Current code calculates ChannelCount based on CAP(NP) value.
It only works when the ports implemented number are <= CAP(NP),
for example, platform has CAP(NP) = 5 (means 6 ports) and ports
implemented are 0, 1, 2, 3, 4 and 5.
But we have some platform that has CAP(NP) = 1 (means 2 ports) and
ports implemented are 1 and 2, and has no port 0 implemented, then
current code does not work.
This patch updates the code to calculate ChannelCount based on PI value.
Cc: Amy Chan <amy.chan@intel.com>
Cc: Hong-chihX Hsueh <hong-chihx.hsueh@intel.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Sami Mujawar <sami.mujawar@arm.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Hao Wu <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Tested-by: Hong-chihX Hsueh <hong-chihx.hsueh@intel.com>
1. Do not use tab characters
2. No trailing white space in one line
3. All files must end with CRLF
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liming Gao <liming.gao@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=256
This commit adds the PEI BlockIo support for NVM Express devices.
The driver will consume the EDKII_NVM_EXPRESS_HOST_CONTROLLER_PPI for NVM
Express host controllers within the system. And then produces the
BlockIo(2) PPIs for each controller.
The implementation of this driver is currently based on the NVM Express 1.1
Specification, which is available at:
http://nvmexpress.org/resources/specifications/
Cc: Star Zeng <star.zeng@intel.com>
Cc: Eric Dong <eric.dong@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The SATA controller driver crashes while accessing the
PCI memory [AHCI Base Registers (ABAR)], as the PCI memory
space is not enabled.
Enable the PCI memory space access to prevent the SATA
Controller driver from crashing.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Per PI spec, the Length value is the length of the ACPI descriptor
in bytes, excluding the first two fields.
The patch fixes the code to report the correct descriptor length
by excluding 3-byte first two fields.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
REF:https://bugzilla.tianocore.org/show_bug.cgi?id=980
Adjust the DEBUG prints within function:
NvmeRead(), NvmeWrite(), AsyncNvmeRead() and AsyncNvmeWrite()
to DEBUG_BLKIO for the consistency with other storage device drivers
(e.g. ATA, USB and etc.).
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Eric Dong <eric.dong@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Per PCIE spec, Memory Write and Invalidate is hardwired to 0b so
PciBus driver shouldn't write 1b to it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Chasel Chiu <chasel.chiu@intel.com>
CreatePciIoDevice() detects whether the PCI device is a PCI Express
device and remembers the device type in PciIoDevice->IsPciExp.
RegisterPciDevice() detects the device type again which is
unnecessary. The detection logic can be removed.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Current code assumes the max bus(0xFF) is under this P2P bridge and
temporarily set it as subordinate bus.
It may cause silicon hangs during PCI enumeration in some specific
case.
Instead, it should get the max bus number from the bus number
resources returned from
PCI_HOST_BRIDGE_RESOURCE_ALLOCATION.StartBusEnumeration() and set it
as subordinate bus.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
RootBridgeIo.PollMem()/PollIo() originally don't count the IO/MMIO
access overhead when delaying.
The patch changes the implementation to count the access overhead
so that the actually delay equals to user required delay.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Cc: Chasel Chiu <chasel.chiu@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=937
In NvmeExpressPassthru.c near line 659:
Prp = NvmeCreatePrpList (
PciIo,
PhyAddr,
EFI_SIZE_TO_PAGES(Offset + Bytes) - 1,
&PrpListHost,
&PrpListNo,
&MapPrpList
);
if (Prp == NULL) {
goto EXIT;
}
Status is not set to an error code - Status is initialized to
EFI_SUCCESS, or set by a PciIo->Map to EFI_SUCCESS above this
code. This error path should set Status to an error code before
goto EXIT.
Change-Id: I8a5cdf981aa609534c205d3676395805ac60a003
Cc: Hao Wu <hao.a.wu@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Michael Turner <Michael.Turner@microsoft.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
According to UEFI spec 2.7, PciIo->GetBarAttributes should return host
address (CPU view ddress) rather than device address (PCI view
address), and
device address = host address + address translation offset,
so we subtract translation from device address before returning.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Signed-off-by: Yi Li <phoenix.liyi@huawei.com>
Reviewed-by: Ni Ruiyu <ruiyu.ni@intel.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
According to UEFI spec 2.7, PciRootBridgeIo->Configuration() should
return host address (CPU view ddress) rather than device address
(PCI view address), so in function GetMmioAddressTranslationOffset we
need to convert the range to device address before comparing.
And device address = host address + translation offset.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Signed-off-by: Yi Li <phoenix.liyi@huawei.com>
Reviewed-by: Ni Ruiyu <ruiyu.ni@intel.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
PCI address translation is necessary for some non-x86 platforms. On
such platforms, address value (denoted as "device address" or "address
in PCI view") set to PCI BAR registers in configuration space might be
different from the address which is used by CPU to access the
registers in memory BAR or IO BAR spaces (denoted as "host address" or
"address in CPU view"). The difference between the two addresses is
called "Address Translation Offset" or simply "translation", and can
be represented by "Address Translation Offset" in ACPI QWORD Address
Space Descriptor (Offset 0x1E). However UEFI and ACPI differs on the
definitions of QWORD Address Space Descriptor, and we will follow UEFI
definition on UEFI protocols, such as PCI root bridge IO protocol and
PCI IO protocol. In UEFI 2.7, "Address Translation Offset" is "Offset
to apply to the Starting address to convert it to a PCI address". This
means:
1. Translation = device address - host address.
2. PciRootBridgeIo->Configuration should return CPU view address, as
well as PciIo->GetBarAttributes.
Summary of addresses used in protocol interfaces and internal
implementations:
1. *Only* the following protocol interfaces assume Address is Device
Address:
(1). PciHostBridgeResourceAllocation.GetProposedResources()
Otherwise PCI bus driver cannot set correct address into PCI
BARs.
