linux-kernel - [PATCH] vmw_vmci: Fix irq allocation

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-next>] [day] [month] [year] [list]

Message-ID: <20170326065112.iv65box7muvifxzw@petr-dev3.eng.vmware.com>
Date:   Sat, 25 Mar 2017 23:51:12 -0700
From:   Petr Vandrovec <petr@...are.com>
To:     <gregkh@...uxfoundation.org>
CC:     <hch@....de>, <linux-kernel@...r.kernel.org>,
        <pv-drivers@...are.com>, <petr@...drovec.name>
Subject: [PATCH] vmw_vmci: Fix irq allocation

From: Petr Vandrovec <petr@...are.com>

Change 3bb434 (vmw_vmci: switch to pci_irq_alloc_vectors) converted
vmw_vmci to use pci_irq_alloc_vectors.  Unfortunately for the code
pci_irq_alloc_vectors returns number of allocated vectors on success,
rather than zero.  So successful allocation was always treated as
an error, causing slew of warning messages, and then error that
device probe method returned unexpected value 2.

Checking that error is < 0 fixes problem.

While I was there, I've simplified allocation to do just one
pci_irq_alloc_vectors call: VMCI device currently supports only
two MSI-X vectors, so we either get two dedicated vectors, or
one shared vector.  Nothing else can happen.  So we can ask
for minimum=1, maximum=2, MSI-X | MSI | Legacy, and get back
best possible choice.

Signed-off-by: Petr Vandrovec <petr@...are.com>
Cc: Christoph Hellwig <hch@....de>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Fixes: 3bb434cdcc6a ("vmw_vmci: switch to pci_irq_alloc_vectors")
---
 drivers/misc/vmw_vmci/vmci_guest.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)

diff --git a/drivers/misc/vmw_vmci/vmci_guest.c b/drivers/misc/vmw_vmci/vmci_guest.c
index 9d659542a335..6aea073841c8 100644
--- a/drivers/misc/vmw_vmci/vmci_guest.c
+++ b/drivers/misc/vmw_vmci/vmci_guest.c
@@ -464,7 +464,6 @@ static int vmci_guest_probe_device(struct pci_dev *pdev,
 	}
 
 	vmci_dev->dev = &pdev->dev;
-	vmci_dev->exclusive_vectors = false;
 	vmci_dev->iobase = iobase;
 
 	tasklet_init(&vmci_dev->datagram_tasklet,
@@ -564,16 +563,11 @@ static int vmci_guest_probe_device(struct pci_dev *pdev,
 	 * Enable interrupts.  Try MSI-X first, then MSI, and then fallback on
 	 * legacy interrupts.
 	 */
-	error = pci_alloc_irq_vectors(pdev, VMCI_MAX_INTRS, VMCI_MAX_INTRS,
-			PCI_IRQ_MSIX);
-	if (error) {
-		error = pci_alloc_irq_vectors(pdev, 1, 1,
-				PCI_IRQ_MSIX | PCI_IRQ_MSI | PCI_IRQ_LEGACY);
-		if (error)
-			goto err_remove_bitmap;
-	} else {
-		vmci_dev->exclusive_vectors = true;
-	}
+	error = pci_alloc_irq_vectors(pdev, 1, VMCI_MAX_INTRS,
+			PCI_IRQ_MSIX | PCI_IRQ_MSI | PCI_IRQ_LEGACY);
+	if (error < 0)
+		goto err_remove_bitmap;
+	vmci_dev->exclusive_vectors = error == VMCI_MAX_INTRS;
 
 	/*
 	 * Request IRQ for legacy or MSI interrupts, or for first
-- 
2.11.0