linux-kernel - Re: [Bug] PCI: Enable INTx if BIOS left them disabled

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]

Date:	Mon, 10 Mar 2014 18:10:03 -0600
From:	Bjorn Helgaas <bhelgaas@...gle.com>
To:	Andreas Noever <andreas.noever@...il.com>
Cc:	Yinghai Lu <yinghai@...nel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>,
	Rajat Jain <rajatxjain@...il.com>,
	"Rafael J. Wysocki" <rjw@...ysocki.net>
Subject: Re: [Bug] PCI: Enable INTx if BIOS left them disabled - triggers
 during rescan

On Tue, Mar 11, 2014 at 12:10:02AM +0100, Andreas Noever wrote:
> On Mon, Mar 10, 2014 at 9:43 PM, Bjorn Helgaas <bhelgaas@...gle.com> wrote:
> > On Sat, Mar 8, 2014 at 2:55 AM, Andreas Noever <andreas.noever@...il.com> wrote:
> >> On Sat, Mar 8, 2014 at 2:04 AM, Bjorn Helgaas <bhelgaas@...gle.com> wrote:
> >
> >>> If we did that, INTX_DISABLE would be cleared by the first
> >>> pci_enable_device() and pci_reenable_device() wouldn't do anything,
> >>> leaving it cleared.  The resulting state (cleared) would be the same,
> >>> but the transitions would be gone, and maybe those are important.
> >> Just a quick note: With pci_intx_for_msi removed no hotplug events are
> >> ever delivered. Everything else still works though. So it is either a
> >> problem specific to Thunderbolt bridges or maybe it just affects
> >> hotplug (and PME?) interrupts.
> >
> > Interesting.  This is on a MacBook, isn't it?  If you have Mac OS on
> > it, is there a way you can do the equivalent of lspci on it?  I'm
> > curious about whether it sets INTx_DISABLE when it enables MSI.
> 
> lspci -vv and lspci -vv -xxxx attached (yes, someone made a port).
> 
> It looks like Mac OS sets DisINTx for all devices that have MSI
> enabled. The only exception is the Thunderbolt controller (no
> idea...). But the hotplug bridges all have DisINTx+.

OK, thanks.  I don't know what to make of that.

Here's a possible patch; can you try it out?


PCI: Do not enable INTx in pci_reenable_device()

From: Bjorn Helgaas <bhelgaas@...gle.com>

Previously we cleared the Interrupt Disable bit in do_pci_enable_device(),
which is used by both pci_enable_device() and pci_reenable_device().  But
we use pci_reenable_device() after the driver may have enabled MSI or
MSI-X, and we *set* Interrupt Disable as part of enabling MSI/MSI-X.

The pciehp hot-plug path uses pci_reenable_device() on the hotplug bridge,
and clearing its Interrupt Disable bit makes its hotplug event-reporting
MSI stop working.

Fixes: 1f42db786b14 PCI: Enable INTx if BIOS left them disabled
Link: https://bugzilla.kernel.org/show_bug.cgi?id=71691
Signed-off-by: Bjorn Helgaas <bhelgaas@...gle.com>
---
 drivers/pci/pci.c |    3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 8dc3e701ec57..79fc89c6c3f3 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -1192,6 +1192,9 @@ static int do_pci_enable_device(struct pci_dev *dev, int bars)
 		return err;
 	pci_fixup_device(pci_fixup_enable, dev);
 
+	if (dev->msi_enabled || dev->msix_enabled)
+		return 0;
+
 	pci_read_config_byte(dev, PCI_INTERRUPT_PIN, &pin);
 	if (pin) {
 		pci_read_config_word(dev, PCI_COMMAND, &cmd);
--
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/