(2). PciRootBridgeIo.Mem.Read() and PciRootBridgeIo.Mem.Write()
(3). PciRootBridgeIo.CopyMem()
UEFI and PI spec have clear statements for all other protocol
interfaces about the address type.
2. Library interfaces and internal implementation:
(1). Base and Limit in PCI_ROOT_BRIDGE_APERTURE are device address.
It is easy to check whether the address is below 4G or above 4G.
(2). Addresses in PCI_ROOT_BRIDGE_INSTANCE.ResAllocNode are host
address, for they are allocated from GCD.
(3). Address passed to PciHostBridgeResourceConflict is host address,
for it comes from PCI_ROOT_BRIDGE_INSTANCE.ResAllocNode.
RESTRICTION: to simplify the situation, we require the alignment of
Translation must be larger than any BAR alignment in the same root
bridge, so that resource allocation alignment can be applied to both
device address and host address.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Signed-off-by: Yi Li <phoenix.liyi@huawei.com>
Reviewed-by: Ni Ruiyu <ruiyu.ni@intel.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Initialize local variable to suppress warning C4703:
potentially uninitialized local pointer variable.
Both reads (dereferences) of "PciRootBridgeIo" in
PciBusDriverBindingStart() are only reached if
"gFullEnumeration" is TRUE on entry *and* we successfully
open the EfiPciRootBridgeIoProtocol interface.
Cc: Star Zeng <star.zeng@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Dandan Bi <dandan.bi@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Invoke the newly introduced SD/MMC override protocol to override
the capabilities register after reading it from the device registers,
and to call the pre/post host init and reset hooks at the appropriate
times.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Within function NvmExpressPassThru():
The data buffer for the below 2 Admin command:
Create I/O Completion Queue command (Opcode 01h)
Create I/O Submission Queue command (Opcode 05h)
are not mapped to the PCI controller specific addresses.
But the current code logic also prevents the below NVM command:
Write (Opcode 01h)
from mapping its data buffer.
Hence, this commit refine the logic to resolve this issue.
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Per PI spec, the PciEnumerationComplete protocol installation
should be after PciIo installation.
Today's implementation installs the PciEnumerationComplete
after hardware enumeration is completed, but before PciIo
installation.
The change corrects the spec/implementation gap.
The change also benefits certain implementation that depends on
the PciIo handle in PciEnumerationComplete callback.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The patch is just a code cleanup with no functionality impact.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
It didn't cause big issues when VT-d was disabled.
But in VT-d enabled platform, lack of EhcFreeUrb call caused
the DMA data was not moved back to user's buffer.
It caused the correct data cannot be got through sync interrupt
transfer.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
EBC compiler doesn't treat EFI_xxx as constant due to these macros
are UINT64 type in 64bit env and UINT32 type in 32bit env.
So it reports error when "case EFI_xxx" is used.
The patch changes to use if-else to fix EBC build failure.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
DUAL_ADDRESS_CYCLE is missing in the EFI_UNSUPPORTED
return status description.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Commit f6b139b added return status handling to PciIo->Mem.Write.
However, the second status handling will override EFI_DEVICE_ERROR
returned in this branch:
//
// Check the NVMe cmd execution result
//
if (Status != EFI_TIMEOUT) {
if ((Cq->Sct == 0) && (Cq->Sc == 0)) {
Status = EFI_SUCCESS;
} else {
Status = EFI_DEVICE_ERROR;
^^^^^^^^^^^^^^^^
Since PciIo->Mem.Write will probably return SUCCESS, it causes
NvmExpressPassThru to return SUCCESS even when DEVICE_ERROR occurs.
Callers of NvmExpressPassThru will then continue executing which may
cause further unexpected results, e.g. DiscoverAllNamespaces couldn't
break out the loop.
So we save previous status before calling PciIo->Mem.Write and restore
the previous one if it already contains error.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Cc: Eric Dong <eric.dong@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Update the UhciPei driver to consume IOMMU_PPI to allocate DMA buffer.
If no IOMMU_PPI exists, this driver still calls PEI service to allocate
DMA buffer, with assumption that DRAM==DMA.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
This patch is following 2c656af04d.
1. Fix typo "XHC" to "EHC".
2. Reinitialize Request(Phy/Map) and Data(Phy/Map)
in Urb, otherwise the last time value of them may
be used in error handling when error happens.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
This reverts commit 5db417ed25.
"MdeModulePkg/PciBusDxe: Enable Bus Master on P2P bridges on demand"
We met some compatibility issues when doing Windows S4 resume.
Reverting the BME disabling patches to fix the S4 resume issue.
Signed-off-by: Ruiyu Ni <Ruiyu.ni@intel.com>
Signed-off-by: Michael Turner <michael.turner@microsoft.com>
Cc: Michael D Kinney <Michael.d.kinney@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
This reverts commit 050763db07.
"MdeModulePkg/PciBus: Disable BME of all devices when entering RT"
We met some compatibility issues when doing Windows S4 resume.
Reverting the BME disabling patches to fix the S4 resume issue.
Signed-off-by: Ruiyu Ni <Ruiyu.ni@intel.com>
Signed-off-by: Michael Turner <michael.turner@microsoft.com>
Cc: Michael D Kinney <Michael.d.kinney@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
1. Call IoMmuInit() after locating gPeiUsbControllerPpiGuid.
2. Call XhcPeiFreeSched() to do cleanup in XhcEndOfPei.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
V2: Halt HC at EndOfPei.
Update the EhciPei driver to consume IOMMU_PPI to allocate DMA buffer.
If no IOMMU_PPI exists, this driver still calls PEI service to allocate
DMA buffer, with assumption that DRAM==DMA.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
For PciIoPciRead interface, memory prior to Buffer would be written
with zeros if Offset was larger than sizeof (Dev->ConfigSpace), which
would cause serious system exception.
So we add a pre-check branch to avoid memory override.
