linux-kernel - [PATCH v3 2/7] xen/pciback: Don't deadlock when unbinding.

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: <1404845909-13563-3-git-send-email-konrad@kernel.org>
Date:	Tue,  8 Jul 2014 14:58:24 -0400
From:	konrad@...nel.org
To:	xen-devel@...ts.xenproject.org, david.vrabel@...rix.com,
	boris.ostrovsky@...cle.com, linux-kernel@...r.kernel.org
Cc:	Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>
Subject: [PATCH v3 2/7] xen/pciback: Don't deadlock when unbinding.

From: Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>

As commit 0a9fd0152929db372ff61b0d6c280fdd34ae8bdb
'xen/pciback: Document the entry points for 'pcistub_put_pci_dev''
explained there are four entry points in this function.
Two of them are when the user fiddles in the SysFS to
unbind a device which might be in use by a guest or not.

Both 'unbind' states will cause a deadlock as the the PCI lock has
already been taken, which then pci_device_reset tries to take.

We can omit this by detecting this and using the  'already
locked' variant of the pci_device_reset - but we don't need to
do either as that is done much later when the PCIback device
structures are destroyed. As such lets use the 'try_reset'
variant. That means we still end up FLR-ing the device in
between swapping it from one guest to another. The FLR when
'unbind' part is still taken care by pci_device_reset.

Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@...cle.com>
---
 drivers/xen/xen-pciback/pci_stub.c |   11 ++++++++---
 1 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/xen/xen-pciback/pci_stub.c b/drivers/xen/xen-pciback/pci_stub.c
index 8a69a9f..b1fb099 100644
--- a/drivers/xen/xen-pciback/pci_stub.c
+++ b/drivers/xen/xen-pciback/pci_stub.c
@@ -277,10 +277,15 @@ void pcistub_put_pci_dev(struct pci_dev *dev)
 	 * (so it's ready for the next domain)
 	 */

-	/* This is OK - we are running from workqueue context
-	 * and want to inhibit the user from fiddling with 'reset'
+	/* We might be holding the PCI lock (see comment at the top of the
+	 * function) - as such try lock and if we can't then don't worry -
+	 * as either:
+	 *  - we are 'unbind' in which case 'pcistub_device_release' gets
+	 *    called which does the locked version of this.
+	 *  - the toolstack has the smarts to do it and has done the
+	 *    reset on SysFS before assigning this device to another guest.
 	 */
-	pci_reset_function(dev);
+	pci_try_reset_function(dev);
 	pci_restore_state(dev);

 	/* This disables the device. */
-- 
1.7.7.6

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/