Following the discussion at [1] and as the ArmLib relies on them,
move ArmPkg/Chipset/ArmV7[|Mmu].h files to the MdePkg.
Update the path to correctly include the moved files.
[1] https://edk2.groups.io/g/devel/message/111566
Continuous-integration-options: PatchCheck.ignore-multi-package
Signed-off-by: Pierre Gondois <pierre.gondois@arm.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Related to https://bugzilla.tianocore.org/show_bug.cgi?id=4121, but not
resolving it. (Nearly?) all of ArmPkg describes industry standard
behaviour, and hence according to general rules, ought to live in MdePkg.
Addressing this will however be a substantial task.
Take a first step by moving the ArmLib interface definition to MdePkg,
as discussed in
https://edk2.groups.io/g/devel/topic/patch_v5_2_6/102725178
Continuous-integration-options: PatchCheck.ignore-multi-package
Cc: Pierre Gondois <pierre.gondois@arm.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Cc: Sami Mujawar <sami.mujawar@arm.com>
Cc: Zhiguang Liu <zhiguang.liu@intel.com>
Signed-off-by: Pierre Gondois <pierre.gondois@arm.com>
Set the rdrand feature bit when faking cpuid for host test cases.
Needed to make the CryptoPkg test cases work.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
PcdCpuSmmApSyncTimeout2 PCD was added in previous patch
(52d0a208), this patch is to override PcdCpuSmmApSyncTimeout2
to 10ms (same as PcdCpuSmmApSyncTimeout) so as to align with
original behavior.
Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Ray Ni <ray.ni@intel.com>
Signed-off-by: Jiaxin Wu <jiaxin.wu@intel.com>
PcdCpuSmmApSyncTimeout is not only used by BSP to wait AP,
but also for AP to wait BSP (APHandler).
This patch is only to refine the PCD comment. No function
impact.
Signed-off-by: Jiaxin Wu <jiaxin.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
This patch is to consume the PcdCpuSmmApSyncTimeout2 to
enhance the flexibility of timeout configuration.
In some cases, certain processors may not be able to enter
SMI, and prolonged waiting could lead to kernel soft/hard
lockup. We have now defined two timeouts. The first timeout
can be set to a smaller value to reduce the waiting period.
Processors that are unable to enter SMI will be woken up
through SMIIPL to enter SMI, followed by a second waiting
period. The second timeout can be set to a larger value to
prevent delays in processors entering SMI case due to the
long instruction execution.
Signed-off-by: Jiaxin Wu <jiaxin.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Provide the capability for platform to specifies the 2nd
timeout value in microseconds for the BSP/AP in SMM to wait for
one another to enter SMM.
The added interface can enhance the flexibility of timeout
configuration. In some cases, certain processors may not be
able to enter SMI, and prolonged waiting could lead to
kernel soft/hard lockup. We have now defined two timeouts.
The first timeout can be set to a smaller value to reduce
the waiting period. Processors that are unable to enter SMI
will be woken up through SMIIPL to enter SMI, followed by
a second waiting period. The second timeout can be set to a
larger value to prevent delays in processors entering SMI
case due to the long instruction execution.
Signed-off-by: Jiaxin Wu <jiaxin.wu@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Add OvmfRng include snippets with the random number generator
configuration for OVMF. Include RngDxe, build with BaseRngLib,
so the rdrand instruction is used (if available).
Also move VirtioRng to the include snippets.
Use the new include snippets for OVMF builds.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Check whenever RngLib actually returns random numbers, only return
a non-zero number of Algorithms if that is the case.
This has the effect that RndDxe loads and installs EFI_RNG_PROTOCOL
only in case it can actually deliver random numbers.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
RDRAND has notoriously been broken many times over its lifespan.
Add a smoketest to RDRAND, in order to better sniff out potential
security concerns.
Also add a proper CPUID test in order to support older CPUs which may
not have it; it was previously being tested but then promptly ignored.
Testing algorithm inspired by linux's arch/x86/kernel/cpu/rdrand.c
:x86_init_rdrand() per commit 049f9ae9..
Many thanks to Jason Donenfeld for relicensing his linux RDRAND detection
code to MIT and the public domain.
>On Tue, Nov 22, 2022 at 2:21 PM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
<..>
> I (re)wrote that function in Linux. I hereby relicense it as MIT, and
> also place it into public domain. Do with it what you will now.
>
> Jason
BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=4163
Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Zhiguang Liu <zhiguang.liu@intel.com>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Starting from certain version of Visual Studio C compiler (I don’t
have the exact version. I am using VS2019), CpuDeadLoop is optimized
quite well by compiler.
The compiler does not generate instructions that jump out of the loop
when the "Index" is non-zero.
It becomes harder/impossible for developers to break out of the dead-loop
in debugger.
The new version of CpuDeadLoop() compares a volatile global to a volatile
local. This forces 2 reads and a comparison on every loop iteration.
The local variable can be set to 1 to exit the loop without modifying the
global variable.
Using VS2019 with max opt enabled, The dead-loop can be exit by setting
Index to 1 in a debugger.
Signed-off-by: Michael D Kinney <michael.d.kinney@intel.com>
Fix calculation of first section in FileFindSection for FILE2 headers
in UefiPayloadEntry module.