Cc: Star Zeng <star.zeng@intel.com>
Cc: Eric Dong <eric.dong@intel.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
The patch ensures all DMA transactions are blocked after
ExitBootService.
If a platform enables IOMMU before and needs disable IOMMU after
ExitBootService, the IOMMU should be disabled after PCI bus driver
disables BME.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Michael Turner <michael.turner@microsoft.com>
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
Cc: Jeff Fan <vanjeff_919@hotmail.com>
Currently, we complete a synchronous operation without unmapping the
DMA mappings, and free the pages using FreePages () rather than calling
EFI_PCI_IO_PROTOCOL::FreeBuffer. This is simply incorrect, but it also
breaks non-coherent DMA as well as DMA protection and/or memory encryption
so let's do it correctly and call SdMmcFreeTrb() instead.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
It's a regression of below commit:
SHA-1: 8be37a5cee
* MdeModulePkg/SecurityStubDxe: Defer 3rd party image before EndOfDxe
When PciBus driver fails to load the Option ROM, it doesn't produce
BusOverride protocol. It was a correct behavior before the above
commit. But due to the above commit, BusOverride protocol never is
produced by PciBus driver.
The patch fixes this issue using the following solution:
1. PciBus records the image device path when LoadImage fails.
2. Override.GetDriver() tries to look for the image handle using
the stored image device path.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The patch shouldn't impact functionality.
Original code records the ImageHandle of Option ROM by creating a
new entry. It's not necessary.
The patch updates the ImageHandle in the old entry.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The patch doesn't impact real functionality.
It only renames EFI_PCI_ROM_IMAGE_MAPPING to PCI_ROM_IMAGE,
and changes prototype of PciRomAddImageMapping so that
no explicit type cast is needed when calling this function.
It also removes unused field RomBase from PCI_IO_DEVICE structure.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The bug was caused by 728d74973c
"MdeModulePkg/PciBus: Count multiple hotplug resource paddings".
The patch firstly updated the Bridge->Alignment to the maximum
alignment of all devices under the bridge, then aligned the
Bridge->Length to Bridge->Alignment.
It caused too much resources were claimed.
The new patch firstly aligns Bridge->Length to Bridge->Alignment,
then updates the Bridge->Alignment to the maximum alignment of all
devices under the bridge.
Because the step to update the Bridge->Alignment is to make sure
the resource allocated to the bus under the Bridge meets all
devices alignment. But the Bridge->Length doesn't have to align
to the maximum alignment.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Eric Dong <eric.dong@intel.com>
Existing code forgot to set Descriptor->SpecificFlag to 0 when
the resource type is non-prefetchable MMIO.
The patch adds the missing assignment.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Eric Dong <eric.dong@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=720
The current implementation assumes there is only one hotplug resource
padding for each resource type. It's not true considering
DegradeResource(): MEM64 resource could be degraded to MEM32
resource.
The patch treat the resource paddings using the same logic as
treating typical/actual resources and the total resource of a bridge
is set to the MAX of typical/actual resources and resource paddings.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The patch dynamically enables Bus Master on P2P bridges only
when requested by a device driver through PciIo.Attribute() to enable
the Bus Master.
Signed-off-by: Sean Brogan <sean.brogan@microsoft.com>
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Update XHCI driver to consume IOMMU_PPI to allocate DMA buffer.
If no IOMMU_PPI exists, this driver still calls PEI service
to allocate DMA buffer, with assumption that DRAM==DMA.
This is a compatible change.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=653
Correct description of Timeout param in XhciReg.h to be matched with
XhciReg.c.
Cc: Alexei Fedorov <Alexei.Fedorov@arm.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
We found there are loops of *2* Maps and only *1* Unmap and
the DMA buffer address is decreasing.
It is caused by the below code flow.
XhcAsyncInterruptTransfer ->
XhcCreateUrb ->
XhcCreateTransferTrb ->
Map Urb->DataMap (1)
Timer: loops of *2* Maps and only *1* Unmap
XhcMonitorAsyncRequests ->
XhcFlushAsyncIntMap ->
Unmap and Map Urb->DataMap (2)
XhcUpdateAsyncRequest ->
XhcCreateTransferTrb ->
Map Urb->DataMap (3)
This patch is to eliminate (3).
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Per NVM Express Spec, software should notify NVME HW when shutdown
occurs.
The NVME controller requires notification for shutdown as part of
its management of internal structures. Even with FUA, failing to
notify the NVME controller to shutdown power off causes the NVME
controller to take quite some time to organize its tables on the
next power on. This time exceeds the normal timeout, so we would
fail to boot the NVME disk.
The host should set the Shutdown Notification (CC.SHN) field to 01b
to indicate a normal shutdown operation. The controller indicates
when shutdown processing is completed by updating the Shutdown Status
(CSTS.SHST) field to 10b.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
PciScanBus() assumes the GetResourcePadding() puts BUS descriptor
in the very beginning, if it's not, the Descriptors will be updated
to point to middle of the pool buffer, which can cause
FreePool(Descriptors) hang in DEBUG image.
No functionality impact to RELEASE image.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=644
According to XHCI spec:
4.10.2.1 Stall Error
4.10.2.4 Babble Detected Error
When a device transmits more data on the USB than the host controller
is expecting for a transaction, it is defined to be babbling.
In general, this is called a Babble Error. When a device sends more
data than the TD Transfer Size bytes (TD Babble), unexpected activity
that persists beyond a specified point in a (micro)frame (Frame Babble),
or a packet greater than Max Packet Size (Packet Babble), the host
controller shall set the Babble Detected Error in the Completion Code
field of the TRB, generate an Error Event, and halt the endpoint
(refer to Section 4.10.2.1).
This patch is to also check for EFI_USB_ERR_BABBLE error returned as
a TransferResult and then proceed to XhcRecoverhaltedEndPoint.
