Pull NTB updates from Jon Mason:
- fixes for switchtec debugability and mapping table entries
- NTB transport improvements
- a reworking of the peer_db_addr for better abstraction
* tag 'ntb-5.1' of git://github.com/jonmason/ntb:
NTB: add new parameter to peer_db_addr() db_bit and db_data
NTB: ntb_transport: Ensure the destination buffer is mapped for TX DMA
NTB: ntb_transport: Free MWs in ntb_transport_link_cleanup()
ntb_hw_switchtec: Added support of >=4G memory windows
ntb_hw_switchtec: NT req id mapping table register entry number should be 512
ntb_hw_switchtec: debug print 64bit aligned crosslink BAR Numbers
NTB door bell usage depends on NTB hardware.
ex: intel NTB gen1 has one peer door bell register which can be controlled
by the bitmap writen to it, while Intel NTB gen3 has a registers
per door bell and the data trigering the each door bell is always 1.
therefore exposing only peer door bell address forcing the user
to be aware of such low level details
Signed-off-by: Leonid Ravich <Leonid.Ravich@emc.com>
Acked-by: Logan Gunthorpe <logang@deltatee.com>
Acked-by: Dave Jiang <dave.jiang@intel.com>
Acked-by: Allen Hubbe <allenbh@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Presently, when ntb_transport is used with DMA and the IOMMU turned on,
it fails with errors from the IOMMU such as:
DMAR: DRHD: handling fault status reg 202
DMAR: [DMA Write] Request device [00:04.0] fault addr
381fc0340000 [fault reason 05] PTE Write access is not set
This is because ntb_transport does not map the BAR space with the IOMMU.
To fix this, we map the entire MW region for each QP after we assign
the DMA channel. This prevents needing an extra DMA map in the fast
path.
Link: https://lore.kernel.org/linux-pci/499934e7-3734-1aee-37dd-b42a5d2a2608@intel.com/
Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
If NTB peer host crashes or reboots, the NTB transport link will be
down and the MWs of NTB transport will be invalid. But the
ntb_transport_link_cleanup() does not free these invalid MWs. When
the NTB peer host is recovered later, NTB transport link will be
up and the ntb_set_mw() will not reset up MWs. Because the MWs of
NTB transport are invalid, the NTB transport will not work.
We can fix it by freeing MWs when NTB transport link is down, then
the ntb_set_mw() will reset up MWs when NTB transport link is up.
Signed-off-by: Joey Zhang <joey.zhang@microchip.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Current Switchtec's BAR setup registers are limited to 32bits,
corresponding to the maximum MW (memory window) size is <4G.
Increase the MW sizes with the addition of the BAR Setup Extension
Register for the upper 32bits of a 64bits MW size. This increases the MW
range to between 4K and 2^63.
Reported-by: Boris Glimcher <boris.glimcher@emc.com>
Signed-off-by: Paul Selles <paul.selles@microchip.com>
Signed-off-by: Wesley Sheng <wesley.sheng@microchip.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Switchtec NTB crosslink BARs are 64bit addressed but they are printed as
32bit addressed BARs. Fix debug log to increment the BAR numbers by 2 to
reflect the 64bit address alignment.
Fixes: 0175250182 ("ntb_hw_switchtec: Add initialization code for crosslink")
Signed-off-by: Paul Selles <paul.selles@microchip.com>
Signed-off-by: Wesley Sheng <wesley.sheng@microchip.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Clean up the ifdefs which conditionally defined the io{read|write}64
functions in favour of the new common io-64-nonatomic-lo-hi header.
Per a nit from Andy Shevchenko, the include list is also made
alphabetical.
Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We already need to zero out memory for dma_alloc_coherent(), as such
using dma_zalloc_coherent() is superflous. Phase it out.
This change was generated with the following Coccinelle SmPL patch:
@ replace_dma_zalloc_coherent @
expression dev, size, data, handle, flags;
@@
-dma_zalloc_coherent(dev, size, handle, flags)
+dma_alloc_coherent(dev, size, handle, flags)
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
[hch: re-ran the script on the latest tree]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Since IDT PCIe-switch temperature sensor is now always available
irregardless of the EEPROM/BIOS settings, Kconfig and in-code
description should be properly altered. In addition lets update
the driver copyright lines.
Signed-off-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
IDT PCIe-switch temperature sensor interface is very broken. First
of all only a few combinations of TMPCTL threshold enable bits
really cause the interrupts unmasked. Even if an individual bit
indicates the event unmasked, corresponding IRQ just isn't generated.
Most of the threshold enable bits combinations are in fact useless and
non of them can help to create a fully functional alarm interface.
So to speak, we can't create a well defined hwmon alarms based on
the IDT PCI-switch threshold IRQs.
Secondly a single threshold IRQ (not a combination of thresholds) can
be successfully enabled without the issue described above. But in this
case we experienced an enormous number of interrupts generated by
the chip if the temperature got near the enabled threshold value. Filter
adjustment didn't help much. It also doesn't provide a hysteresis settings.
Due to the temperature sample fluctuations near the threshold the
interrupts spate makes the system nearly unusable until the temperature
value finally settled so being pushed either to be fully higher or lower
the threshold.
All of these issues makes the temperature sensor alarm interface useless
and even at some point dangerous to be used in the driver. In this case
it is safer to completely discard it and disable the temperature alarm
interrupts.
Signed-off-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
IDT PCIe switches provide an embedded temperature sensor working
within [0; 127.5]C with resolution of 0.5C. They also can generate
a PCIe upstream interrupt in case if the temperature passes through
specified thresholds. Since this thresholds interface is very broken
the created hwmon-sysfs interface exposes only the next set of hwmon
nodes: current input temperature, lowest and highest values measured,
history resetting, value offset. HWmon alarm interface isn't provided.
