Use the new pci_alloc_irq_vectors API to allocate MSI-X and MSI vectors.
The big advantage over the old code is that we can use the same API for
MSI and MSI-X, and that we don't need to store the MSI-X vector mapping
in driver-private data structures.
This first conversion keeps the probe order as-is: MSI-X multi vector,
MSI multi vector, MSI single vector, MSI-X single vector and last a
single least legacy interrupt line. There is one small change of
behavior: we now check the "MSI Revert to Single Message" flag for
MSI-X in addition to MSI.
Because the API to find the Linux IRQ number for a MSI/MSI-X vector
is PCI specific, but libahaci is bus-agnostic I had to a
get_irq_vector function pointer to struct ahci_host_priv. The
alternative would be to move the multi-vector case of ahci_host_activate
to ahci.c and just call ata_host_activate directly from the others
users of ahci_host_activate.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Tejun Heo <tj@kernel.org>
irq already contains the interrupt number for the port, don't add the
port index to it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: d684a90d38 ("ahci: per-port msix support")
Cc: stable@vger.kernel.org v4.5+
Due to PCI subsystem behaviour, unloading AHCI driver will disable
MSI and enable INTx. When HBA supports MSIx or Multiple MSI, Driver's
irq handler doesn't clear GHC.IS register. It works well when reading or
writing data and GHC.IS is always non-zero. But when unloading driver
(or any other operation which causes disable MSIx and enable INTx), PCI
subsystem uses config write(Rx04.bit10) to enable INTx. Because
GHC.IS is non-zero, HBA will falsely assume some port needs interrupt
service. Then it asserts INTx. To make things worse, when AHCI controller
shares the same interrupt pin with other PCI device, that PCI device's ISR
will be called and nobody de-asserts previous INTx.
This patch clears GHC.IS in ahci_port_stop() even when using MSIx or
MMSI to prevent this case. It ensures GHC.IS is zero before PCI subsystem
enables INTx.
tj: Minor updates to the comment.
Signed-off-by: Raymond Pang <raymond_rule@hotmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Use accessor functions instead of the raw value.
Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Tejun Heo <tj@kernel.org>
In usecases where force_port_map is used saved_port_map is never set,
resulting in not programming the PORTS_IMPL register as part of initial
config. This patch fixes this by setting it to port_map even in case
where force_port_map is used, making it more inline with other parts of
the code.
Fixes: 566d1827df ("libata: disable forced PORTS_IMPL for >= AHCI 1.3")
Cc: stable@vger.kernel.org # v4.5+
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andy Gross <andy.gross@linaro.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
The HPCP bit is set by bioses for on-board sata ports either because
they think sata is hotplug capable in general or to allow Windows
to display a "device eject" icon on ports which are routed to an
external connector bracket.
However in Redhat Bugzilla #1310682, users report that with kernel 4.4,
where this bit test first appeared, a lot of partitions on sata drives
are now mounted automatically.
This patch should fix redhat and a lot of other distros which
unconditionally automount all devices which have the "removable"
bit set.
Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: 8a3e33cf92 ("ata: ahci: find eSATA ports and flag them as removable" changes userspace behavior)
Link: http://lkml.kernel.org/g/56CF35FA.1070500@redhat.com
Cc: stable@vger.kernel.org #v4.4+
Add new functions ahci_rpm_get_port()/ahci_rpm_put_port() that change
runtime PM status of AHCI ports. Depending if the AHCI host has runtime PM
enabled or disabled calling these may trigger runtime suspend/resume of the
host controller.
We also call these functions in appropriate places to make sure host
controller registers are available before using them.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This allows sysfs nodes to read the cached value directly instead of
powering up possibly runtime suspended controller.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
The flexibility to override the irq handles in the LLD's are already
present, so controllers implementing a edge trigger latch can
implement their own interrupt handler inside the driver. This patch
removes the AHCI_HFLAG_EDGE_IRQ support from libahci and moves edge
irq handling to ahci_xgene.
tj: Minor update to description.
Signed-off-by: Suman Tripathi <stripathi@apm.com>
Signed-off-by: Tejun Heo <tj@kenrel.org>
This patch implements the capability to override the generic AHCI
interrupt handler so that specific ahci drivers can implement their
own custom interrupt handler routines. It also exports
ahci_handle_port_intr so that custom irq_handler implementations can
use it.
tj: s/ahci_irq_handler/irq_handler/ and updated description.
Signed-off-by: Suman Tripathi <stripathi@apm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Some early controllers incorrectly reported zero ports in PORTS_IMPL
register and the ahci driver fabricates PORTS_IMPL from the number of
ports in those cases. This hasn't mattered but with the new nvme
controllers there are cases where zero PORTS_IMPL is valid and should
be honored.