Cc: Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
https://bugzilla.tianocore.org/show_bug.cgi?id=433
When a blocking NVMe PassThru request experiences timeout, the current
codes in function NvmExpressPassThru() do not abort the timeout request
while advancing synchronous Submission Queue tail. Therefore, it is
possible to submit a new blocking PassThru request when the synchronous
Submission Queue is full.
The commit adds logic to abort the timeout request by resetting the NVMe
controller when a timeout occurs for a blocking PassThru request.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
This fixes BULK data loss when transfer is detected as timeout but
finished just before stopping endpoint.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Cc: Feng Tian <feng.tian@intel.com>
The patch separates the common logic in XhcControlTransfer,
XhcBulkTransfer and XhcSyncIntTransfer to a sub-routine
XhcTransfer. It doesn't have functionality impact.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Cc: Feng Tian <feng.tian@intel.com>
Current implementation of IsTransferRingTrb only checks whether
the TRB is in the RING of the URB.
The patch enhanced the logic to check that whether the TRB belongs
to the transaction of URB.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Cc: Feng Tian <feng.tian@intel.com>
Some USB devices don't report data periodically through Int
Transfer. They report data only when be asked. If the TRB
is not removed from the XHCI HW, when next time HOST asks
data again, the data is reported but consumed by the previous
TRB, which results the HOST thinks data never comes.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao A Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
https://bugzilla.tianocore.org/show_bug.cgi?id=510
The commit fills the 'Interval' field of the Endpoint Context data for
isochronous endpoints. It will resolve the error when a Configure
Endpoint Command is sent to an isochronous endpoint.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
If IOMMU protocol is installed, PciBus need call IOMMU
to set access attribute for the PCI device in Map/Ummap.
Only after the access attribute is set, the PCI device can
access the DMA memory.
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Leo Duran <leo.duran@amd.com>
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Previous patch Tested-by: Brijesh Singh <brijesh.singh@amd.com>
Previous patch Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Leo Duran <leo.duran@amd.com>
If IOMMU protocol is installed, PciHostBridge just calls
IOMMU AllocateBuffer/FreeBuffer/Map/Unmap.
PciHostBridge does not set IOMMU access attribute,
because it does not know which device request the DMA.
This work is done by PciBus driver.
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Leo Duran <leo.duran@amd.com>
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Previous patch Tested-by: Brijesh Singh <brijesh.singh@amd.com>
Previous patch Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Leo Duran <leo.duran@amd.com>
In function UfsHcDriverBindingStart(), the return value 'Status' may be
overridden during the original PCI attributes restore process.
This commit refines the logic to avoid such override.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The commit removes the unused field 'EFI_HANDLE Handle' in Ufs host
controller private data structure 'UFS_HOST_CONTROLLER_PRIVATE_DATA'.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
In case of an async command if updating the submission queue tail
doorbell fails then the command will not be picked up by device and
no completion response will be created. This scenario has to be handled.
Also if we create an AsyncRequest element and insert in the async queue,
it will never receive a completion so in the timer routine this element
won't be freed, resulting in memory leak. Also in case of blocking calls
we should capture the status of updating completion queue head doorbell
register and return it to caller of PassThru.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Suman Prakash <suman.p@samsung.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
In function ReadCapacity(), the following expression:
MediaInfo->LastBlock = (Data.LastLba3 << 24) |
(Data.LastLba2 << 16) |
(Data.LastLba1 << 8) |
Data.LastLba0;
(There is also a similar case in this function.)
will involve undefined behavior in signed left shift operations.
Since Data.LastLbaX is of type UINT8, and MediaInfo->LastBlock is of type
UINTN. Therefore, Data.LastLbaX will be promoted to int (32 bits, signed)
first, and then perform the left shift operation.
According to the C11 spec, Section 6.5.7:
4 The result of E1 << E2 is E1 left-shifted E2 bit positions; vacated
bits are filled with zeros. If E1 has an unsigned type, the value
of the result is E1 * 2^E2 , reduced modulo one more than the
maximum value representable in the result type. If E1 has a signed
type and nonnegative value, and E1 * 2^E2 is representable in the
result type, then that is the resulting value; otherwise, the
behavior is undefined.
So if bit 7 of Data.LastLba3 is 1, (Data.LastLba3 << 24) will be out of
the range within int type. The undefined behavior of the signed left shift
will lead to a potential of setting the high 32 bits of
MediaInfo->LastBlock to 1 during the cast from type int to type UINT64
for X64 builds.
This commit will add an explicit UINT32 type cast for Data.LastLba3 to
resolve this issue.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
For async commands, the buffer allocated for Prp list is
not getting freed, which will cause memory leak for async
read write command. For example testing async command flow
with custom application to send multiple read write commands
were resulting in decrease of available memory page in memmap,
which eventually resulted in system hang. Hence freeing
AsyncRequest->MapData, AsyncRequest->MapMeta, AsyncRequest->MapPrpList and
AsyncRequest->PrpListHost when async command is completed.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Suman Prakash <suman.p@samsung.com.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
At worst case, OCR register may always not set BIT31. It will cause
original code enter to dead loop. Adding a break for such case.
Cc: Hao Wu <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
There are cases that the operands of an expression are all with rank less
than UINT64/INT64 and the result of the expression is explicitly cast to
UINT64/INT64 to fit the target size.
An example will be:
UINT32 a,b;
// a and b can be any unsigned int type with rank less than UINT64, like
// UINT8, UINT16, etc.
UINT64 c;
c = (UINT64) (a + b);
Some static code checkers may warn that the expression result might
overflow within the rank of "int" (integer promotions) and the result is
then cast to a bigger size.
