PCI Express ASPM defines a protocol for PCI Express components in the D0
state to reduce Link power by placing their Links into a low power state
and instructing the other end of the Link to do likewise. This
capability allows hardware-autonomous, dynamic Link power reduction
beyond what is achievable by software-only controlled power management.
However, The device should be configured by software appropriately.
Enabling ASPM will save power, but will introduce device latency.
This patch adds ASPM support in Linux. It introduces a global policy for
ASPM, a sysfs file /sys/module/pcie_aspm/parameters/policy can control
it. The interface can be used as a boot option too. Currently we have
below setting:
-default, BIOS default setting
-powersave, highest power saving mode, enable all available ASPM
state
and clock power management
-performance, highest performance, disable ASPM and clock power
management
By default, the 'default' policy is used currently.
In my test, power difference between powersave mode and performance mode
is about 1.3w in a system with 3 PCIE links.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If the fakephp driver is used to emulate removal of a PCI device by
writing text string "0" to the "power" sysfs attribute file, this causes
its parent directory and its contents (including the "power" file) to be
deleted before the write operation returns. Unfortunately, it ends up
in a deadlock waiting for itself to complete.
The deadlock is as follows: sysfs_write_file calls flush_write_buffer
which calls sysfs_get_active_two before calling power_write_file in
pci_hotplug_core.c via the sysfs store operation. The power_write_file
function calls disable_slot in fakephp.c via the slot operation. The
disable_slot function calls remove_slot which calls pci_hp_deregister
(back in pci_hotplug_core.c) which calls fs_remove_slot which calls
sysfs_remove_file to remove the "power" file. The sysfs_remove_file
function calls sysfs_hash_and_remove which calls sysfs_addrm_finish
which calls sysfs_deactivate. The sysfs_deactivate function sees that
something has an active reference on the sysfs_dirent (from the
previous call to sysfs_get_active_two back up the call stack somewhere)
so waits for the active reference to go away, which is of course
impossible.
The problem has been present since 2.6.21.
This patch breaks the deadlock by queuing work queue items on a single-
threaded work queue to remove a slot from sysfs, and to rescan the PCI
buses. There is also some protection against disabling a slot that is
already being removed.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Cc: Kristen Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
SB700 SATA MSI bug will be fixed in SB700 revision A21 at hardware
level, but the SB700 revision older than A21 will also be found in the
market. This patch modify the original quirk commit
bc38b411fe instead of withdrawing it.
The patch also removes quirk to 0x4395 because 0x4395 is SB800 device
ID.
Signed-off-by: Shane Huang <shane.huang@amd.com>
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
According to the PCI Firmware Specification Revision 3.0 section 4.5, _OSC
should only be called on a root brdige. Here is the relevant passage: "The
_OSC interface defined in this section applies only to Host Bridge ACPI
devices that originate PCI, PCI-X, or PCI Express hierarchies". Changed the
code to find the parent root bridge of the device and call _OSC on that.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
AER is only used with PCIe devices so we should only check PCIe devices for
_OSC support.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The function pci_osc_support_set() traverses every root bridge when
checking for _OSC support for a capability. It quits as soon as it finds a
device/bridge that doesn't support the requested capability. This won't
work for systems that have mixed PCI and PCIe bridges when checking for
PCIe features. I split this function into two -- pci_osc_support_set() and
pcie_osc_support_set(). The latter is used when only PCIe devices should be
traversed.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Disable Bus Master, SERR# and INTx to ensure that no new Requests will
be generated from the device before turning power off, in accordance
with the specification.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Set Bad DLLP Mask bit in Correctable Error Mask Register during
turning power off the slot.
This is the workaround against Bad DLLP error that sometimes happen
during turning power off on the slot which conforms to PCI Express
1.0a spec. The cause of this error seems that PCI Express 1.0a spec
doesn't have the following consideration that was added to PCI Express
1.1 spec.
"If the port is associated with a hot-pluggable slot (Hot-Plug
Capable bit in the Slot Capabilities register set to 1b), and
Power Controller Control bit in Slot Control register is 1b(Off),
then any transition to DL Inactive must not be considered an
error."
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
After turning power off, we must wait for at least 1 second *before*
LED operation.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Now that all in-tree users are gone, this removes pci_enable_device_bars()
completely.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch converts users of pci_enable_device_bars() to the new
pci_enable_device_{io,mem} interface.
The new API fits nicely, except maybe for the QLA case where a bit of
code re-organization might be a good idea but I prefer sticking to the
simple patch as I don't have hardware to test on.
I'll also need some feedback on the cs5520 change.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The pci_enable_device_bars() interface isn't well suited to PCI
because you can't actually enable/disable BARs individually on
a device. So for example, if a device has 2 memory BARs 0 and 1,
and one of them (let's say 1) has not been successfully allocated
by the firmware or the kernel, then enabling memory decoding
shouldn't be permitted for the entire device since it will decode
whatever random address is still in that BAR 1.
So a device must be either fully enabled for IO, for Memory, or
for both. Not on a per-BAR basis.
This provides two new functions, pci_enable_device_io() and
pci_enable_device_mem() to replace pci_enable_device_bars(). The
implementation internally builds a BAR mask in order to be able
to use existing arch infrastructure.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Avoid adding the same type of cap multiple times, otherwise we will see dead loop.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
save_state->cap_nr should be correctly set, otherwise we can't find the
saved cap at resume.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
pci_save/store_state has multiple bugs, which will cause cap can't be
saved/restored correctly. Below 3 patches fix them.
fix the typo in pci_save_pcix_state
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Convert quirk printks to dev_printk().
