REF:https://bugzilla.tianocore.org/show_bug.cgi?id=4578
The implementation of XhcGetElapsedTicks did not account for
non-zero start and stop values for the performance counter
timer, potentially resulting in an incorrect elapsed tick
count getting returned to the caller. Account for non-zero
start and stop values when calculating the elapsed tick
count.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Signed-off-by: Patrick Henz <patrick.henz@hpe.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Hao A Wu <hao.a.wu@intel.com>
Message-ID: <c3038878c4d30c54e60cce7192cf1aa60c30ad2e.1698770394.git.patrick.henz@hpe.com>
Resolve rebase conflict from commit ff4c49a5ee ("MdeModulePkg/Bus: Fix
XhciDxe Linker Issues", 2023-12-06): rename "mPerformanceCounter*" to
"mXhciPerformanceCounter*".
Signed-off-by: Laszlo Ersek <laszlo.ersek@posteo.net>
The issue with locating the expected interface and endpoint descriptors
arises because `configDesc` (USB_CONFIG_DESCRIPTOR) and `IfDesc`
(USB_INTERFACE_DESCRIPTOR) are incremented by structure size rather than
by actual descriptor length.
Specifically:
- `configDesc` should be incremented by its actual length.
- `IfDesc` should be incremented by its actual length.
This incorrect increment causes misalignment, preventing access to the
subsequent interface and endpoint descriptors.
[Suggested Solution]
Update the code to increment the pointers by the actual descriptor lengths,
ensuring proper access to all descriptors in the USB configuration.
Signed-off-by: Aniket Surekar <Aniket.Surekar@Dell.com>
REF:https://bugzilla.tianocore.org/show_bug.cgi?id=4858
Ph52a PCIE to SATA card inserted on Intel MTL/ARL causes system hanged.
Root cause of this issue is because Ph52a's driver only uses DevicePath
protocol alone and EDK2 driver only uses PciIo protocol alone. Both
drivers start and try to manage SATA controller.
Signed-off-by: Paul Chang <paulchang@ami.com>
On AARCH64, there is no ordering guarantee between configuration
space (ECAM) writes and memory space reads (MMIO). ARM AMBA CHI
only guarantees ordering for reads and writes within a single address
region, however, on some systems MMIO and ECAM may be split into
separateaddress regions.
A problem may arise when an ECAM write is issued a completion before a
subsequent MMIO read is issued and receives a completion.
For example, a typical PCI software flow is the following:
1. ECAM write to device command register to enable memory space
2. MMIO read from device memory space for which access was enabled
in step 1.
There is no guarantee that step 2. will not begin before the completion
of step 1.
on systems where ECAM/MMIO are specified as separate address regions,
even
if both spaces have the memory attributes device-nGnRnE.
- Add a barrier after the final PCI Configuration space write
in RootBridgeIoPciAccess. Configuration space reads should not have
side-efects.
- When configuration space is strongly ordered, this ensures
that program execution cannot continue until the completion
is received for the previous Cfg-Write, which may have side-effects.
- Risk of reading a "write-only" register and causing a CA which leaves
the device unresponsive. The expectation based on the PCI Base Spec
v6.1 section 7.4 is that all PCI Spec-defined registers will be readable,
however, there may exist design-specific registers that fall into
this category.
Signed-off-by: Aaron Pop <aaronpop@microsoft.com>
Co-authored-by: Ard Biesheuvel <ardb@kernel.org>
Implementation of MEDIA_SANITIZE_PROTOCOL for NIST
purge/clear actions with mapping to NVM Express native
commands.
Signed-off-by: Aaron Pop <aaronpop@microsoft.com>
When allocating memory for a non-discoverable PCI device's IO, the
current core code removes the XP attribute, allowing code to execute
from that region. This is a security vulnerability and unneeded. This
change updates to mark the region as XP when allocating memory for the
non-discoverable PCI device.
These allocations in this function are limited to `EfiBootServicesData`
and `EfiRuntimeServicesData`, which we expect to be XP.
Signed-off-by: Aaron Pop <aaronpop@microsoft.com>
Add Report status code events to Nvme to trigger when timeouts occur.