The commit refines codes by the following rules:
1). When the expression is possible to overflow the range of unsigned int/
int:
c = (UINT64)a + b;
2). When the expression will not overflow within the rank of "int", remove
the explicit type casts:
c = a + b;
3). When the expression will be cast to pointer of possible greater size:
UINT32 a,b;
VOID *c;
c = (VOID *)(UINTN)(a + b); --> c = (VOID *)((UINTN)a + b);
4). When one side of a comparison expression contains only operands with
rank less than UINT32:
UINT8 a;
UINT16 b;
UINTN c;
if ((UINTN)(a + b) > c) {...} --> if (((UINT32)a + b) > c) {...}
For rule 4), if we remove the 'UINTN' type cast like:
if (a + b > c) {...}
The VS compiler will complain with warning C4018 (signed/unsigned
mismatch, level 3 warning) due to promoting 'a + b' to type 'int'.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
For pointer subtraction, the result is of type "ptrdiff_t". According to
the C11 standard (Committee Draft - April 12, 2011):
"When two pointers are subtracted, both shall point to elements of the
same array object, or one past the last element of the array object; the
result is the difference of the subscripts of the two array elements. The
size of the result is implementation-defined, and its type (a signed
integer type) is ptrdiff_t defined in the <stddef.h> header. If the result
is not representable in an object of that type, the behavior is
undefined."
In our codes, there are cases that the pointer subtraction is not
performed by pointers to elements of the same array object. This might
lead to potential issues, since the behavior is undefined according to C11
standard.
Also, since the size of type "ptrdiff_t" is implementation-defined. Some
static code checkers may warn that the pointer subtraction might underflow
first and then being cast to a bigger size. For example:
UINT8 *Ptr1, *Ptr2;
UINTN PtrDiff;
...
PtrDiff = (UINTN) (Ptr1 - Ptr2);
The commit will refine the pointer subtraction expressions by casting each
pointer to UINTN first and then perform the subtraction:
PtrDiff = (UINTN) Ptr1 - (UINTN) Ptr2;
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
Compiler calculates the PciBar[BarIndex] using
sizeof (PciBar[0]) * BarIndex, when BarIndex is type of UINT64,
the above calculation generates assembly code using _allmul.
Change BarIndex to UINTN to avoid the build failure.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Wu Jiaxin <jiaxin.wu@intel.com>
When the VendorId/DeviceId/RevisionId/SubsystemVendorId
/SubsystemDeviceId is MAX_UINTN, IncompatiblePciDeviceSupport
driver doesn't use it to match any IDs.
The patch fixes this bug.
Since PciBus driver always calls IncompatiblePciDeviceSupport
using IDs read from HW, MAX_UINTN is never passed to this
driver.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
The patch replaces the following macros:
DEVICE_ID_NOCARE (0xFF) --> MAX_UINT64
PCI_ACPI_UNUSED (0) --> 0
PCI_BAR_ALL (0xFF) --> MAX_UINT64
PCI_BAR_NOCHANGE (0) --> 0
PCI_BAR_EVEN_ALIGN --> EVEN_ALIGN (local definition)
Since the PciBus driver was updated to accept Spec defined values
in previous commit, the above replacements don't impact
functionality.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
PI spec IncompatiblePciSupport part defines (UINT64)-1 as all BARs
and 0 to use existing alignment. PciBus driver didn't accept these
values. It treated 0xFF as all BARs and 0xFFFFFFFFFFFFFFFFULL to use
existing alignment.
The patch changes the code to still accept old values while also
accept values defined in PI spec. So that the driver can provide
backward compatibility and follow spec.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
When BarIndex equals to 0xFF, default value 0 is used as the BAR
index. Though PCI_BAR_ALL and MAX_UINT8 shares the same value,
using PCI_BAR_ALL is like to match any BAR not BAR 0, it's more
proper to use MAX_UINT8 here.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
If the code eventually returns "Status" anyway, it does not make
sense to explicitly return "Status" in case of an error, too.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Refine the codes to compare the definition 'SIZE_4GB' with type
EFI_PHYSICAL_ADDRESS.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Add support for non-coherent DMA, either by performing explicit cache
maintenance when DMA mappings are aligned to the CPU's DMA buffer alignment,
or by bounce buffering via uncached mappings otherwise.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Fix switch/case statement type mismatch in functions PciIoMemRead &
PciIoMemWrite.
Parameter 'Width' is of enum type EFI_PCI_IO_PROTOCOL_WIDTH, but the enum
type provided in 'switch (Width)' block is of type
EFI_PCI_ROOT_BRIDGE_IO_PROTOCOL_WIDTH.
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Add missing EFIAPI modifiers to the functions that are exposed via the
PCI I/O protocol.
At the same time, add a missing UINT8 cast which breaks the build on
Visual Studio.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Dandan Bi <dandan.bi@intel.com>
This implements support for non-discoverable PCI compatible devices, i.e,
devices that are not on a PCI bus but that can be controlled by generic PCI
drivers in EDK2.
This is implemented as a UEFI driver, which means we take full advantage
of the UEFI driver model, and only instantiate those devices that are
necessary for booting.
Care is taken to deal with DMA addressing limitations: DMA mappings and
allocations are moved below 4 GB if the PCI driver has not informed us
that the device being driven is 64-bit DMA capable. DMA is implemented as
coherent, support for non-coherent DMA is implemented by a subsequent patch.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Port status bits are clear in original code, so no enumeration
takes place.
Changing this to prevent the status bits from being cleared
allows enumeration to proceed normally.
Cc: Star Zeng <star.zeng@intel.com>
Cc: Mike Turner <Michael.Turner@microsoft.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Mike Turner <Michael.Turner@microsoft.com>
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The commit e27cca has a typo on DEBUG level macro. And this debug
message should be DEBUG_INFO rather than DEBUG_ERROR.
Cc: Jan Dabros <jsd@semihalf.com>
Cc: Marcin Wojtas <mw@semihalf.com>
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
reviewed-by: Marcin Wojtas <mw@semihalf.com>
EFI_D_INFO, EFI_D_VERBOSE, EFI_D_WARN and EFI_D_ERROR are replaced
with currently recommended values.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jan Dabros <jsd@semihalf.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
We send ADDRESS DEVICE CMD in XhcInitializeDeviceSlot(), which will
cause XHC issue a USB SET_ADDRESS request to the USB Device.