I made the MSI disable messages a little more consistent:
- always use "disabled", not "deactivated"
- specify "device MSI disabled" or "subordinate MSI disabled" when
disabling MSI for only a specific device or subordinate bus
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Instead of printing this:
PCI: Calling quirk c023b250 for 0000:00:00.0
we can print this:
pci 0000:00:00.0: calling quirk 0xc023b270: quirk_cardbus_legacy+0x0/0x30()
The address is superfluous because sprint_symbol() includes the
address if the symbol lookup fails, but this is the same style used
in do_initcalls() and pnp_fixup_device().
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Check that the e100 is in the D0 power state. If it's not, it won't
respond to MMIO accesses and we end up with master-abort machine
checks on some platforms.
Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
A HOWTO that hasn't been updated for half a dozen years no longer
"contains valuable information about which PCI hardware does work under
Linux and which doesn't".
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes the following problem present with older gcc versions:
<-- snip -->
...
CC drivers/pci/msi.o
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/pci/msi.c:692: warning: weak declaration of `arch_msi_check_device' after first use results in unspecified behavior
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/pci/msi.c:704: warning: weak declaration of `arch_setup_msi_irqs' after first use results in unspecified behavior
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/pci/msi.c:724: warning: weak declaration of `arch_teardown_msi_irqs' after first use results in unspecified behavior
...
<-- snip -->
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This adds appropriate casts to avoid a warning and print the correct
values in pr_debug.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The current pci_assign_unassigned_resources() code doesn't work properly
on 32 bits platforms with 64 bits resources. The main reason is the use
of unsigned long in various places instead of resource_size_t.
This is a pre-requisite for making powerpc use the generic code instead of
its own half-useful implementation.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove needless members from struct controller. This has no functional
changes.
Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
acpi_get_name() is called before and after dbg(). The latter is
useless and should be removed.
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Earlier patches to split out the hardware init for PCIe hotplug resulted in
some one-time initializations being redone on every resume cycle. Eg.
irq/polling initialization.
This patch splits the hardware init into two parts, and separates the
one-time initializations from those so that they only ever get done once,
as intended.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make use of the previously split out pcie_init_enable_events() function
to reinitialize the hotplug hardware on resume from suspend, but only
when pciehp_force==1. Otherwise behaviour is unmodified.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Split out the hotplug hardware initialization code from pcie_init()
into pcie_init_enable_events(), without changing any functionality.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix pciehp_probe() to deal with ExpressCard cards
that were inserted prior to the driver being loaded.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
PCI error recovery usually involves the PCI adapter being reset.
If the device is using MSI, the reset will cause the MSI state
to be lost; the device driver needs to restore the MSI state.
The pci_restore_msi_state() routine is currently protected
by CONFIG_PM; remove this, and also export the symbol, so
that it can be used in a modle.
Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix presentation of the slot number in the /sys/bus/pci/slots
directory to match that used in the majority of other drivers.
> Greg said:
> How is anyone supposed to write sane managability tools in the
> presence
> of such anarchy?
>
> > ~ # cat /sys/bus/pci/slots/0000:00:02.2/phy_location
> > U787A.001.DNZ00Z5-P1-C2
>
> Right. This should look like:
>
> # cat /sys/bus/pci/slots/U787A.001.DNZ00Z5-P1-C2/address
> 0000:00:02
This patch implements exactly what you describe. Boot tested.
I assume you really mean it -- if so, then please review and
ack the patch !?
I have absolutely no clue if this breaks any existing IBM tools.
I'm pretty sure it doesn't ... but attention Mike Strosaker! does it?
Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Cc: <strosake@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Unhide the SMBus on the HP xw4100. This gives access to a hardware
monitoring chip (ADT7463) and to the memory module SPD EEPROMs. I
checked that ACPI wasn't accessing the SMBus, so it should be safe.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add and changes a few sanity checks in dmar.c.
1. The haw field in ACPI DMAR table in VT-d spec doesn't describe the
range of haw. But since DMA page size is 4KB in DMA remapping, haw
should be at least 4KB. The current VT-d code in dmar.c returns failure
when haw==0. This sanity check is not accurate and execution can pass
when haw is less than one page size 4KB. This patch changes the haw
sanity check to validate if haw is less than 4KB.
2. Add dmar_rmrr_units verification.
3. Add parse_dmar_table() verification.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Acked-by: mark gross <mgross@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove lots of space-before-) instances. Perhaps these were a workaround for
problems in some long-dead cpp version.
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
There's already a prototype for pci_scan_child_bus() at the correct place in
pci.h, so there's no reason for an additional one.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
sys_pciconfig_{read,write}() are protected against PCI removal with the
reference count in struct pci_dev. The concurrency of
pci_user_{read,write}_config_* functions are already protected by pci_lock
in drivers/pci/access.c.
Signed-off-by: Diego Woitasen <diego@woitasen.com.ar>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The vendor_compatible and device_compatible fields in struct pci_dev aren't
used anywhere, and are somewhat pointless. Assuming that these are
historical artifacts, remove them.
Signed-off-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In an attempt to ensure memory allocation from the local node, the pci
driver temporarily replaces the current task's memory policy with the
system default policy. Trying to be a good citizen, the driver then call's
mpol_get() on the new policy. When it's finished probing, it undoes the
'_get by calling mpol_free() [on the system default policy] and then
restores the current task's saved mempolicy.
A couple of issues here:
1) it's never necessary to set a task's mempolicy to the
system default policy in order to get system default
allocation behavior. Simply set the current task's
mempolicy to NULL and allocations will fall back to
system default policy.
2) we should never [need to] call mpol_free() on the system
default policy. [I plan on trapping this with a VM_BUG_ON()
in a subsequent patch.]
This patch removes the calls to mpol_get() and mpol_free()
and uses NULL for the temporary task mempolicy to effect
default allocation behavior.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
