If the port supports Link speeds greater than 5.0 GT/s, we must wait
for 100 ms after Link training completes before sending configuration
request.
Acked-by: Yinghai Lu <yinghai@kernel.org>
Tested-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We need to wait for 1000 ms after Data Link Layer Link Active (DLLLA)
bit reads 1b before sending configuration request. Currently pciehp
does this wait after checking Link Training (LT) bit. But we need it
before checking LT bit because LT is still set even after DLLLA bit is
set on some platforms.
Acked-by: Yinghai Lu <yinghai@kernel.org>
Tested-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
During hot plug, board_added will call pciehp_power_on_slot().
But link speed is updated in pciehp_power_on_slot().
We should not update link speed there, because that is too early.
So move the link speed update to pciehp_check_link_status() after making
sure the link has been trained.
-v2: fix compile warning that Kenji found.
Signed-off-by: Yinghai Lu <yinghai.lu@oracle.com>
Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Tested-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This avoids the PCI_PRI question in 'make config' when PCI
is not selected.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
If the kernel has requested control of the PCIe native hotplug
feature for a given root complex, the acpiphp driver should not try
to handle that root complex and it should leave it to pciehp.
Failing to do so causes problems to happen if acpiphp is loaded
before pciehp on such systems.
To address this issue make find_root_bridges() ignore PCIe root
complexes with PCIe native hotplug enabled and make add_bridge()
return error code if PCIe native hotplug is enabled for the given
root port. This causes acpiphp to refuse to load if PCIe native
hotplug is enabled for all complexes and to refuse binding to
the root complexes with PCIe native hotplug is enabled.
Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'next-rebase' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci:
PCI: Clean-up MPS debug output
pci: Clamp pcie_set_readrq() when using "performance" settings
PCI: enable MPS "performance" setting to properly handle bridge MPS
PCI: Workaround for Intel MPS errata
PCI: Add support for PASID capability
PCI: Add implementation for PRI capability
PCI: Export ATS functions to modules
PCI: Move ATS implementation into own file
PCI / PM: Remove unnecessary error variable from acpi_dev_run_wake()
PCI hotplug: acpiphp: Prevent deadlock on PCI-to-PCI bridge remove
PCI / PM: Extend PME polling to all PCI devices
PCI quirk: mmc: Always check for lower base frequency quirk for Ricoh 1180:e823
PCI: Make pci_setup_bridge() non-static for use by arch code
x86: constify PCI raw ops structures
PCI: Add quirk for known incorrect MPSS
PCI: Add Solarflare vendor ID and SFC4000 device IDs
Clean-up MPS debug output to make it a single line and aligned, thus
making it more readable for a large number of buses and devices in a
single system.
Suggested by Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
When configuring the PCIe settings for "performance", we allow parents
to have a larger Max Payload Size than children and rely on children
Max Read Request Size to not be larger than their own MPS to avoid
having the host bridge generate responses they can't cope with.
However, various drivers in Linux call pci_set_readrq() with arbitrary
values, assuming this to be a simple performance tweak. This breaks
under our "performance" configuration.
Fix that by making sure the value programmed by pcie_set_readrq() is
never larger than the configured MPS for that device.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Rework the "performance" MPS option to configure the device MPS with the
smaller of the device MPSS or the bridge MPS (which is assumed to be
properly configured at this point to the largest allowable MPS based on
its parent bus).
Also, rework the MRRS setting to report an inability to set the MRRS to
a valid setting.