Disable the workaround for >= AHCI 1.3.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Andy Lutomirski <luto@amacapital.net>
Link: http://lkml.kernel.org/g/CALCETrU7yMvXEDhjAUShoHEhDwifJGapdw--BKxsP0jmjKGmRw@mail.gmail.com
Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Cc: stable@vger.kernel.org
The AHCI driver code stops and starts port DMA engines at will
without considering the power state of the particular port. The
AHCI specification isn't very clear on how to handle this scenario,
leaving implementation open to interpretation.
Broadcom's STB SATA host controller is unable to handle port DMA
controller restarts when the port in question is in low power mode.
When a port enters partial or slumber mode, its PHY is powered down.
When a controller restart is requested, the controller's internal
state machine expects the PHY to be brought back up by software which
never happens in this case, resulting in failures.
To avoid this situation, logic is added to manually wake up the port
just before its DMA engine is stopped, if the port happens to be in
a low power state. HBA initiated power management ensures that the port
eventually returns to its configured low power state, when the link is
idle (as per the conditions listed in the spec). A new host flag is also
added to ensure this logic is only exercised for hosts with the above
limitation.
tj: Formatting changes.
Signed-off-by: Danesh Petigara <dpetigara@broadcom.com>
Reviewed-by: Markus Mayer <mmayer@broadcom.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Pull libata updates from Tejun Heo:
"Mostly low level driver specific changes.
Two changes are somewhat noteworthy. First, Dan's patchset to support
per-port msix interrupt handling for ahci, which was tried last cycle
but had to be backed out due to a couple issues, is back and seems to
be working fine. Second, libata exception handling now uses
usleep_range() instead of msleep() for sleeps < 20ms which can make
things snappier in some corner cases"
* 'for-4.5' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
libata: skip debounce delay on link resume
ata: ahci_brcmstb: disable DIPM support
ata: ahci_brcmstb: enable support for ALPM
drivers: libata-core: Use usleep_range() instead of msleep() for short sleeps (<20 ms)
sata_sx4: correctly handling failed allocation
ata: ahci_brcmstb: add support for MIPS-based platforms
ahci: qoriq: Adjust the default register values on ls1021a
ahci: qoriq: Update the default Rx watermark value
ahci: qoriq: Adjust the default register values on ls1043a
ahci: compile out msi/msix infrastructure
ata: core: fix irq description on AHCI single irq systems
ata: ahci_brcmstb: remove unused definitions
ata: ahci_brcmstb: add a quirk for MIPS-based platforms
ata: ahci_brcmstb: disable NCQ for MIPS-based platforms
ata: sata_rcar: Remove obsolete platform_device_id entries
sata_rcar: Add compatible string for r8a7795
ahci: kill 'intr_status'
ahci: switch from 'threaded' to 'hardirq' interrupt handling
ahci: per-port msix support
Quoting Arnd:
The AHCI driver is used for some on-chip devices that do not use PCI
for probing, and it can be built even when CONFIG_PCI is disabled, but
that now results in a build failure:
ata/libahci.c: In function 'ahci_host_activate_multi_irqs':
ata/libahci.c:2475:4: error: invalid use of undefined type 'struct msix_entry'
ata/libahci.c:2475:21: error: dereferencing pointer to incomplete type 'struct msix_entry'
Add ifdef CONFIG_PCI_MSI infrastructure to compile out the multi-msi and
multi-msix code.
Reported-by: Arnd Bergmann <arnd@arndb.de>
Tested--by: Arnd Bergmann <arnd@arndb.de>
[arnd: fix up pci enabled case]
Reported-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Fixes: d684a90d38 ("ahci: per-port msix support")
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Current code doesn't update port value of Port Multiplier(PM) when
sending FIS of softreset to device, command will fail if FBS is
enabled.
There are two ways to fix the issue: the first is to disable FBS
before sending softreset command to PM device and the second is
to update port value of PM when sending command.
For the first way, i can't find any related rule in AHCI Spec. The
second way can avoid disabling FBS and has better performance.
Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
For high frequency I/O the overhead of threaded interrupts impacts
performance. A quick out-of-the-box test (i.e. no affinity tuning)
shows ~10% random read performance at ~20% less cpu. The cpu wins
appear to be from reduced lock contention.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Some AHCI controllers support per-port MSI-X vectors. At the same time
the Linux AHCI driver needs to support one-off architectures that
implement a single MSI-X vector for all ports. The heuristic for
enabling AHCI ports becomes, in order of preference:
1/ per-port multi-MSI-X
2/ per-port multi-MSI
3/ single MSI
4/ single MSI-X
5/ legacy INTX
This all depends on AHCI implementations with potentially broken MSI-X
requesting less vectors than the number of ports. If this assumption is
violated we will need to start explicitly white-listing AHCI-MSIX
implementations.
