| Message ID | 20161021153714.GA4221@wunner.de (mailing list archive) |
|---|---|
| State | New, archived |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
On Fri, Oct 21, 2016 at 05:37:14PM +0200, Lukas Wunner wrote: > On Tue, Sep 27, 2016 at 04:23:32PM -0400, Keith Busch wrote: > > --- a/drivers/pci/hotplug/pciehp_pci.c > > +++ b/drivers/pci/hotplug/pciehp_pci.c > > @@ -109,6 +109,8 @@ int pciehp_unconfigure_device(struct slot *p_slot) > > break; > > } > > } > > + if (!presence) > > + dev->is_removed = 1; > > pci_stop_and_remove_bus_device(dev); > > /* > > * Ensure that no new Requests will be generated from > > Sorry for the delay Keith, I finally got around to test v3 of your > series with hot-removed Thunderbolt devices on Apple Macs. > > I've found that the above isn't sufficient, it's necessary to also > set the is_removed bit on any child devices. E.g. on my system > when an Apple Gigabit Ethernet adapter is plugged in, the topology > looks like this: Thanks for the catch. Your proposal looks good to me. I'll send a new revision incorporating something like this that the dpc driver can also use. > With your patch above, the is_removed bit is only set on 0000:09:00.0 > but not on its children. Consequently the "tg3" driver tries to > access the hot-removed Broadcom 57762 Ethernet chip as before, > causing a soft lockup. Is that something that can be fixed in the tg3 driver? I don't think drivers can rely on this patch to fense off their unintended access since we can't stop tg3 from accesses a removed device before 'is_removed' is set. -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, Oct 21, 2016 at 12:15:16PM -0400, Keith Busch wrote: > On Fri, Oct 21, 2016 at 05:37:14PM +0200, Lukas Wunner wrote: > > With your patch above, the is_removed bit is only set on 0000:09:00.0 > > but not on its children. Consequently the "tg3" driver tries to > > access the hot-removed Broadcom 57762 Ethernet chip as before, > > causing a soft lockup. > > Is that something that can be fixed in the tg3 driver? I don't think > drivers can rely on this patch to fense off their unintended access since > we can't stop tg3 from accesses a removed device before 'is_removed' > is set. I haven't tested yet what happens when the adapter is unplugged while packets are in-flight, but at least unplugging works fine when the adapter is idle (with your series plus the small changes I outlined). *Without* your series, I have to set the interface to down with ifconfig before unplugging. If I ever forget that, the machine locks up: NMI watchdog: BUG: soft lockup - CPU#2 stuck for 22s! [kworker/2:2:299] ... Workqueue: pciehp-4 pciehp_power_thread RIP: 0010:[<ffffffffa105b01d>] [<ffffffffa105b01d>] tg3_read32+0xd/0x10 [tg3] ... Call Trace: [<ffffffffa1063610>] ? tg3_stop_block.constprop.126+0x80/0x110 [tg3] [<ffffffffa1066298>] ? tg3_abort_hw+0x68/0x2f0 [tg3] [<ffffffffa106654d>] ? tg3_halt+0x2d/0x180 [tg3] [<ffffffffa1072a07>] ? tg3_stop+0x157/0x210 [tg3] [<ffffffffa1072aeb>] ? tg3_close+0x2b/0xe0 [tg3] [<ffffffff81465ff4>] ? __dev_close_many+0x84/0xd0 [<ffffffff814660b4>] ? dev_close_many+0x74/0x100 [<ffffffff8146790b>] ? rollback_registered_many+0xfb/0x2e0 [<ffffffff81467b19>] ? rollback_registered+0x29/0x40 [<ffffffff81468950>] ? unregister_netdevice_queue+0x40/0x90 [<ffffffff814689b8>] ? unregister_netdev+0x18/0x20 [<ffffffffa106604b>] ? tg3_remove_one+0x8b/0x130 [tg3] [<ffffffff8130b556>] ? pci_device_remove+0x36/0xb0 [<ffffffff813df92a>] ? __device_release_driver+0x9a/0x140 [<ffffffff813df9ee>] ? device_release_driver+0x1e/0x30 [<ffffffff81304bf4>] ? pci_stop_bus_device+0x84/0xa0 [<ffffffff81304b9b>] ? pci_stop_bus_device+0x2b/0xa0 [<ffffffff81304b9b>] ? pci_stop_bus_device+0x2b/0xa0 [<ffffffff81304cee>] ? pci_stop_and_remove_bus_device+0xe/0x20 [<ffffffff8131ecea>] ? pciehp_unconfigure_device+0x9a/0x180 [<ffffffff8131e7ef>] ? pciehp_disable_slot+0x3f/0xb0 [<ffffffff8131e8e5>] ? pciehp_power_thread+0x85/0xa0 [<ffffffff810855df>] ? process_one_work+0x19f/0x3d0 [<ffffffff8108585d>] ? worker_thread+0x4d/0x450 [<ffffffff81085810>] ? process_one_work+0x3d0/0x3d0 [<ffffffff8108b32d>] ? kthread+0xbd/0xe0 [<ffffffff8108b270>] ? kthread_create_on_node+0x170/0x170 [<ffffffff8155ee1f>] ? ret_from_fork+0x3f/0x70 [<ffffffff8108b270>] ? kthread_create_on_node+0x170/0x170 Being able to just unplug without having to think of ifconfig is already a massive improvement. Thanks, Lukas -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/pci/hotplug/pciehp_pci.c b/drivers/pci/hotplug/pciehp_pci.c index 299ea5e..ec26eb7 100644 --- a/drivers/pci/hotplug/pciehp_pci.c +++ b/drivers/pci/hotplug/pciehp_pci.c @@ -73,6 +73,12 @@ int pciehp_configure_device(struct slot *p_slot) return ret; } +static int set_is_removed_cb(struct pci_dev *pdev, void *unused) +{ + pdev->is_removed = 1; + return 0; +} + int pciehp_unconfigure_device(struct slot *p_slot) { int rc = 0; @@ -109,8 +115,11 @@ int pciehp_unconfigure_device(struct slot *p_slot) break; } } - if (!presence) + if (!presence) { dev->is_removed = 1; + if (pci_has_subordinate(dev)) + pci_walk_bus(dev->subordinate, set_is_removed_cb, NULL); + } pci_stop_and_remove_bus_device(dev); /* * Ensure that no new Requests will be generated from