Signed-off-by: Jon Mason <mason@myri.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'core-iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86, ioapic: Consolidate the explicit EOI code
x86, ioapic: Restore the mask bit correctly in eoi_ioapic_irq()
x86, kdump, ioapic: Reset remote-IRR in clear_IO_APIC
iommu: Rename the DMAR and INTR_REMAP config options
x86, ioapic: Define irq_remap_modify_chip_defaults()
x86, msi, intr-remap: Use the ioapic set affinity routine
iommu: Cleanup ifdefs in detect_intel_iommu()
iommu: No need to set dmar_disabled in check_zero_address()
iommu: Move IOMMU specific code to intel-iommu.c
intr_remap: Call dmar_dev_scope_init() explicitly
x86, x2apic: Enable the bios request for x2apic optout
* 'stable/drivers-3.2' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xenbus: don't rely on xen_initial_domain to detect local xenstore
xenbus: Fix loopback event channel assuming domain 0
xen/pv-on-hvm:kexec: Fix implicit declaration of function 'xen_hvm_domain'
xen/pv-on-hvm kexec: add xs_reset_watches to shutdown watches from old kernel
xen/pv-on-hvm kexec: update xs_wire.h:xsd_sockmsg_type from xen-unstable
xen/pv-on-hvm kexec+kdump: reset PV devices in kexec or crash kernel
xen/pv-on-hvm kexec: rebind virqs to existing eventchannel ports
xen/pv-on-hvm kexec: prevent crash in xenwatch_thread() when stale watch events arrive
* 'stable/drivers.bugfixes-3.2' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/pciback: Check if the device is found instead of blindly assuming so.
xen/pciback: Do not dereference psdev during printk when it is NULL.
xen: remove XEN_PLATFORM_PCI config option
xen: XEN_PVHVM depends on PCI
xen/pciback: double lock typo
xen/pciback: use mutex rather than spinlock in vpci backend
xen/pciback: Use mutexes when working with Xenbus state transitions.
xen/pciback: miscellaneous adjustments
xen/pciback: use mutex rather than spinlock in passthrough backend
xen/pciback: use resource_size()
* 'stable/pci.fixes-3.2' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/pci: support multi-segment systems
xen-swiotlb: When doing coherent alloc/dealloc check before swizzling the MFNs.
xen/pci: make bus notifier handler return sane values
xen-swiotlb: fix printk and panic args
xen-swiotlb: Fix wrong panic.
xen-swiotlb: Retry up three times to allocate Xen-SWIOTLB
xen-pcifront: Update warning comment to use 'e820_host' option.
Devices supporting Process Address Space Identifiers
(PASIDs) can use an IOMMU to access multiple IO address
spaces at the same time. A PCIe device indicates support for
this feature by implementing the PASID capability. This
patch adds support for the capability to the Linux kernel.
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Implement the necessary functions to handle PRI capabilities
on PCIe devices. With PRI devices behind an IOMMU can signal
page fault conditions to software and recover from such
faults.
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch makes the ATS functions usable for modules.
They will be used by a module implementing some advanced
AMD IOMMU features.
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
ATS does not depend on IOV support, so move the code into
its own file. This file will also include support for the
PRI and PASID capabilities later.
Also give ATS its own Kconfig variable to allow selecting it
without IOV support.
Reviewed-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The result returned by acpi_dev_run_wake() is always either -EINVAL
or -ENODEV, while obviously it should return 0 on success. The
problem is that the leftover error variable, that's not really used
in the function, is initialized with -ENODEV and then returned
without modification.
To fix this issue remove the error variable from acpi_dev_run_wake()
and make the function return 0 on success as appropriate.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
I originally submitted a patch to workaround this by pushing all Ejection
Requests and Device Checks onto the kacpi_hotplug queue.
http://marc.info/?l=linux-acpi&m=131678270930105&w=2
The patch is still insufficient in that Bus Checks also need to be added.
Rather than add all events, including non-PCI-hotplug events, to the
hotplug queue, mjg suggested that a better approach would be to modify
the acpiphp driver so only acpiphp events would be added to the
kacpi_hotplug queue.
It's a longer patch, but at least we maintain the benefit of having separate
queues in ACPI. This, of course, is still only a workaround the problem.
As Bjorn and mjg pointed out, we have to refactor a lot of this code to do
the right thing but at this point it is a better to have this code working.
The acpi core places all events on the kacpi_notify queue. When the acpiphp
driver is loaded and a PCI card with a PCI-to-PCI bridge is removed the
following call sequence occurs:
cleanup_p2p_bridge()
-> cleanup_bridge()
-> acpi_remove_notify_handler()
-> acpi_os_wait_events_complete()
-> flush_workqueue(kacpi_notify_wq)
which is the queue we are currently executing on and the process will hang.