Reported-by: Ricardo Neri <ricardo.neri@intel.com>
[ricardo: fix struct msix_entry handling]
Reported-by: kernel test robot <ying.huang@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
With devm there is no need to explicitly free irqs on error.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
MSI messages are per-device, so there will never be another "shared"
device in the interrupt chain.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
If the AHCI ports' HPCP or ESP bits are set, the port
should be considered external (e.g. eSATA) and is marked
as removable. Userspace tools like udisks then treat it
like an usb drive.
With this patch applied, when I plug a drive into the esata port,
KDE pops up a window asking what to do with the drives(s), just
like it does for any random USB stick.
Removability is indicated to the upper layers by way of the
SCSI RMB bit, as I haven't found another way to signal
userspace to treat a sata disk like any usb stick.
Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Pull libata updates from Tejun Heo:
- a number of libata core changes to better support NCQ TRIM.
- ahci now supports MSI-X in single IRQ mode to support a new
controller which doesn't implement MSI or INTX.
- ahci now supports edge-triggered IRQ mode to support a new controller
which for some odd reason did edge-triggered IRQ.
- the usual controller support additions and changes.
* 'for-4.2' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata: (27 commits)
libata: Do not blacklist Micron M500DC
ata: ahci_mvebu: add suspend/resume support
ahci, msix: Fix build error for !PCI_MSI
ahci: Add support for Cavium's ThunderX host controller
ahci: Add generic MSI-X support for single interrupts to SATA PCI driver
libata: finally use __initconst in ata_parse_force_one()
drivers: ata: add support for Ceva sata host controller
devicetree:bindings: add devicetree bindings for ceva ahci
ahci: added support for Freescale AHCI sata
ahci: Store irq number in struct ahci_host_priv
ahci: Move interrupt enablement code to a separate function
Doc: libata: Fix spelling typo found in libata.xml
ata:sata_nv - Change 1 to true for bool type variable.
ata: add Broadcom AHCI SATA3 driver for STB chips
Documentation: devicetree: add Broadcom SATA binding
libata: Fix regression when the NCQ Send and Receive log page is absent
ata: hpt366: fix constant cast warning
ata: ahci_xgene: potential NULL dereference in probe
ata: ahci_xgene: Add AHCI Support for 2nd HW version of APM X-Gene SoC AHCI SATA Host controller.
libahci: Add support to handle HOST_IRQ_STAT as edge trigger latch.
...
Currently, ahci supports only msi and intx. To also support msix the
handling of the irq number need to be changed. The irq number for msix
devices is taken from msi_list instead of pci_dev. Thus, the irq
number of a device needs to be stored in struct ahci_host_priv now.
This allows the host controller to be activated in a generic way.
This change is only intended for ahci drivers. For that reason the irq
number is stored in struct ahci_host_priv used only by ahci drivers.
Thus, the ABI changes only for ahci_host_activate(), but existing ata
drivers (about 50) are unaffected and keep unchanged. All users of
ahci_host_activate() have been updated.
While touching drivers/ata/libahci.c, doing a small code cleanup in
ahci_port_start().
Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This patch adds the support to handle HOST_IRQ_STAT as edge trigger
latch.
Signed-off-by: Suman Tripathi <stripathi@apm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This patch refactors the ahci_single_irq_intr function and also rename
it to ahci_single_level_irq_intr as it handles a level triggered latch.
Signed-off-by: Suman tripathi <stripathi@apm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This is a preparation commit that will allow to add other criteria
according to which PHY events should be dropped.
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
According to the Serial ATA AHCI specification, Device Sleep is an
optional feature and as such no errors should be printed if it's
missing. Keep informing users, but use dev_info() instead of dev_err().
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This reverts commit 33fb0d01ce.
18dcf433f3 ("AHCI: Optimize single IRQ interrupt processing") is
scheduled to be reverted. This is an optimization dependent on the
mentioned commit. Revert it first.
Signed-off-by: Tejun Heo <tj@kernel.org>
There is no need to acquire ata_host::lock spinlock from
hardware context single IRQ interrupt handler since the
handler does not access host data that could be altered
by concurrent processors.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: linux-ide@vger.kernel.org
Split interrupt service routine into hardware context handler
and threaded context handler. That allows to protect ports with
individual locks rather than with a single host-wide lock and
move port interrupts handling out of the hardware interrupt
context.
Testing was done by transferring 8GB on two hard drives in
parallel using command 'dd if=/dev/sd{a,b} of=/dev/null'. With
lock_stat statistics I measured access times to ata_host::lock
spinlock (since interrupt handler code is fully embraced with
this lock). The average lock's holdtime decreased eight times
while average waittime decreased two times.