According to USB spec, there should have a 10ms delay before this
operation after resetting a given port.
But in original code, there is a possible path which may have no such
10ms delay:
UsbHubResetPort()->UsbHubSetPortFeature()->Stall(20)->UsbHubGetPortSt
atus()->XhcPollPortStatusChange()->(if RESET_C bit is set)->
XhcInitializeDeviceSlot()->(if RESET_C bit is set)->Stall(10)
So this patch is used to fix above issue.
Cc: Star Zeng <star.zeng@intel.com>
Cc: Baraneedharan Anbazhagan <anbazhagan@hp.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Tested-by: Baraneedharan Anbazhagan <anbazhagan@hp.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
When PciSioSerial is firstly started with a non-NULL remaining
device path, the UART instance is created using the parameters
specified in the remaining device path. Later when the driver
is started again on the same UART controller with NULL remaining
device path, the correct logic is to directly return SUCCESS
instead of current buggy implementation which wrongly produces
another UART using the default parameters.
The bug causes two UARTs are created when the UART is configured
in 57600 baud rate.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Cc: Eric Dong <eric.dong@intel.com>
Short Packet case is a normal case, we shouldn't print it as an error
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
The 'universal' PCI bus driver in MdeModulePkg contains a quirk to
degrade 64-bit PCI MMIO BARs to 32-bit in the presence of an option
ROM on the same PCI controller.
This quirk is highly specific to not just the X64 architecture in general,
but to the PC platform in particular, given that only X64 platforms that
require legacy PC BIOS compatibility require it. However, making the
quirk dependent on the presence of the legacy BIOS protocol met with
resistance, due to the fact that it introduces a dependency on the
IntelFrameworkModulePkg package.
So instead, make the quirk configurable, by introducing a feature flag PCD
'PcdPciDegradeResourceForOptionRom' which defaults to TRUE only for X64.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Some XHCI host controllers require to have extra 1ms delay before
accessing any MMIO register during HC reset.
As this delay is not defined by XHCI spec, we use this workaround
to fix the issue.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Some XHCI host controllers require to have extra 1ms delay before
accessing any MMIO register during HC reset.
As this delay is not defined by XHCI spec, we use this workaround
to fix the issue.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Currently, the EFI_PCI_ATTRIBUTE_DUAL_ADDRESS_CYCLE attribute is completely
ignored by the PCI host bridge driver, which means that, on an implementation
that supports DMA above 4 GB, allocations above 4 GB may be provided to
devices that have not expressed support for it.
So in addition to checking 'RootBridge->DmaAbove4G' to establish whether the
root bridge itself supports DMA above 4 GB, we must also take into account
the operation type (EfiPciOperationBusMaster{Read|Write|CommonBuffer}64),
and the EFI_PCI_ATTRIBUTE_DUAL_ADDRESS_CYCLE attribute, when mapping and
allocating DMA memory, respectively.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Ruiyu Ni <Ruiyu.ni@intel.com>
Since current codes in NvmExpressDxe already support the non-blocking I/O
feature for EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL interface, the relative bit
in the 'Attributes' field of EFI_NVM_EXPRESS_PASS_THRU_MODE should be set
to reflect this.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
This commit fixes the issue that the caller event passed to
EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.PassThru() will not be signaled for
NVME Admin commands.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to the UEFI spec, EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.PassThru()
should return EFI_INVALID_PARAMETER if the input 'NamespaceId' is invalid
for the NVM Express controller. This commit adds check in PassThru() to
follow this rule.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
This commit adds serveral checks for the 'Packet' parameter passed to the
EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.PassThru() API:
The check for the 'TransferLength' field in
EFI_NVM_EXPRESS_PASS_THRU_COMMAND_PACKET to make sure the value will not
exceed the maximum data transfer size allowed by a controller.
The check for the 'TransferBuffer' and 'TransferLength' fields in
EFI_NVM_EXPRESS_PASS_THRU_COMMAND_PACKET when the Opcode of an NVME
command indicates a data transfer between controller and host.
The check for the 'MetadataLength' field in
EFI_NVM_EXPRESS_PASS_THRU_COMMAND_PACKET to make sure the value is not 0
when the corresponding 'MetadataBuffer' field has a non-NULL value.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to UEFI spec, an EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL with neither
EFI_NVM_EXPRESS_PASS_THRU_ATTRIBUTES_LOGICAL nor
EFI_NVM_EXPRESS_PASS_THRU_ATTRIBUTES_PHYSICAL set in the Attributes field
is an illegal configuration.
This commit adds this check in the PassThru API to follow the spec.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to the UEFI spec, the 'TransferBuffer' and 'MetadataBuffer' used
in a data transfer should be aligned on the boundary specified by the
IoAlign field in the EFI_NVM_EXPRESS_PASS_THRU_MODE structure.
This commit adds this check to follow the spec.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to the UEFI spec,
EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.GetNextNamespace() should return
EFI_NOT_FOUND when the value pointed to by NamespaceId is the namespace ID
of the last namespace on the NVM Express controller. This commit modifies
the check for NamespaceId to follow this rule.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to the UEFI spec,
EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.GetNamespace() should return
EFI_NOT_FOUND when the input DevicePath is a device path node type that
the NVM Express Pass Thru driver supports, but there is not a valid
translation from DevicePath to a namespace ID. Current code will return
EFI_SUCCESS. This commit adds additional check in the GetNameSpace API to
make sure correct status is returned.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
According to the UEFI spec,
EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL.BuildDevicePath() should return
EFI_NOT_FOUND when the input NamespaceId is not valid. However, current
code returns EFI_DEVICE_ERROR instead. This commit modifies the check for
input NamespaceId to return the correct status.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
The gBS->OpenProtocol() calls to open EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL
in NvmExpress.c will crash the data in 'Mode' field of
'Private->Passthru'.