This improves traceability when critical errors happen.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
Fire an event if a long delay occurs when starting an NVMe device.
This can be used by platforms to draw pictures on the screen or
take other actions to notify a user or move boot forward.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
Eui64 is a 64 bit value, so the "L" or "l" is required for
format specifier, otherwise only lower 32 bit will be converted.
Signed-off-by: Mike Maslenkin <mike.maslenkin@gmail.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
https://bugzilla.tianocore.org/show_bug.cgi?id=4727
Recently some of XHCI host controllers require to have
extra 1ms delay before accessing any MMIO register
during reset. PHY transition from P3 to P0 can take
around 1.3ms and the xHCI reset can take around 1.5ms.
Add PCD to control the delay, the default is 2 ms.
Cc: Ray Ni <ray.ni@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Krzysztof Lewandowski <krzysztof.lewandowski@intel.com>
Cc: Jenny Huang <jenny.huang@intel.com>
Cc: More Shih <more.shih@intel.com>
Cc: Ian Chiu <ian.chiu@intel.com>
Signed-off-by: Xianglei Cai <xianglei.cai@intel.com>
Reviewed-by: Krzysztof Lewandowski <krzysztof.lewandowski@intel.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
https://bugzilla.tianocore.org/show_bug.cgi?id=4556
Based on XHCI spec 4.8.3, software should do the
reset endpoint while USB Transaction occur.
Add the error code for USB Transaction error
since UEFI spec don't have the related definition.
Cc: Ray Ni <ray.ni@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Krzysztof Lewandowski <krzysztof.lewandowski@intel.com>
Cc: Jenny Huang <jenny.huang@intel.com>
Cc: More Shih <more.shih@intel.com>
Signed-off-by: Xianglei Cai <xianglei.cai@intel.com>
Reviewed-by: Krzysztof Lewandowski <krzysztof.lewandowski@intel.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
REF:https://bugzilla.tianocore.org/show_bug.cgi?id=4560
Commit f36e1ec1f0 had fixed the DXE_ASSERT
caused by the TRB size round up from 16 to 64 for most cases.
However, there is a remaining case that the TRB size is also rounded up
during setting TR dequeue pointer that would trigger DXE_ASSERT.
This patch sets the alignment flag to FALSE in XhcSetTrDequeuePointer to
fix this issue as well.
Cc: Gao Cheng <gao.cheng@intel.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Signed-off-by: Dat Mach <dmach@nvidia.com>
Reviewed-by: Gao Cheng <gao.cheng@intel.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
AtaBusDxe, NvmExpressDxe, ScsiDiskDxe and EmmcDxe is used to back the
EFI_STORAGE_SECURITY_COMMAND_PROTOCOL, update the parameter 'MediaId'
description for the protocol function ReceiveData and SendData as
described in UEFI Spec 2.10 section 13.14.
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Signed-off-by: Qingyu Shang <qingyu.shang@intel.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
PciIoMap () need to feedback the status of
mIoMmuProtocol->SetAttribute () return value.
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4652
Reviewed-by: Ray Ni <ray.ni@intel.com>
Reviewed-by: Huang Jenny <jenny.huang@intel.com>
Cc: Chiang Chris <chris.chiang@intel.com>
Signed-off-by: Sheng Wei <w.sheng@intel.com>
The DXE & MM standalone variant of AcpiTimerLib defines a global
named mPerformanceCounterFrequency. A global with an identical
name is also present in MdeModulePkg/Bus/Pci/XhciDxe/Xhci.c
Since XhciDxe has a dependency on TimerLib, this can cause link
errors due to the same symbol being defined twice if the platform
DSC chooses to use AcpiTimerLib as the TimerLib implementation for
any given platform.
To resolve this, I noted that some of the globals in Xhci.c are not
used outside of the Xhci.c compilation unit:
- mPerformanceCounterStartValue
- mPerformanceCounterEndValue
- mPerformanceCounterFrequency
- mPerformanceCounterValuesCached
I have changed the definition for all of these to static and added
an Xhci prefix. Since they are not used outside of the Xhci.c
compilation unit, there is no reason to have them exported as
globals.