Move all hotplug acpiphp events onto the kacpi_hotplug workqueue. In
handle_hotplug_event_bridge() and handle_hotplug_event_func() we can simply
push the rest of the work onto the kacpi_hotplug queue and then avoid the
deadlock.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Cc: mjg@redhat.com
Cc: bhelgaas@google.com
Cc: linux-acpi@vger.kernel.org
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The land of PCI power management is a land of sorrow and ugliness,
especially in the area of signaling events by devices. There are
devices that set their PME Status bits, but don't really bother
to send a PME message or assert PME#. There are hardware vendors
who don't connect PME# lines to the system core logic (they know
who they are). There are PCI Express Root Ports that don't bother
to trigger interrupts when they receive PME messages from the devices
below. There are ACPI BIOSes that forget to provide _PRW methods for
devices capable of signaling wakeup. Finally, there are BIOSes that
do provide _PRW methods for such devices, but then don't bother to
call Notify() for those devices from the corresponding _Lxx/_Exx
GPE-handling methods. In all of these cases the kernel doesn't have
a chance to receive a proper notification that it should wake up a
device, so devices stay in low-power states forever. Worse yet, in
some cases they continuously send PME Messages that are silently
ignored, because the kernel simply doesn't know that it should clear
the device's PME Status bit.
This problem was first observed for "parallel" (non-Express) PCI
devices on add-on cards and Matthew Garrett addressed it by adding
code that polls PME Status bits of such devices, if they are enabled
to signal PME, to the kernel. Recently, however, it has turned out
that PCI Express devices are also affected by this issue and that it
is not limited to add-on devices, so it seems necessary to extend
the PME polling to all PCI devices, including PCI Express and planar
ones. Still, it would be wasteful to poll the PME Status bits of
devices that are known to receive proper PME notifications, so make
the kernel (1) poll the PME Status bits of all PCI and PCIe devices
enabled to signal PME and (2) disable the PME Status polling for
devices for which correct PME notifications are received.
Tested-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Commit 15bed0f2f added a quirk for the e823 Ricoh card reader to lower the
base frequency. However, the quirk first checks to see if the proprietary
MMC controller is disabled, and returns if so. On some devices, such as the
Lenovo X220, the MMC controller is already disabled by firmware it seems,
but the frequency change is still needed so sdhci-pci can talk to the cards.
Since the MMC controller is disabled, the frequency fixup was never being run
on these machines.
This moves the e823 check above the MMC controller check so that it always
gets run.
This fixes https://bugzilla.redhat.com/show_bug.cgi?id=722509
Signed-off-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The "powernv" platform of the powerpc architecture needs to assign PCI
resources using a specific algorithm to fit some HW constraints of
the IBM "IODA" architecture (related to the ability to create error
handling domains that encompass specific segments of MMIO space).
For doing so, it wants to call pci_setup_bridge() from architecture
specific resource management in order to configure bridges after all
resources have been assigned. So make it non-static.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Using legacy interrupts and TLPs > 256 bytes on the SFC4000 (all
revisions) may cause interrupt messages to be replayed. In some
systems this results in a non-recoverable MCE. Early boards using the
SFC4000 set the maximum payload size supported (MPSS) to 1024 bytes
and we should override that.
There are probably other devices with similar issues, so give this
quirk a generic name.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add the ability to disable PCI-E MPS turning and using the BIOS
configured MPS defaults. Due to the number of issues recently
discovered on some x86 chipsets, make this the default behavior.
Also, add the option for peer to peer DMA MPS configuration. Peer to
peer DMA is outside the scope of this patch, but MPS configuration could
prevent it from working by having the MPS on one root port different
than the MPS on another. To work around this, simply make the system
wide MPS the smallest possible value (128B).
Signed-off-by: Jon Mason <mason@myri.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In pcie_find_smpss(), we have the following statement:
if (dev->is_hotplug_bridge && (!list_is_singular(&dev->bus->devices) ||
dev->bus->self->pcie_type != PCI_EXP_TYPE_ROOT_PORT))
The problem is that at least on my machine, this gets called for the
root complex (virtual P2P bridge), and dev->bus->self is NULL since
the parent bus for this is not itself anchor to a PCI device.