The third parameter of gBS->OpenProtocol() is an output parameter that
stores the address where a pointer to the corresponding Protocol
Interface. The current code mistakenly pass '&Private->Passthru' (a
pointer of the EFI_NVM_EXPRESS_PASS_THRU_PROTOCOL) as the third
parameter. This will crash the data in 'Mode' filed.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
The last TRB in transfer ring is a LINK type TRB, which shouldn't
be accounted as a valid item in IsAsyncIntTrb().
Without this fix, the original algo will bring issue on those URBs
whose TRBs crosses the transfer ring.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
> cc1: warnings being treated as errors
> MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c:
> In function 'GetProposedResources':
> MdeModulePkg/Bus/Pci/PciHostBridgeDxe/PciHostBridge.c:1388:
> error: integer constant is too large for 'long' type
Append a ULL suffix to the replacement text of PCI_RESOURCE_LESS that is
aligned with the style of the EFI_RESOURCE_SATISFIED and
EFI_RESOURCE_NOT_SATISFIED macros.
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Liming Gao <liming.gao@intel.com>
Section 7.8.2 of the PCI Express specification (r4.0 v0.3), entitled "PCI
Express Capabilities Register (Offset 02h)", and section 7.8.9 "Slot
Capabilities Register (Offset 14h)" of the same, describe the conditions
when a PCIe port should be considered "supporting hotplug":
- it should be a root complex port or a switch downstream port, and
- it should have the "Slot Implemented" bit set in the Express
Capabilities Register, and
- it should have the "Hot-Plug Capable" bit set in the Slot Capabilities
Register.
The first two sub-conditions are already implemented in at least two open
source projects I could find:
- in SeaBIOS by Marcel Apfelbaum: "hw/pci: reserve IO and mem for pci
express downstream ports with no devices attached"
<https://code.coreboot.org/p/seabios/source/commit/3aa31d7d6375>,
- in edk2 itself, in the implementation of the "PCI" UEFI Shell command:
see the "PcieExplainTypeSlot" case label in function
PciExplainPciExpress(), file
"ShellPkg/Library/UefiShellDebug1CommandsLib/Pci.c".
PciBusDxe recognizes such PCIe ports as bridges, but it doesn't realize
they support hotplug. In turn PciBusDxe omits getting any resource padding
information from the platform's EFI_PCI_HOT_PLUG_INIT_PROTOCOL for these
bridges:
GatherPpbInfo() [PciEnumeratorSupport.c]
GetResourcePaddingPpb() [PciResourceSupport.c]
GetResourcePaddingForHpb() [PciHotPlugSupport.c]
IsPciHotPlugBus() [PciHotPlugSupport.c]
//
// returns FALSE
//
//
// the following is not reached:
//
gPciHotPlugInit->GetResourcePadding()
Implement a function called SupportsPcieHotplug() for identifying such
ports, and call it from IsPciHotPlugBus() (after the call to IsSHPC()).
Cc: "Johnson, Brian J." <bjohnson@sgi.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Andrew Fish <afish@apple.com>
Cc: Feng Tian <feng.tian@intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Marcel Apfelbaum <marcel@redhat.com>
Cc: Michael Kinney <michael.d.kinney@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Ruiyu Ni <Ruiyu.ni@intel.com>
The PCI Hot Plug capability register block is marked with capability ID
0x0C (EFI_PCI_CAPABILITY_ID_SHPC), not 0x06
(EFI_PCI_CAPABILITY_ID_HOTPLUG).
This bug prevents PciBusDxe from recognizing whether a PCI-to-PCI bridge
supports hotplug. In turn the platform's EFI_PCI_HOT_PLUG_INIT_PROTOCOL is
not consulted for resource padding information:
GatherPpbInfo() [PciEnumeratorSupport.c]
GetResourcePaddingPpb() [PciResourceSupport.c]
GetResourcePaddingForHpb() [PciHotPlugSupport.c]
IsPciHotPlugBus() [PciHotPlugSupport.c]
IsSHPC() [PciHotPlugSupport.c]
//
// returns FALSE
//
//
// the following is not reached:
//
gPciHotPlugInit->GetResourcePadding()
Look for the correct capability ID.
Cc: "Johnson, Brian J." <bjohnson@sgi.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Andrew Fish <afish@apple.com>
Cc: Feng Tian <feng.tian@intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Marcel Apfelbaum <marcel@redhat.com>
Cc: Michael Kinney <michael.d.kinney@intel.com>
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Ruiyu Ni <Ruiyu.ni@intel.com>
- abstrated to abstracted
- accessibla to accessible
- addres to address
- apropriate to appropriate
- arry to array
- availabe to available
- avaliable to available
- becasue to because
- correponding to corresponding
- etablished to established
- exeuction to execution
- extensiable to extensible
- fileds to fields
- loadding to loading
- ptototypes to prototypes
- prococol protocol
- requried to required
- resoruce to resource
- runing to running
- uild to build
Cc: Star Zeng <star.zeng@intel.com>
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Giri P Mudusuru <giri.p.mudusuru@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
We have to upgrade the TPL level used by SdMmc stack because the
following flow:
DiskIo2ReadWriteDisk() in logical partition -> PartitionReadBlocksEx()
in logical partition at TPL callback level -> ProbeMediaStatusEx()
with sync request -> DiskIo2ReadWriteDisk() in physical partition ->
waiting for async task completion.
if the low layer driver doesn't run at TPL_NOTIFY level, it will have
no time to trigger async task and cause system hang.
Cc: Hao Wu <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Reviewed-by: Hao Wu <hao.a.wu@intel.com>
This delay is necessary for eMMC reset to working properly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Joe Zhou <shjzhou@marvell.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
The Maximum Queue Entries Supported (MQES) field in the CAP (Controller
Capabilities) register for a NVMe controller restrict the maximum
individual queue size that the controller supports.