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Signed-off-by: Nate DeSimone <nathaniel.l.desimone@intel.com>
REF:https://bugzilla.tianocore.org/show_bug.cgi?id=4560
TRB Template is 16 bytes. When boundary checking is 64 bytes for xHCI
device/host memory address, it may exceed xHCI host memory pool and
cause unwanted DXE_ASSERT. Introduce a new input parameter to indicate
whether to enforce 64byte size alignment and round up. For TRB case,
should set it to FALSE to skip the size round up.
Signed-off-by: Gao Cheng <gao.cheng@intel.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
https://bugzilla.tianocore.org/show_bug.cgi?id=4552
Following XHCI spec 4.6.1.2, software may abort the
execution of Address Device Command when command failed
due to timeout.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: More Shih <more.shih@intel.com>
Cc: Jenny Huang <jenny.huang@intel.com>
Signed-off-by: Xianglei Cai <xianglei.cai@intel.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
REF:https://bugzilla.tianocore.org/show_bug.cgi?id=2948
XhciDxe uses the timer functionality provided by the
boot services table to detect timeout conditions. This
breaks the driver's ExitBootServices call back, as
CoreExitBootServices halts the timer before signaling
the ExitBootServices event. If the host controller
fails to halt in the call back, the timeout condition
will never occur and the boot gets stuck in an indefinite
spin loop. Use the free running timer provided by
TimerLib to calculate timeouts, avoiding the potential
hang.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Signed-off-by: Patrick Henz <patrick.henz@hpe.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
The function UsbHcGetPciAddressForHostMem has
ASSERT ((Block != NULL)); and
and the function UsbHcFreeMem has
ASSERT (Block != NULL);
statement after for loop, but these are applicable only in DEBUG mode.
In RELEASE mode, if for whatever reasons there is no match inside for
loop and the loop exits because of Block != NULL; condition, then there
is no "Block" NULL pointer check afterwards and the code proceeds to do
dereferencing "Block" which will lead to CRASH.
Hence, for safety add NULL pointer checks always.
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4211
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Co-authored-by: Veeresh Sangolli <veeresh.sangolli@dellteam.com>
Signed-off-by: Ranbir Singh <Ranbir.Singh3@Dell.com>
Signed-off-by: Ranbir Singh <rsingh@ventanamicro.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
The function UhciConvertPollRate has a check
ASSERT (Interval != 0);
but this comes into play only in DEBUG mode. In Release mode, there is
no handling if the Interval parameter value is ZERO. To avoid shifting
by a negative amount later in the code flow in this undesirable case,
it is better to handle it as well by treating it same as if 1 is sent.
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4211
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Co-authored-by: Veeresh Sangolli <veeresh.sangolli@dellteam.com>
Signed-off-by: Ranbir Singh <Ranbir.Singh3@Dell.com>
Signed-off-by: Ranbir Singh <rsingh@ventanamicro.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
The code can reach line 65 only through the else path above at line 53.
The else path already has the same NULL check at line 55 and hence the
duplicate code lines are totally redundant which can be deleted.
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4220
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Co-authored-by: Veeresh Sangolli <veeresh.sangolli@dellteam.com>
Signed-off-by: Ranbir Singh <Ranbir.Singh3@Dell.com>
Signed-off-by: Ranbir Singh <rsingh@ventanamicro.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
A faulty PCI device has the Option ROM image size set to 0. UEFI reads
two headers PCI_EXPANSION_ROM_HEADER and PCI_DATA_STRUCTURE to get the
Option ROM information. Because the image size is 0, the Option ROM
header address never changes. As a result, UEFI keeps reading the same
two headers definitely. This patch is intended to fix it.
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Signed-off-by: Nhi Pham <nhi@os.amperecomputing.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
The function UsbHcGetPciAddressForHostMem has
ASSERT ((Block != NULL));
and the UsbHcFreeMem has
ASSERT (Block != NULL);
statement after for loop, but these are applicable only in DEBUG mode.
