linux-kernel - Re: Aw: Re: Re: [PATCH] pci: mediatek: fix warning in msi.h

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]

Date:   Mon, 02 Nov 2020 13:58:12 +0000
From:   Marc Zyngier <maz@...nel.org>
To:     Frank Wunderlich <frank-w@...lic-files.de>
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ryder Lee <ryder.lee@...iatek.com>,
        linux-mediatek@...ts.infradead.org,
        Frank Wunderlich <linux@...web.de>,
        linux-kernel@...r.kernel.org,
        Matthias Brugger <matthias.bgg@...il.com>,
        linux-pci@...r.kernel.org, Bjorn Helgaas <bhelgaas@...gle.com>
Subject: Re: Aw: Re:  Re: [PATCH] pci: mediatek: fix warning in msi.h

On 2020-11-02 11:56, Frank Wunderlich wrote:
> looks good on bananapi-r2, no warning, pcie-card and hdd recognized

Thanks for giving it a shot. Still needs a bit of tweaking, as I expect
it to break configurations that select CONFIG_PCI_MSI_ARCH_FALLBACKS
(we have to assume that MSIs can be handled until we hit the 
arch-specific
stuff).

There is also a small nit in the way we allow userspace to mess with
this flag via sysfs, and similar restrictions should probably apply.

Updated patch below.

         M.

diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
index d15c881e2e7e..5bb1306162c7 100644
--- a/drivers/pci/pci-sysfs.c
+++ b/drivers/pci/pci-sysfs.c
@@ -387,10 +387,20 @@ static ssize_t msi_bus_store(struct device *dev, 
struct device_attribute *attr,
  		return count;
  	}

-	if (val)
+	if (val) {
+		/*
+		 * If there is no possibility for this bus to deal with
+		 * MSIs, then allowing them to be requested would lead to
+		 * the kernel complaining loudly. In this situation, don't
+		 * let userspace mess things up.
+		 */
+		if (!pci_bus_is_msi_capable(subordinate))
+			return -EINVAL;
+
  		subordinate->bus_flags &= ~PCI_BUS_FLAGS_NO_MSI;
-	else
+	} else {
  		subordinate->bus_flags |= PCI_BUS_FLAGS_NO_MSI;
+	}

  	dev_info(&subordinate->dev, "MSI/MSI-X %s for future drivers of 
devices on this bus\n",
  		 val ? "allowed" : "disallowed");
diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 4289030b0fff..28861cc6435a 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -871,6 +871,8 @@ static void pci_set_bus_msi_domain(struct pci_bus 
*bus)
  		d = pci_host_bridge_msi_domain(b);

  	dev_set_msi_domain(&bus->dev, d);
+	if (!pci_bus_is_msi_capable(bus))
+		bus->bus_flags |= PCI_BUS_FLAGS_NO_MSI;
  }

  static int pci_register_host_bridge(struct pci_host_bridge *bridge)
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 22207a79762c..6aadb863dff4 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -2333,6 +2333,12 @@ pci_host_bridge_acpi_msi_domain(struct pci_bus 
*bus) { return NULL; }
  static inline bool pci_pr3_present(struct pci_dev *pdev) { return 
false; }
  #endif

+static inline bool pci_bus_is_msi_capable(struct pci_bus *bus)
+{
+	return (IS_ENABLED(CONFIG_PCI_MSI_ARCH_FALLBACKS) ||
+		dev_get_msi_domain(&bus->dev));
+}
+
  #ifdef CONFIG_EEH
  static inline struct eeh_dev *pci_dev_to_eeh_dev(struct pci_dev *pdev)
  {

-- 
Jazz is not dead. It just smells funny...