This adds the necessary NULL check.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Jon Mason <mason@myri.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Modifying the Maximum Read Request Size to 0 (value of 128Bytes) has
massive negative ramifications on some devices. Without knowing which
devices have this issue, do not modify from the default value when
walking the PCI-E bus in pcie_bus_safe mode. Also, make pcie_bus_safe
the default procedure.
Tested-by: Sven Schnelle <svens@stackframe.org>
Tested-by: Simon Kirby <sim@hostway.ca>
Tested-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Reported-and-tested-by: Eric Dumazet <eric.dumazet@gmail.com>
Reported-and-tested-by: Niels Ole Salscheider <niels_ole@salscheider-online.de>
References: https://bugzilla.kernel.org/show_bug.cgi?id=42162
Signed-off-by: Jon Mason <mason@myri.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit b03e7495a8 ("PCI: Set PCI-E Max Payload Size on fabric")
introduced a potential NULL pointer dereference in calls to
pcie_bus_configure_settings due to attempts to access pci_bus self
variables when the self pointer is NULL.
To correct this, verify that the self pointer in pci_bus is non-NULL
before dereferencing it.
Reported-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Shyam Iyer <shyam_iyer@dell.com>
Signed-off-by: Jon Mason <mason@myri.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With Xen changeset 23428 "libxl: Add 'e820_host' option to config file"
the E820 as seen from the host can now be passed into the guest.
This means that a PV guest can now:
- Use the correct PCI I/O gap. Before these patches, Linux guest would
boot up and would tell:
[ 0.000000] Allocating PCI resources starting at 40000000 (gap: 40000000:c0000000)
while in actuality the PCI I/O gap should have been:
[ 0.000000] Allocating PCI resources starting at b0000000 (gap: b0000000:4c000000)
- The PV domain with PCI devices was limited to 3GB. It now can be booted
with 4GB, 8GB, or whatever number you want. The PCI devices will now _not_ conflict
with System RAM. Meaning the drivers can load.
CC: Jesse Barnes <jbarnes@virtuousgeek.org>
CC: linux-pci@vger.kernel.org
CC: stable@kernel.org
[v2: Made the string less broken up. Suggested by Joe Perches]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Fix new kernel-doc warning in pci.c:
Warning(drivers/pci/pci.c:3259): No description found for parameter 'mps'
Warning(drivers/pci/pci.c:3259): Excess function parameter 'rq' description in 'pcie_set_mps'
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
PCI: OF: Don't crash when bridge parent is NULL.
PCI: export pcie_bus_configure_settings symbol
PCI: code and comments cleanup
PCI: make cardbus-bridge resources optional
PCI: make SRIOV resources optional
PCI : ability to relocate assigned pci-resources
PCI: honor child buses add_size in hot plug configuration
PCI: Set PCI-E Max Payload Size on fabric
In pcibios_get_phb_of_node(), we will crash while booting if
bus->bridge->parent is NULL.
Check for this case and avoid dereferencing the NULL pointer.
Signed-off-by: David Daney <david.daney@cavium.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (28 commits)
ACPI: delete stale reference in kernel-parameters.txt
ACPI: add missing _OSI strings
ACPI: remove NID_INVAL
thermal: make THERMAL_HWMON implementation fully internal
thermal: split hwmon lookup to a separate function
thermal: hide CONFIG_THERMAL_HWMON
ACPI print OSI(Linux) warning only once
ACPI: DMI workaround for Asus A8N-SLI Premium and Asus A8N-SLI DELUX
ACPI / Battery: propagate sysfs error in acpi_battery_add()
ACPI / Battery: avoid acpi_battery_add() use-after-free
ACPI: introduce "acpi_rsdp=" parameter for kdump
ACPI: constify ops structs
ACPI: fix CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS
ACPI: fix 80 char overflow
ACPI / Battery: Resolve the race condition in the sysfs_remove_battery()
ACPI / Battery: Add the check before refresh sysfs in the battery_notify()
ACPI / Battery: Add the hibernation process in the battery_notify()
ACPI / Battery: Rename acpi_battery_quirks2 with acpi_battery_quirks
ACPI / Battery: Change 16-bit signed negative battery current into correct value
ACPI / Battery: Add the power unit macro
...
pcie_bus_configure_settings needs to be exported if the PCI hotplug
driver is being compiled as a module.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
a) adjust_resource_sorted() is now called reassign_resource_sorted()
b) nice-to-have is now called optional
c) add_list is now called realloc_list.