In RELEASE mode, if for whatever reasons there is no match inside the
for loop and the loop exits because of Block != NULL; condition, then
there is no "Block" NULL pointer check afterwards and the code proceeds
to do dereferencing "Block" which will lead to CRASH.
Hence, for safety add NULL pointer checks always.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4210
Signed-off-by: Ranbir Singh <Ranbir.Singh3@Dell.com>
Signed-off-by: Ranbir Singh <rsingh@ventanamicro.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
ASSERT (Private != NULL) (where Private = CR(...)) is ineffective as
CR(Ptr, Type, Member, Sig) either returns Ptr - offsetof(Type, Member),
or ASSERTS on the signature, so it's unlikely to ever return NULL (must
be passed a pointer = member's offset, or in this case, 0x4).
ASSERT on This != NULL instead.
Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
When a UEFI_DRIVER attempts to open a protocol interface with BY_DRIVER
attribute that it already has open with BY_DRIVER attribute,
OpenProtocol() returns EFI_ALREADY_STARTED. This is not an error. The
UEFI-2.7 spec currently says,
> EFI_ALREADY_STARTED -- Attributes is BY_DRIVER and there is an item on
> the open list with an attribute of BY_DRIVER
> whose agent handle is the same as AgentHandle.
Downgrade the log mask for this one condition to DEBUG_INFO, in
SataControllerStart(). This will match the log mask of the other two
informative messages in this function.
(ported from commit 5dfba97)
Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=4394
Fix DumpCapabilityReg() debug log to print 64 bit capability instead of
32 bit pointer
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Andrew Fish <afish@apple.com>
Signed-off-by: Giri Mudusuru <girim@apple.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
This applies the volatile keyword and appropriate casts
to the NvmExpressPei's Passthru CQ.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Michael Kubacki <mikuback@linux.microsoft.com>
Cc: Sean Brogan <sean.brogan@microsoft.com>
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
Acked-by: Michael Kubacki <michael.kubacki@microsoft.com>
This updates the relevant functions that expect a non-volatile
structure to be passed to them to take casts of the CQ
now that it is volatile.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Michael Kubacki <mikuback@linux.microsoft.com>
Cc: Sean Brogan <sean.brogan@microsoft.com>
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
Acked-by: Michael Kubacki <michael.kubacki@microsoft.com>
__FUNCTION__ is a pre-standard extension that gcc and Visual C++ among
others support, while __func__ was standardized in C99.
Since it's more standard, replace __FUNCTION__ with __func__ throughout
MdeModulePkg.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
Reviewed-by: Michael D Kinney <michael.d.kinney@intel.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
In V4: Update the copyright to 2023.
In V3: Add AMD copyright.
In V2: Remove the signed-off-by: Abner Chang
Display PCIe Vendor ID and Device ID in DEBUG message.
Signed-off-by: Jiangang He <jiangang.he@amd.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Garrett Kirkendall <garrett.kirkendall@amd.com>
Cc: Abner Chang <abner.chang@amd.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Acked-by: Michael D Kinney <michael.d.kinney@intel.com>
Unlink the XhciPei memory block when it has been freed.
Signed-off-by: Jiangang He <jiangang.he@amd.com>
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Garrett Kirkendall <garrett.kirkendall@amd.com>
Cc: Abner Chang <abner.chang@amd.com>
Cc: Kuei-Hung Lin <Kuei-Hung.Lin@amd.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
Force resetting the port by clearing the USB_PORT_STAT_C_RESET bit in
PortChangeStatus when XhcPollPortStatusChange fails
Signed-off-by: Sean Rhodes <sean@starlabs.systems>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
On some platforms, including Sky Lake and Kaby Lake, the PSIV (Protocol
Speed ID Value) indices are shared between Protocol Speed ID DWORD' in
the extended capabilities registers for both USB2 (Full Speed) and USB3
(Super Speed).