IDT PCIe switch also've got an ADC/filter settings of the sensor.
This driver doesn't expose them to the hwmon-sysfs interface at the
moment, except the offset node.
Signed-off-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
In order to create a hwmon interface for the IDT PCIe-switch temperature
sensor the already available reader method should be improved. Particularly
we need to redesign it so one would be able to read temperature/offset
values from registers of the passed types. Since IDT sensor interface
provides temperature in unsigned format 0:7:1 (7 bits for real value
and one for fraction) we also need to have helpers for the typical sysfs
temperature data type conversion to and from this format. Even though
the IDT PCIe-switch provided temperature offset got the same but signed
type it can be translated by these methods too.
Signed-off-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Be a little wasteful if the (likely CMA) message window buffer is not
suitably aligned after our first attempt; allocate a buffer twice as big
as we need and manually align our MW buffer within it.
This was needed on Intel Broadwell DE platforms with intel_iommu=off
Signed-off-by: Aaron Sierra <asierra@xes-inc.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Addresses-Coverity-ID: 1373888 ("Missing break in switch")
Addresses-Coverity-ID: 1373889 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Allen Hubbe <allenbh@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
IDT NTB driver sets the upper limit of actual translation address
being written to the corresponding memory window setup. It is achieved
by BARLIMITx register initialization. Needless to say, that the register
works within PCIe bus address space.
In general CPU and PCIe address spaces are different. It means,
that addresses used for Memory TLPs routine can be different from
CPU addresses. While in most of cases they are the same, there are
exceptions when the proper mapping must be performed to have the
portable driver code. There used to be a virt_to_bus()/bus_to_virt()
interface for this purpose. But it's deprecated now. It was also a
mistake to use pci_resource_start() since the return address of the
method is at the CPU address space. In order to achieve the desired
purpose we need to use pci_bus_address() helper. This method shall
return a PCIe bus base address of the corresponding BAR resource.
Signed-off-by: Serge Semin <fancer.lancer@gmail.com>
Acked-by: Allen Hubbe <allenbh@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Both devm_kcalloc() and devm_kzalloc() return NULL on error. They
never return error pointers.
The use of IS_ERR_OR_NULL is currently applied to the wrong
context.
Fix this by replacing IS_ERR_OR_NULL with regular NULL checks.
Fixes: bf2a952d31 ("NTB: Add IDT 89HPESxNTx PCIe-switches support")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
ndev_vec_mask() should be returning u64 mask value instead of int.
Otherwise the mask value returned can be incorrect for larger
vectors.
Fixes: e26a5843f7 ("NTB: Split ntb_hw_intel and ntb_transport drivers")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Tested-by: Lucas Van <lucas.van@intel.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Move the Microsemi Switchtec PCI Vendor ID (same as
PCI_VENDOR_ID_PMC_Sierra) to pci_ids.h. Also, replace Microsemi class
constants with the standard PCI definitions.
Signed-off-by: Doug Meyer <dmeyer@gigaio.com>
[bhelgaas: restore SPDX (I assume it was removed by mistake), remove
device ID definitions]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
Pull more overflow updates from Kees Cook:
"The rest of the overflow changes for v4.18-rc1.
This includes the explicit overflow fixes from Silvio, further
struct_size() conversions from Matthew, and a bug fix from Dan.
But the bulk of it is the treewide conversions to use either the
2-factor argument allocators (e.g. kmalloc(a * b, ...) into
kmalloc_array(a, b, ...) or the array_size() macros (e.g. vmalloc(a *
b) into vmalloc(array_size(a, b)).
Coccinelle was fighting me on several fronts, so I've done a bunch of
manual whitespace updates in the patches as well.
Summary:
- Error path bug fix for overflow tests (Dan)
- Additional struct_size() conversions (Matthew, Kees)
- Explicitly reported overflow fixes (Silvio, Kees)
- Add missing kvcalloc() function (Kees)
- Treewide conversions of allocators to use either 2-factor argument
variant when available, or array_size() and array3_size() as needed
(Kees)"
* tag 'overflow-v4.18-rc1-part2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux: (26 commits)
treewide: Use array_size in f2fs_kvzalloc()
treewide: Use array_size() in f2fs_kzalloc()
treewide: Use array_size() in f2fs_kmalloc()
treewide: Use array_size() in sock_kmalloc()
treewide: Use array_size() in kvzalloc_node()
treewide: Use array_size() in vzalloc_node()
treewide: Use array_size() in vzalloc()
treewide: Use array_size() in vmalloc()
treewide: devm_kzalloc() -> devm_kcalloc()
treewide: devm_kmalloc() -> devm_kmalloc_array()
treewide: kvzalloc() -> kvcalloc()
treewide: kvmalloc() -> kvmalloc_array()
treewide: kzalloc_node() -> kcalloc_node()
treewide: kzalloc() -> kcalloc()
treewide: kmalloc() -> kmalloc_array()
mm: Introduce kvcalloc()
video: uvesafb: Fix integer overflow in allocation
UBIFS: Fix potential integer overflow in allocation
leds: Use struct_size() in allocation
Convert intel uncore to struct_size
...
ntb_transport_create_queue() is never called in atomic context.
ntb_transport_create_queue() is only called by ntb_netdev_probe(),
which is set as ".probe" in struct ntb_transport_client.
Despite never getting called from atomic context,
ntb_transport_create_queue() calls kzalloc_node() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
ntb_transport_setup_qp_mw() is never called in atomic context.
ntb_transport_setup_qp_mw() is only called by ntb_transport_link_work(),
which is set as a parameter of INIT_DELAYED_WORK()
in ntb_transport_probe().
Despite never getting called from atomic context,
ntb_transport_setup_qp_mw() calls kzalloc_node() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