Signed-off-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Allocate resources to cardbus bridge only after all other genuine
resources requests are satisfied. Dont retry if resource allocation
for cardbus-bridges fail.
Signed-off-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
From: Yinghai Lu <yinghai@kernel.org>
Allocate resources to SRIOV BARs only after all other required
resource-requests are satisfied. Dont retry if resource allocation for SRIOV
BARs fail.
Signed-off-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Currently pci-bridges are allocated enough resources to satisfy their immediate
requirements. Any additional resource-requests fail if additional free space,
contiguous to the one already allocated, is not available. This behavior is not
reasonable since sufficient contiguous resources, that can satisfy the request,
are available at a different location.
This patch provides the ability to expand and relocate a allocated resource.
v2: Changelog: Fixed size calculation in pci_reassign_resource()
v3: Changelog : Split this patch. The resource.c changes are already
upstream. All the pci driver changes are in here.
Signed-off-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
git commit c8adf9a3e8
"PCI: pre-allocate additional resources to devices only after
successful allocation of essential resources."
fails to take into consideration the optional-resources needed by children
devices while calculating the optional-resource needed by the bridge.
This can be a problem on some setup. For example, if a hotplug bridge has 8
children hotplug bridges, the bridge should have enough resources to accomodate
the hotplug requirements for each of its children hotplug bridges. Currently
this is not the case.
This patch fixes the problem.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Reviewed-by: Ram Pai <linuxram@us.ibm.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
On a given PCI-E fabric, each device, bridge, and root port can have a
different PCI-E maximum payload size. There is a sizable performance
boost for having the largest possible maximum payload size on each PCI-E
device. However, if improperly configured, fatal bus errors can occur.
Thus, it is important to ensure that PCI-E payloads sends by a device
are never larger than the MPS setting of all devices on the way to the
destination.
This can be achieved two ways:
- A conservative approach is to use the smallest common denominator of
the entire tree below a root complex for every device on that fabric.
This means for example that having a 128 bytes MPS USB controller on one
leg of a switch will dramatically reduce performances of a video card or
10GE adapter on another leg of that same switch.
It also means that any hierarchy supporting hotplug slots (including
expresscard or thunderbolt I suppose, dbl check that) will have to be
entirely clamped to 128 bytes since we cannot predict what will be
plugged into those slots, and we cannot change the MPS on a "live"
system.
- A more optimal way is possible, if it falls within a couple of
constraints:
* The top-level host bridge will never generate packets larger than the
smallest TLP (or if it can be controlled independently from its MPS at
least)
* The device will never generate packets larger than MPS (which can be
configured via MRRS)
* No support of direct PCI-E <-> PCI-E transfers between devices without
some additional code to specifically deal with that case
Then we can use an approach that basically ignores downstream requests
and focuses exclusively on upstream requests. In that case, all we need
to care about is that a device MPS is no larger than its parent MPS,
which allows us to keep all switches/bridges to the max MPS supported by
their parent and eventually the PHB.
In this case, your USB controller would no longer "starve" your 10GE
Ethernet and your hotplug slots won't affect your global MPS.
Additionally, the hotplugged devices themselves can be configured to a
larger MPS up to the value configured in the hotplug bridge.
To choose between the two available options, two PCI kernel boot args
have been added to the PCI calls. "pcie_bus_safe" will provide the
former behavior, while "pcie_bus_perf" will perform the latter behavior.
By default, the latter behavior is used.
NOTE: due to the location of the enablement, each arch will need to add
calls to this function. This patch only enables x86.
This patch includes a number of changes recommended by Benjamin
Herrenschmidt.