An example can be found below:
XhcCheckUsbPortSpeedUsedPsic: checking for USB2 ext caps
XhciPsivGetPsid: found 3 PSID entries
XhciPsivGetPsid: looking for port speed 1
XhciPsivGetPsid: PSIV 1 PSIE 2 PLT 0 PSIM 12
XhciPsivGetPsid: PSIV 2 PSIE 1 PLT 0 PSIM 1500
XhciPsivGetPsid: PSIV 3 PSIE 2 PLT 0 PSIM 480
XhcCheckUsbPortSpeedUsedPsic: checking for USB3 ext caps
XhciPsivGetPsid: found 3 PSID entries
XhciPsivGetPsid: looking for port speed 1
XhciPsivGetPsid: PSIV 1 PSIE 3 PLT 0 PSIM 5
XhciPsivGetPsid: PSIV 2 PSIE 3 PLT 0 PSIM 10
XhciPsivGetPsid: PSIV 34 PSIE 2 PLT 0 PSIM 1248
The result is edk2 detecting USB2 devices as USB3 devices, which
consequently causes enumeration to fail.
To avoid incorrect detection, check the Compatible Port Offset to find
the starting Port of Root Hubs that support the protocol.
Signed-off-by: Sean Rhodes <sean@starlabs.systems>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
PSID matching relies on comparing the PSIV against the PortSpeed
value. This patch stops edk2 from checking for a PSIV of 0, as it
is not valid; this reduces the number of register access by
approximately 6 per second.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Sean Rhodes <sean@starlabs.systems>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
SD1.0 cards don't support CMD8 and CMD6
CMD8 result can be used to distinguish the card is SD1.0 or not.
CMD8 result can be used to decide following CMD6 is sent or skip.
Cc: Hao A Wu <hao.a.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Signed-off-by: Chevron Li <chevron.li@bayhubtech.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
With the pending commit of UsbNetworkPkg, it will become common for
UsbBulkTransfer calls to timeout, given that the drivers are called from
MnpSystemPoll every MNP_SYS_POLL_INTERVAL milliseconds: the drivers
check for network packets by calling UsbBulkTransfer with a timeout of
1ms.
Avoid console spam by moving DEBUG messages that occur each time a bulk
transfer request times out from DEBUG_ERROR to DEBUG_VERBOSE, for both
EHCI and XHCI drivers.
Signed-off-by: Rebecca Cran <rebecca@quicinc.com>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
gSerialDevTempate should be gSerialDevTemplate
Cc: Ray Ni <ray.ni@intel.com>
Cc: Zhichao Gao <zhichao.gao@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Signed-off-by: Nate DeSimone <nathaniel.l.desimone@intel.com>
Reviewed-by: Michael D Kinney <michael.d.kinney@intel.com>
According the Xhci Spec, TRB Rings may be larger than a Page, however they
shall not cross a 64K byte boundary, so add a parameter to indicate
whether the memory allocation is for TRB Rings or not. It will ensure the
allocation not crossing 64K boundary in UsbHcAllocMemFromBlock if the
memory is allocated for TRB Rings.
Signed-off-by: jdzhang <jdzhang@kunluntech.com.cn>
Reviewed-by: Hao A Wu <hao.a.wu@intel.com>
If a device which support both features SR-IOV/ARI has multi
functions, which maybe support 8-255. After enable ARI forwarding in
the root port and ARI Capable Hierarchy in the SR-IOV PF0.
The device will support and expose multi functions(0-255) with ARI ID routing.
In next device loop in below for() code, actually it still be in the
same SR-IOV device, and just some PF which is over 8 or higher
one(n*8), PciAllocateBusNumber() will allocate bus
number(ReservedBusNum - TempReservedBusNum)) for this PF. if reset
TempReservedBusNum as 0 in this case,it will allocate wrong bus number
for this PF because TempReservedBusNum should be total previous PF's
reserved bus numbers.
code:
for (Device = 0; Device <= PCI_MAX_DEVICE; Device++) {
TempReservedBusNum = 0;
for (Func = 0; Func <= PCI_MAX_FUNC; Func++) {
//
// Check to see whether a pci device is present
//
Status = PciDevicePresent (
PciRootBridgeIo,
&Pci,
StartBusNumber,
Device,
Func
);
...
