linux-kernel - [PATCH 1/1] nvme-pci: fix hang during error recovery when the PCI device is isolated

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20220712124453.2227362-2-schnelle@linux.ibm.com>
Date:   Tue, 12 Jul 2022 14:44:53 +0200
From:   Niklas Schnelle <schnelle@...ux.ibm.com>
To:     Christoph Hellwig <hch@....de>, Keith Busch <kbusch@...nel.org>
Cc:     Stefan Roese <sr@...x.de>, Matthew Rosato <mjrosato@...ux.ibm.com>,
        linux-nvme@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: [PATCH 1/1] nvme-pci: fix hang during error recovery when the PCI device is isolated

On s390 and powerpc PCI devices are isolated when an error is detected
and driver->err_handler->error_detected is called with an inaccessible
PCI device and PCI channel state set to pci_channel_io_frozen
(see Step 1 in Documentation/PCI/pci-error-recovery.rst).

In the case of NVMe devices nvme_error_detected() then calls
nvme_dev_disable(dev, false) and requests a reset. After a successful
reset the device is accessible again and nvme_slot_reset() resets the
controller and queues nvme_reset_work() which then recovers the
controller.

Since commit b98235d3a471 ("nvme-pci: harden drive presence detect in
nvme_dev_disable()") however nvme_dev_disable() no longer freezes the
queues if pci_device_is_present() returns false. This is the case for an
isolated PCI device. In principle this makes sense as there are no
accessible hardware queues to run. The problem though is that for
a previously live reset controller with online queues nvme_reset_work()
calls nvme_wait_freeze() which, without the freeze having been
initiated, then hangs forever. Fix this by starting the freeze in
nvme_slot_reset() which is the earliest point where we know the device
should be accessible again.

Fixes: b98235d3a471 ("nvme-pci: harden drive presence detect in nvme_dev_disable()")
Signed-off-by: Niklas Schnelle <schnelle@...ux.ibm.com>
---
 drivers/nvme/host/pci.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 193b44755662..7c0c61b74c30 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -3399,6 +3399,7 @@ static pci_ers_result_t nvme_slot_reset(struct pci_dev *pdev)
 	dev_info(dev->ctrl.device, "restart after slot reset\n");
 	pci_restore_state(pdev);
 	nvme_reset_ctrl(&dev->ctrl);
+	nvme_start_freeze(&dev->ctrl);
 	return PCI_ERS_RESULT_RECOVERED;
 }

-- 
2.34.1