Tested-by: Jordan_Hargrave@dell.com
Signed-off-by: Jon Mason <mason@myri.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
* 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
PCI: remove printks about disabled bridge windows
PCI: fold pci_calc_resource_flags() into decode_bar()
PCI: treat mem BAR type "11" (reserved) as 32-bit, not 64-bit, BAR
PCI: correct pcie_set_readrq write size
PCI: pciehp: change wait time for valid configuration access
x86/PCI: Preserve existing pci=bfsort whitelist for Dell systems
PCI: ARI is a PCIe v2 feature
x86/PCI: quirks: Use pci_dev->revision
PCI: Make the struct pci_dev * argument of pci_fixup_irqs const.
PCI hotplug: cpqphp: use pci_dev->vendor
PCI hotplug: cpqphp: use pci_dev->subsystem_{vendor|device}
x86/PCI: config space accessor functions should not ignore the segment argument
PCI: Assign values to 'pci_obff_signal_type' enumeration constants
x86/PCI: reduce severity of host bridge window conflict warnings
PCI: enumerate the PCI device only removed out PCI hieratchy of OS when re-scanning PCI
PCI: PCIe AER: add aer_recover_queue
x86/PCI: select direct access mode for mmconfig option
PCI hotplug: Rename is_ejectable which also exists in dock.c
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
merge fchmod() and fchmodat() guts, kill ancient broken kludge
xfs: fix misspelled S_IS...()
xfs: get rid of open-coded S_ISREG(), etc.
vfs: document locking requirements for d_move, __d_move and d_materialise_unique
omfs: fix (mode & S_IFDIR) abuse
btrfs: S_ISREG(mode) is not mode & S_IFREG...
ima: fmode_t misspelled as mode_t...
pci-label.c: size_t misspelled as mode_t
jffs2: S_ISLNK(mode & S_IFMT) is pointless
snd_msnd ->mode is fmode_t, not mode_t
v9fs_iop_get_acl: get rid of unused variable
vfs: dont chain pipe/anon/socket on superblock s_inodes list
Documentation: Exporting: update description of d_splice_alias
fs: add missing unlock in default_llseek()
This allows us to move duplicated code in <asm/atomic.h>
(atomic_inc_not_zero() for now) to <linux/atomic.h>
Signed-off-by: Arun Sharma <asharma@fb.com>
Reviewed-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial: (43 commits)
fs: Merge split strings
treewide: fix potentially dangerous trailing ';' in #defined values/expressions
uwb: Fix misspelling of neighbourhood in comment
net, netfilter: Remove redundant goto in ebt_ulog_packet
trivial: don't touch files that are removed in the staging tree
lib/vsprintf: replace link to Draft by final RFC number
doc: Kconfig: `to be' -> `be'
doc: Kconfig: Typo: square -> squared
doc: Konfig: Documentation/power/{pm => apm-acpi}.txt
drivers/net: static should be at beginning of declaration
drivers/media: static should be at beginning of declaration
drivers/i2c: static should be at beginning of declaration
XTENSA: static should be at beginning of declaration
SH: static should be at beginning of declaration
MIPS: static should be at beginning of declaration
ARM: static should be at beginning of declaration
rcu: treewide: Do not use rcu_read_lock_held when calling rcu_dereference_check
Update my e-mail address
PCIe ASPM: forcedly -> forcibly
gma500: push through device driver tree
...
