This patch is to improve IOMMU performance.
All WBINVD is removed due to performance issue.
CLFLUSH by WriteBackDataCacheRange() is used to
only flush the context table or
second level page table if they are changed.
This patch also removed some unused functions.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 4ad5f59715)
There are VS2015 NOOPT IA32 build failure like below in IntelVTdDxe.
XXX.lib(XXX.obj) : error LNK2001: unresolved external symbol __allshl
XXX.lib(XXX.obj) : error LNK2001: unresolved external symbol __aullshr
This patch is to update Vtd.h to use UINT32 instead of UINT64 for
bitfields in structure definition, and also update IntelVTdDxe code
accordingly.
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Star Zeng <star.zeng@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
(cherry picked from commit 76c6f69cca)
Add explicit NULL pointer check to make the codes more straight-forward.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Hao Wu <hao.a.wu@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
(cherry picked from commit be61fcd2b0)
It provides sample on Platform VTd policy protocol.
This protocol is optional.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 5071fb9cd9)
It provides AllocateBuffer/FreeBuffer/Map/Unmap function.
It also provides VTd capability yet.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
(cherry picked from commit c049fc9909)
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>
(cherry picked from commit a012bf6e3e)
Install IOMMU PPI for pre-memory phase and return
EFI_NOT_AVAILABLE_YET to indicate that DMA protection has been enabled,
but DMA buffer are not available to be allocated yet.
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>
(cherry picked from commit 20b58eb850)
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>
(cherry picked from commit 748cd9a680)
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>
(cherry picked from commit 42e8bc7d16)
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>
(cherry picked from commit 2632981783)
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>
(cherry picked from commit c74805f1e7)
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>
(cherry picked from commit 8284b1791e)
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>
(cherry picked from commit c34a5aab53)
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>
(cherry picked from commit 0aa1794118)
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>
(cherry picked from commit 2c656af04d)
Update the SdBlockIoPei 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: 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: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 77af86688c)
Update the EmmcBlockIoPei 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: 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: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 85ad9a6e0a)
V2 changes:
Resource cleanup logic update in UfsEndOfPei().
V1 history:
Update the UfsBlockIoPei 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: 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: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 44a0857ec6)
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>
(cherry picked from commit b575ca32c8)
This IOMMU_PPI is to provide IOMMU abstraction in PEI.
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>
(cherry picked from commit 2b620ee1ff)
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>
(cherry picked from commit 0b9c0c6540)
This protocol is to abstract DMA access from IOMMU.
1) Intel "DMAR" ACPI table.
2) AMD "IVRS" ACPI table
3) ARM "IORT" ACPI table.
There might be multiple IOMMU engines on one platform.
For example, one for graphic and one for rest PCI devices
(such as ATA/USB).
All IOMMU engines are reported by one ACPI table.
All IOMMU protocol provider should be based upon ACPI table.
This single IOMMU protocol can handle multiple IOMMU engines on one system.
This IOMMU protocol provider can use UEFI device path to distinguish
if the device is graphic or ATA/USB, and find out corresponding
IOMMU engine.
The IOMMU protocol provides 2 capabilities:
A) Set DMA access attribute - such as write/read control.
B) Remap DMA memory - such as remap above 4GiB system memory address
to below 4GiB device address.
It provides AllocateBuffer/FreeBuffer/Map/Unmap for DMA memory.
The remapping can be static (fixed at build time) or dynamic (allocate
at runtime).
4) AMD "SEV" feature.
We can have an AMD SEV specific IOMMU driver to produce IOMMU protocol,
and manage SEV bit.
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>
(cherry picked from commit d1fddc4533)
sha256 is not the standard option. It should be replaced by sha -sha256.
Otherwise, it doesn't work in MAC OS.
In V2, update the option to sha1 -sha256.
In late openssl version >= 1.1, there is no sha option, but has sha1,sha256.
In previous openssl version < 1.1, there is no sha256, but has sha,sha1.
To work with all openssl version, use sha1 -sha256 for it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liao Jui-peng <jui-pengx.liao@intel.com>
Signed-off-by: Liming Gao <liming.gao@intel.com>
Cc: Michael Kinney <michael.d.kinney@intel.com>
Cc: Yonghong Zhu <yonghong.zhu@intel.com>
Reviewed-by: Yonghong Zhu <yonghong.zhu@intel.com>
(cherry picked from commit 1d574dfc15)
Add empty TempRamInitApi function to fix
build error with WHOLEARCHIVE option
Cc: Jiewen Yao <jiewen.yao@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Bell Song <binx.song@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
(cherry picked from commit c69071bd7e)
XCODE5 doesn't support absolute addressing in the assembly code.
This change uses lea instruction to get the address.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liming Gao <liming.gao@intel.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
(cherry picked from commit c45f4c5e75)
Ovmf appended option -mno-mmx -mno-sse, but these two options were enabled
in Openssl. The compiler option becomes -mmmx ?msse -mno-mmx -mno-sse. It
trig mac clang compiler hang when compile one source file in openssl.
This issue is found when SECURE_BOOT_ENABLE is TRUE. This may be the compiler
issue. To work around it, don't add these two options for XCODE5 tool chain.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liming Gao <liming.gao@intel.com>
Cc: Andrew Fish <afish@apple.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
(cherry picked from commit 4a64cbda86)
https://bugzilla.tianocore.org/show_bug.cgi?id=849
In V2, use "mov rax, strict qword 0" to replace the hard code db.
1. Use lea instruction to get the address instead of mov instruction.
2. Use the dummy address as jmp destination, and add the logic to fix up
the address to the absolute address at boot time.
3. On MpFuncs.nasm, use ExchangeInfo to record InitializeFloatingPointUnits.
This way is same to MpInitLib.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liming Gao <liming.gao@intel.com>
Cc: Andrew Fish <afish@apple.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Eric Dong <eric.dong@intel.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Cc: Michael Kinney <michael.d.kinney@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
(cherry picked from commit e21e355e2c)
# Conflicts:
# UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.h
XCODE5 doesn't support absolute addressing in the assembly code.
This change uses lea instruction to get the address.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Liming Gao <liming.gao@intel.com>
Cc: Andrew Fish <afish@apple.com>
Cc: Star Zeng <star.zeng@intel.com>
Reviewed-by: Star Zeng <star.zeng@intel.com>
(cherry picked from commit 62382925c9)