Status = PciAllocateBusNumber (PciDevice, *SubBusNumber,
(UINT8)(PciDevice->ReservedBusNum - TempReservedBusNum), SubBusNumber);
The solution is add a new flag IsAriEnabled to help handle this case.
if ARI is enabled, then TempReservedBusNum will not be reset again
during all functions(1-255) scan with checking flag IsAriEnabled.
Signed-off-by: Foster Nong <foster.nong@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Below code will calculate the reserved bus number for the each PF.
Based on the VF routing ID algorithm, PFRid and LastVF in below code
already sure that "All VFs and PFs must have distinct Routing IDs".
PF will be assigned Routing ID based on secBusNumber, ReservedBusNum
will add into SubBusNumber directly. So the SR-IOV device will be
assigned bus range as SecBusNumber ~ (SubBusNumber=(SecBusNumber +
ReservedBusNum)).
Thus "+1" in below code will cause extra 1 bus, and introduce a bus hole.
PFRid = EFI_PCI_RID (Bus, Device, Func);
LastVF = PFRid + FirstVFOffset + (PciIoDevice->InitialVFs - 1) * VFStride;
PciIoDevice->ReservedBusNum = (UINT16)(EFI_PCI_BUS_OF_RID (LastVF) -
Bus + 1);
In SR-IOV spec, there is a note in section 2.1.2:
Note: Bus Numbers are a constrained resource. Devices are strongly
encouraged to avoid leaving ?holes? in their Bus Number usage to avoid
wasting Bus Numbers
So the issue can be fixed with below code change.
PciIoDevice->ReservedBusNum = (UINT16)(EFI_PCI_BUS_OF_RID (LastVF) -
Bus);
https://bugzilla.tianocore.org/show_bug.cgi?id=4069
Signed-off-by: Foster Nong <foster.nong@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Per the section 3.3.5 SR-IOV spec v1.1, InitialVFs (0ch).
InitialVFs indicates to SR-PCIM the number of VFs that are initially associated with the PF.
The minimum value of InitialVFs is 0.
Below code is used to calculate SR-IOV reserved bus number,
if InitialVFs =0, it maybe calculate the wrong bus number in this case.
LastVF = PFRid + FirstVFOffset + (PciIoDevice->InitialVFs - 1) * VFStride
we can fix it with below code:
if (PciIoDevice->InitialVFs == 0) {
PciIoDevice->ReservedBusNum = 0;
} else {
PFRid = EFI_PCI_RID (Bus, Device, Func);
LastVF = PFRid + FirstVFOffset + (PciIoDevice->InitialVFs - 1) * VFStride;
//
// Calculate ReservedBusNum for this PF
//
PciIoDevice->ReservedBusNum = (UINT16)(EFI_PCI_BUS_OF_RID (LastVF) - Bus + 1);
//
// Calculate ReservedBusNum for this PF
//
PciIoDevice->ReservedBusNum = (UINT16)(EFI_PCI_BUS_OF_RID (LastVF) - Bus + 1);
}
https://bugzilla.tianocore.org/show_bug.cgi?id=4069
Signed-off-by: Foster Nong <foster.nong@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Add support for partial free of non cached buffers.
If a request for less than the full size is requested new allocations
for the remaining head and tail of the buffer are added to the list.
Added verification that Buffer is EFI_PAGE_SIZE aligned.
The XHCI driver does this if the page size for the controller is >4KB.
Signed-off-by: Jeff Brasen <jbrasen@nvidia.com>
Acked-by: Liming Gao <gaoliming@byosoft.com.cn>
Henz, Patrick
Aniket_Surekar
Paul Chang
joe
Rebecca Cran
Ray Robles
Oliver Smith-Denny
Sean Brogan
Mike Maslenkin
Xianglei Cai
Dat Mach
Qingyu Shang
Sheng Wei
Nate DeSimone
Gao Cheng
Ranbir Singh
Nhi Pham
likun su
Pedro Falcato
Giri Mudusuru
Oliver Smith-Denny
Rebecca Cran
Michael Kubacki
Abner Chang
Sean Rhodes
Matt DeVillier
Chevron Li (WH)
Rebecca Cran
jdzhang
Foster Nong
Jeff Brasen via groups.io