Fix up trivial conflicts:
- arch/arm/mach-ep93xx/dma-m2p.c (deleted)
- drivers/gpio/gpio-ep93xx.c (renamed and context nearby)
- drivers/net/r8169.c (just context changes)
* 'core-iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
iommu/core: Fix build with INTR_REMAP=y && CONFIG_DMAR=n
iommu/amd: Don't use MSI address range for DMA addresses
iommu/amd: Move missing parts to drivers/iommu
iommu: Move iommu Kconfig entries to submenu
x86/ia64: intel-iommu: move to drivers/iommu/
x86: amd_iommu: move to drivers/iommu/
msm: iommu: move to drivers/iommu/
drivers: iommu: move to a dedicated folder
x86/amd-iommu: Store device alias as dev_data pointer
x86/amd-iommu: Search for existind dev_data before allocting a new one
x86/amd-iommu: Allow dev_data->alias to be NULL
x86/amd-iommu: Use only dev_data in low-level domain attach/detach functions
x86/amd-iommu: Use only dev_data for dte and iotlb flushing routines
x86/amd-iommu: Store ATS state in dev_data
x86/amd-iommu: Store devid in dev_data
x86/amd-iommu: Introduce global dev_data_list
x86/amd-iommu: Remove redundant device_flush_dte() calls
iommu-api: Add missing header file
Fix up trivial conflicts (independent additions close to each other) in
drivers/Makefile and include/linux/pci.h
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6: (51 commits)
PM: Improve error code of pm_notifier_call_chain()
PM: Add "RTC" to PM trace time stamps to avoid confusion
PM / Suspend: Export suspend_set_ops, suspend_valid_only_mem
PM / Suspend: Add .suspend_again() callback to suspend_ops
PM / OPP: Introduce function to free cpufreq table
ARM / shmobile: Return -EBUSY from A4LC power off if A3RV is active
PM / Domains: Take .power_off() error code into account
ARM / shmobile: Use genpd_queue_power_off_work()
ARM / shmobile: Use pm_genpd_poweroff_unused()
PM / Domains: Introduce function to power off all unused PM domains
OMAP: PM: disable idle on suspend for GPIO and UART
OMAP: PM: omap_device: add API to disable idle on suspend
OMAP: PM: omap_device: add system PM methods for PM domain handling
OMAP: PM: omap_device: conditionally use PM domain runtime helpers
PM / Runtime: Add new helper function: pm_runtime_status_suspended()
PM / Domains: Queue up power off work only if it is not pending
PM / Domains: Improve handling of wakeup devices during system suspend
PM / Domains: Do not restore all devices on power off error
PM / Domains: Allow callbacks to execute all runtime PM helpers
PM / Domains: Do not execute device callbacks under locks
...
* 'of-pci' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
pci/of: Consolidate pci_bus_to_OF_node()
pci/of: Consolidate pci_device_to_OF_node()
x86/devicetree: Use generic PCI <-> OF matching
microblaze/pci: Move the remains of pci_32.c to pci-common.c
microblaze/pci: Remove powermac originated cruft
pci/of: Match PCI devices to OF nodes dynamically
I don't think there's enough value in the fact of a bridge window
being disabled to justify cluttering the dmesg log with it.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
decode_bar() and pci_calc_resource_flags() both looked at the PCI BAR
type information, and it's simpler to just do it all in one place.
decode_bar() sets IORESOURCE_IO, IORESOURCE_MEM, and IORESOURCE_MEM_64
as appropriate, so res->flags contains all the information pci_bar_type
does, so we don't need to test the pci_bar_type return value.
decode_bar() used to return pci_bar_type, which we no longer need. We
can simplify it a bit by returning the struct resource flags rather than
updating them internally.
In pci_update_resource(), there's no need to decode the BAR type bits
again; we can just test for IORESOURCE_MEM_64 directly.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This fixes a minor regression where broken PCI devices that use the
reserved "11" memory BAR type worked before e354597cce but not after.
The low four bits of a memory BAR are "PTT0" where P=1 for prefetchable
BARs, and TT is as follows:
00 32-bit BAR, anywhere in lower 4GB
01 anywhere below 1MB (reserved as of PCI 2.2)
10 64-bit BAR
11 reserved
Prior to e354597cce, we treated "0100" as a 64-bit BAR and all others,
including prefetchable 64-bit BARs ("1100") as 32-bit BARs. The e354597cce
fix, which appeared in 2.6.28, treats "x1x0" as 64-bit BARs, so the
reserved "x110" types are treated as 64-bit instead of 32-bit.
This patch returns to treating the reserved "11" type as a 32-bit BAR and
adds a warning if we see it.
It also logs a note if we see a 1M BAR. This is not a warning, because
such hardware conforms to pre-PCI 2.2 spec, but I think it's worth noting
because Linux ignores the 1M restriction if it ever has to assign the BAR.
CC: Peter Chubb <peterc@gelato.unsw.edu.au>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=35952
Reported-by: Jan Zwiegers <jan@radicalsystems.co.za>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Kenji Kaneshige