432.585106] e1000e_pm_suspend+0x21/0x30 [e1000e] [ 432.585113] pci_pm_suspend+0x71/0x140 [ 432.585118] dpm_run_callback+0x6f/0x330 [ 432.585122] ? pci_pm_freeze+0xe0/0xe0 [ 432.585125] __device_suspend+0xea/0x330 [ 432.585128] async_suspend+0x1a/0x90 [ 432.585132] async_run_entry_fn+0x34/0x160 [ 432.585137] process_one_work+0x1f4/0x6d0 [ 432.585140] ? process_one_work+0x16e/0x6d0 [ 432.585143] worker_thread+0x49/0x4a0 [ 432.585145] kthread+0x107/0x140 [ 432.585148] ? process_one_work+0x6d0/0x6d0 [ 432.585150] ? kthread_create_on_node+0x40/0x40 [ 432.585154] ret_from_fork+0x2e/0x40 [ 432.585156] ---[ end trace 6712df7f8c4b9124 ]--- The unwind failures stems from commit 2800209994f8 ("e1000e: Refactor PM flows"), but it may be a later patch that introduced the non-recoverable behaviour. Fixes: 2800209994f8 ("e1000e: Refactor PM flows") Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99847 Signed-off-by: Chris Wilson Signed-off-by: Jani Nikula Tested-by: Aaron Brown Signed-off-by: Jeff Kirsher Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman --- drivers/net/ethernet/intel/e1000e/netdev.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'drivers/net/ethernet/intel/e1000e/netdev.c') diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c index 2577a47fe65b..20d8806d2bff 100644 --- a/drivers/net/ethernet/intel/e1000e/netdev.c +++ b/drivers/net/ethernet/intel/e1000e/netdev.c @@ -6595,12 +6595,17 @@ static int e1000e_pm_thaw(struct device *dev) static int e1000e_pm_suspend(struct device *dev) { struct pci_dev *pdev = to_pci_dev(dev); + int rc; e1000e_flush_lpic(pdev); e1000e_pm_freeze(dev); - return __e1000_shutdown(pdev, false); + rc = __e1000_shutdown(pdev, false); + if (rc) + e1000e_pm_thaw(dev); + + return rc; } static int e1000e_pm_resume(struct device *dev) -- cgit v1.2.3