The origin code does not check this value and always uses a hardcode value
when creating I/O submission/completion queues for asynchronous
transmission. The hardcode value might be larger than the MQES field, this
will lead to an 'Invalid Queue Size' error when creating I/O
submission/completion queues.
The patch will add checks to make sure proper queue size is passed when
creating I/O submission/completion queues.
Cc: Feng Tian <feng.tian@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@Intel.com>
Move NvmExpressHci.h definitions to a public industry standard
header in MdePkg. The NvmExpressHci.h contains definitions based
on the NVMe v1.1 specification. These definitions should be in
a public header so that clients of the NVMe passthru protocol
can use them to execute NVMe commands.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Darbin Reyes <darbin.emm.reyes@hpe.com>
Reviewed-by: Samer El-Haj-Mahmoud <samer.el-haj-mahmoud@hpe.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
The driver entry point calls gDS->SetMemorySpaceAttributes().
This interface may return EFI_NOT_AVAILABLE_YET when CPU Arch
protocol is not available.
So we need to list CpuArch protocol in its INF dependency section.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Liming Gao <liming.gao@intel.com>
PciBus driver originally always degrade (64->32) the MMIO resource
for PCI BAR when the PCI device contains option ROM.
But the degrade causes the PCI device can only use resource below 4GB
which makes the resource allocation fails when the PCI device wants
very big MMIO.
The patch follows the PI spec (ECR 1529) to honor the granularity
setting for PCI BAR from IncompatiblePciDeviceSupport so that even
for PCI device which contains option ROM, the degrade doesn't happen
if IncompatiblePciDeviceSupport returns 64 as granularity.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Remove "Efi" from gEfiIncompatiblePciDeviceSupport to shorten
the global variable name.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Jeff Fan <jeff.fan@intel.com>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Together with EFI_BLOCK_IO_PROTOCOL, EFI_BLOCK_IO2_PROTOCOL is also
produced on NVMe devices.
The following Block I/O 2 functions are implemented:
Reset
ReadBlocksEx
WriteBlocksEx
FlushBlocksEx
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
When PcdPciDisableBusEnumeration is enabled, the PciBus driver
might get into a dead loop if the secondary bus register on PCI
bridge is not programmed or programmed improperly. Adding this
check to avoid any potential dead loop caused by this.
Cc: Feng Tian <feng.tian@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Maurice Ma <maurice.ma@intel.com>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Lee Leahy <Leroy.p.leahy@intel.com>
Change PciHostBridgeDxe driver to not install the
PciHostBridgeResourceAllocation protocol and let
PciRootBridgeIo.Configuration() return the correct PCI resource
assignment information when the ResourceAssigned is TRUE.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
When the aperture base equals to aperture limit, the old code treats
the aperture as non-existent. It's not correct because it indicates
a range starting with base and the length is 1.
The new code corrects the comparing bug.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Cc: Jeff Fan <jeff.fan@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
The patch aligns to the IntelFrameworkModulePkg/Bus/Isa/IsaSerialDxe
driver not flush the UART in Reset() and SetAttributes() function.
It was found the flush causes hang on certain PCI serial devices.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ruiyu Ni <ruiyu.ni@intel.com>
Reviewed-by: Eric Jin <eric.jin@intel.com>
Before switch to a bus mode, we need check if the SD device supports
this bus mode.
Cc: Wu, Hao A <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
The original code is using ADMA mode to do clock tuning procedure. It
may have problem on some SD/MMC host controllers as there is no way to
know when to send next tuning cmd.
Update it to PIO mode to strictly follow SD Host Controller Simplified
Specification 3.0 Figure 2-29. By this way, if the Buffer Read Ready
interrupt is set, we could know it's ok to send the next clock tuning
cmd.
Cc: Wu, Hao A <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
The original code has a bug to calculate which clock freq should be
used when the target clock freq is larger than the BaseClock Freq
provided by the system.
Cc: Wu, Hao A <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
Check if the card is identified/initialized correctly. if not, break
the following cmd execution through PassThru()/ResetDevice().
Cc: Wu, Hao A <hao.a.wu@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Feng Tian <feng.tian@intel.com>
The "IoAlign" field in EFI_BLOCK_IO_MEDIA of an NVMe device is not
initialized properly, leading to a zero value for this field.
It should be initialized from the "IoAlign" field in the
EFI_NVM_EXPRESS_PASS_THRU_MODE structure maintained by the NVMe
controller.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Feng Tian <feng.tian@intel.com>
If the current PCI configuration requires no resources to be allocated at
all (i.e., unpopulated bus), the PCI enumeration code creates a single
ACPI_ADDRESS_SPACE_DESCRIPTOR memory descriptor with all fields cleared.
This is rejected by the SubmitResources() implementation of the generic
PciHostBridgeDxe in the following way:
PciHostBridge: SubmitResources for PcieRoot(0x0)
Mem: Granularity/SpecificFlag = 0 / 00
Length/Alignment = 0x0 / 0x0
PciBus: HostBridge->SubmitResources() - Invalid Parameter
ASSERT_EFI_ERROR (Status = Invalid Parameter)
ASSERT [PciBusDxe] .../PciBusDxe/PciLib.c(561): !EFI_ERROR (Status)
So instead, create the empty configuration as a single entry of type
EFI_ACPI_END_TAG_DESCRIPTOR.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
Star Zeng
Liming Gao
Hao Wu
Sami Mujawar
Ruiyu Ni
Heyi Guo
Dandan Bi
Ard Biesheuvel
Jiewen Yao
Bi, Dandan
Suman Prakash
Feng Tian
Thomas Huth
Laszlo Ersek
Gary Lin
Giri P Mudusuru
Joe Zhou
Darbin Reyes
Ma, Maurice