Both before and after the change the transfer time is the same,
while 'perf record -e cycles:k ...' shows 1%-4% CPU time spent
in ahci_single_irq_intr() routine before the update and not even
sampled/shown ahci_single_irq_intr() after the update.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: linux-ide@vger.kernel.org
As described in AHCI v1.0 specification chapter 10.6.2.2
"Multiple MSI Based Messages" generation of interrupts
is not controlled through the HOST_IRQ_STAT register.
Considering MMIO access is expensive remove unnecessary
reading and writing of HOST_IRQ_STAT register.
Further, serializing access to the host data is no longer
needed and the interrupt service routine can avoid competing
on the host lock.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Suggested-by: "Jiang, Dave" <dave.jiang@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: "Jiang, Dave" <dave.jiang@intel.com>
Cc: linux-ide@vger.kernel.org
Currently host activation done by calling either function
ahci_host_activate() or ata_host_activate(). Consolidate
the code by only calling ahci_host_activate() for all AHCI
devices.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: linux-ide@vger.kernel.org
This update is a prerequisite for consolidation of
AHCI host activation code within ahci_host_activate()
function.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: linux-ide@vger.kernel.org
This patch moves force_port_map and mask_port_map into the
ahci_host_priv structure. This allows to modify them into the AHCI
framework. This is needed by the new dt bindings representing ports as
the port_map mask is computed automatically.
Parameters modifying force_port_map, mask_port_map and flags have been
removed from the ahci_platform_init_host() function, and inputs in the
ahci_host_priv structure are now directly filed.
Signed-off-by: Antoine Ténart <antoine.tenart@free-electrons.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
The subsequent patch will make use of them.
Signed-off-by: Loc Ho <lho@apm.com>
Signed-off-by: Suman Tripathi <stripathi@apm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Pull libata updates from Tejun Heo:
"Nothing too interesting - another ahci platform driver variant,
additional controller support, minor fixes and cleanups"
* 'for-3.16' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
ahci: Add Device ID for HighPoint RocketRaid 642L
ata: ep93xx: use dmaengine_prep_slave_sg api instead of internal callback
ahci: add PCI ID for Marvell 88SE91A0 SATA Controller
sata_fsl: remove check for CONFIG_MPC8315_DS
ahci: add support for Hisilicon sata
libahci_platform: add host_flags parameter in ahci_platform_init_host()
ata: ahci: append new hflag AHCI_HFLAG_NO_FBS
ata: use CONFIG_PM_SLEEP instead of CONFIG_PM where applicable in host drivers
ata: ahci_mvebu: new driver for Marvell Armada 380 AHCI interfaces
Documentation: dt-bindings: reformat and order list of ahci-platform compatibles
libata-sff: remove dead code
ata: SATL compliance for Inquiry Product Revision
pata_octeon_cf: use devm_kzalloc() to allocate cf_port
Append AHCI_HFLAG_NO_FBS to force turning off FBS flag.
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>
Signed-off-by: Kefeng Wang <kefeng.wang@linaro.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
On Intel Valleyview SoC, SATA device sleep is not reliable. When
DEVSLP is attempted on certain SSDs, port_devslp write would fail
and result in malfunction of AHCI controller. AHCI controller may
be not shown in PCI enumeration after reset. Complete power source
removal may be required to recover from this failure. So we blacklist
this device and override host device reported capabilities such that
device LPM will only attempt slumber but not DEVSLP.
Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Unreferenced casts of void * types are unnecessary so remove them.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Allwinner A10 and A20 ARM SoCs have an AHCI sata controller which needs a
special register to be poked before starting the DMA engine.
This register gets reset on an ahci_stop_engine call, so there is no other
place then ahci_start_engine where this poking can be done.
This commit allows drivers to override ahci_start_engine behavior for use by
the Allwinner AHCI driver (and potentially other drivers in the future).
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
sparse says:
drivers/ata/libahci.c:1390:5: warning:
symbol 'ahci_pmp_retry_softreset' was not declared. Should it be static?
Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>. Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Cc: linux-ide@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This patch marks the function ahci_port_intr() and
ahci_hw_port_interrupt() as static in libahci.c because they are not
used outside this file.
Thus, it also eliminates the following warnings in libahci.c:
drivers/ata/libahci.c:1767:6: warning: no previous prototype for ‘ahci_port_intr’ [-Wmissing-prototypes]
drivers/ata/libahci.c:1800:6: warning: no previous prototype for ‘ahci_hw_port_interrupt’ [-Wmissing-prototypes]
Signed-off-by: Rashika Kheria <rashika.kheria@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
