Re: Info: mapping multiple BARs. Your kernel is fine.

[Bjorn Helgaas <bhelgaas@google.com>]

On Wed, Apr 16, 2014 at 06:31:22PM -0400, Dave Jones wrote:
> On Wed, Apr 16, 2014 at 02:31:38PM -0600, Bjorn Helgaas wrote:
>  > On Wed, Apr 16, 2014 at 09:04:04PM +0200, Borislav Petkov wrote:
>  > > On Thu, Mar 20, 2014 at 02:48:30PM -0600, Bjorn Helgaas wrote:
>  > > > Right.  Even if we had this long-term solution, we'd still have
>  > > > Stephane's current problem, because the PNP0C02 _CRS is still wrong.
>  > > > 
>  > > > We do have a drivers/pnp/quirks.c where we could conceivably adjust
>  > > > the PNP resource if we found the matching PCI device and MCHBAR.  That
>  > > > should solve Stephane's problem even with the current
>  > > > drivers/pnp/system.c.
>  > > 
>  > > Guys, this still triggers in -rc1. Do we have a fix or something
>  > > testable at least?
>  > 
>  > Hi Boris,
>  > 
>  > Can you try the patch below?
> 
> I'm seeing the exact same message on my thinkpad t430s.
> When I try your patch, modesetting no longer works. When it tries
> to change to the framebuffer I get a black screen and lockup.
> If I boot with nomodeset it locks up when it gets to X.
> It all scrolls by too fast to read, but it looks like there's still
> a backtrace present.

Ouch, sorry about that.  I do see a bug in my patch (fixed below), but I
don't see how that could cause what you're seeing.  Maybe I could figure
out something from this info (this can be from a kernel without my patch):

    - dmesg log
    - output of "find /sys/devices/pnp0 -name id -o -name resources | xargs grep ."
    - output of "sudo lspci -s00:00.0 -xxx"



PNP: Work around Haswell BIOS defect in MCH area reporting

From: Bjorn Helgaas <bhelgaas@google.com>

Work around a Haswell BIOS defect that causes part of the MCH area to be
unreported.

MCHBAR is not an architected PCI BAR, so MCH space is usually reported as a
PNP0C02 resource.  The MCH space was 16KB prior to Haswell, but it is 32KB
in Haswell.  Some Haswell BIOSes still report a PNP0C02 resource that is
only 16KB, which means the rest of the MCH space is consumed but
unreported.

This can cause resource map sanity check warnings or (theoretically) a
device conflict if we assigned the unreported space to another device.

The Intel perf event uncore driver tripped over this when it claimed the
MCH region:

  resource map sanity check conflict: 0xfed10000 0xfed15fff 0xfed10000 0xfed13fff pnp 00:01
  Info: mapping multiple BARs. Your kernel is fine.

To prevent this, if we find a PNP0C02 resource that covers part of the MCH
space, extend it to cover the entire space.

Link: http://lkml.kernel.org/r/20140224162400.GE16457@pd.tnic
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
---
 drivers/pnp/quirks.c |   55 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)

diff --git a/drivers/pnp/quirks.c b/drivers/pnp/quirks.c
index 258fef272ea7..8402088d4145 100644
--- a/drivers/pnp/quirks.c
+++ b/drivers/pnp/quirks.c
@@ -334,6 +334,60 @@ static void quirk_amd_mmconfig_area(struct pnp_dev *dev)
 }
 #endif
 
+static void quirk_intel_haswell_mch(struct pnp_dev *dev)
+{
+	struct pci_dev *host;
+	u32 addr_lo, addr_hi;
+	struct pci_bus_region region;
+	struct resource mch;
+	struct pnp_resource *pnp_res;
+	struct resource *res;
+
+	host = pci_get_device(PCI_VENDOR_ID_INTEL, 0x0c00, NULL);
+	if (!host)
+		return;
+
+	/*
+	 * MCHBAR is not an architected PCI BAR, so MCH space is usually
+	 * reported as a PNP0C02 resource.  The MCH space was 16KB prior to
+	 * Haswell, but it is 32KB in Haswell.  Some Haswell BIOSes still
+	 * report a PNP0C02 resource that is only 16KB, which means the
+	 * rest of the MCH space is consumed but unreported.
+	 */
+
+	/*
+	 * Read MCHBAR for Host Member Mapped Register Range Base
+	 * https://www-ssl.intel.com/content/www/us/en/processors/core/4th-gen-core-family-desktop-vol-2-datasheet
+	 * Sec 3.1.12.
+	 */
+	pci_read_config_dword(host, 0x48, &addr_lo);
+	region.start = addr_lo & ~0x7fff;
+	pci_read_config_dword(host, 0x4c, &addr_hi);
+	region.start |= (dma_addr_t) addr_hi << 32;
+	region.end = region.start + 32*1024 - 1 ;
+
+	memset(&mch, 0, sizeof(mch));
+	mch.flags = IORESOURCE_MEM;
+	pcibios_bus_to_resource(host->bus, &mch, &region);
+
+	list_for_each_entry(pnp_res, &dev->resources, list) {
+		res = &pnp_res->res;
+		if (res->end < mch.start || res->start > mch.end)
+			continue;	/* no overlap */
+		if (res->start == mch.start && res->end == mch.end)
+			continue;	/* exact match */
+
+		dev_info(&dev->dev, FW_BUG
+			 "%pR covers only part of Intel Haswell MCH; extending to %pR\n",
+			 res, &mch);
+		res->start = mch.start;
+		res->end = mch.end;
+		break;
+	}
+
+	pci_dev_put(host);
+}
+
 /*
  *  PnP Quirks
  *  Cards or devices that need some tweaking due to incomplete resource info
@@ -364,6 +418,7 @@ static struct pnp_fixup pnp_fixups[] = {
 #ifdef CONFIG_AMD_NB
 	{"PNP0c01", quirk_amd_mmconfig_area},
 #endif
+	{"PNP0c02", quirk_intel_haswell_mch},
 	{""}
 };