Signed-off-by: Sebastian Witt <sebastian.witt@siemens.com>
Reviewed-by: Guo Dong <guo.dong@intel.com>
Cc: Sean Rhodes <sean@starlabs.systems>
Cc: James Lu <james.lu@intel.com>
Reviewed-by: Gua Guo <gua.guo@intel.com>
This reverts commit 32460bb5b1, which is
no longer needed as ArmVirtQemu now has its own special implementation
for ArmMonitorLib.
Continuous-integration-options: PatchCheck.ignore-multi-package
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
This reverts commit 865229bcc8, and
restores the old state where dynamic PCDs are only used when TPM support
is configured.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
This reverts commit c98f7f7550, which is
no longer needed: the SMCCC conduit will be converted back to being
hardcoded, as PrePi based ArmVirtPkg build cannot support dynamic PCDs,
and falling back to patchable PCDs does not work either.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Whether SMCCC calls use HVC or SMC generally depends on the exception
level that the firmware executes at, but also on whether or not EL2 is
implemented.
This is almost always known at build time, which is why the default
ArmMonitorLib used to model this as a feature PCD. However, on QEMU,
things are not that simple.
However, SMCCC specifies that the conduit is the same as the one used
for PSCI calls (which has been retrofitted into SMCCC when it was
defined). Given that QEMU provides this information via the device tree,
let's use it to select the conduit, using a special ArmMonitorLib
implementation.
This also removes the need to set the associated PCD at runtime, given
that its updated value will no longer be used.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
When building ArmVirtQemuKernel with CLANGDWARF (which does not require
a GCC workaround where -mgeneral-regs-only is needed to ensure
-mstrict-align works as expected), the C code invoked from the PrePi
startup code may contain instructions that access the FP/SIMD register
file. This means that the FP/SIMD must be enabled before making such
calls, and this is currently not the case. So fix that, by moving the
call to ArmEnableVFP() early into the asm startup code.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
The guest os which depends on vbeshim for video support is -- according
to the comments -- Windows 2008 R2. Which went EOL in January 2020,
more than four years ago.
Time to retire VbeShim. RIP.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
The ESRT table is not required in UEFI firmware. In such cases, the
table may not be present in the UEFI Configuration Table. The mEsrtTable
is to check if the IsNestedFmpCapsule() function is invoked at runtime
to determine whether to use gEsrtManagementProtocolGuid or the ESRT
table from the Configuration Table. Unfortunately, the check does not
cover situations where the ESRT is not present, potentially resulting in
a kernel crash. This patch is intended to fix this issue.
Signed-off-by: Nhi Pham <nhi@os.amperecomputing.com>
Following https://bugzilla.tianocore.org/show_bug.cgi?id=2817 this
bug could also apply to strncpy and strcat.
For strncpy use count+1 if smaller than MAX_STRING_SIZE. This still
restricts the destination size to MAX_STRING_SIZE as before but allows
a strncpy when the source is close after destination without triggering
the InternalSafeStringNoAsciiStrOverlap check in AsciiStrnCpyS.
For strcat use the destination string length + the size of the source
string including the terminator as destination size if smaller than
MAX_STRING_SIZE.
Also move both functions to CrtWrapper.c as they do not return the
correct return value. AsciiStrnCpyS and AsciiStrCatS return
RETURN_VALUE instead of a char * to the destination buffer.
Signed-off-by: Sebastian Witt <sebastian.witt@siemens.com>
strcpy fails when strSource is closer than 4096 bytes after strDest.
This is caused by an overlap check in AsciiStrCpyS:
//
// 5. Copying shall not take place between objects that overlap.
//
SAFE_STRING_CONSTRAINT_CHECK (InternalSafeStringNoAsciiStrOverlap
(Destination, DestMax, (CHAR8 *)Source, SourceLen + 1),
RETURN_ACCESS_DENIED);
Since DestMax is MAX_STRING_SIZE (0x1000) and with a Source
that is in this area behind Destination, AsciiStrCpyS will fail
and strcpy will do nothing.
When called by CRYPTO_strdup in openssl this leads to uninitialzed
memory that gets accessed instead of the copied string.
BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=2817
Signed-off-by: Sebastian Witt <sebastian.witt@siemens.com>
Emit a DEBUG_WARN message if there is not enough flash space left to
write/update a variable. This condition is currently not logged
appropriately in all cases, given that full variable store can easily
render the system unbootable.
This new message helps identifying this condition.
Signed-off-by: Oliver Steffen <osteffen@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Fix logic error that changes the commit range checked depending
on the verbosity level set.
Signed-off-by: Michael D Kinney <michael.d.kinney@intel.com>
Add support for GuestPhysBits (cpuid 0x80000008, eax, bits 23:16).
GuestPhysBits is a field which can be set by the hypervisor to inform
the guest about the /usable/ physical address space bits. This can be
smaller than the PhysBits of the CPU, for example because of nested
paging limitations.
OVMF will read GuestPhysBits, log the value, in case it is set use it
as upper limit.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
It is legal for the VarStoreId of a question to
be 0 per the UEFI spec:
"Specifies the identifier of a previously
declared variable store to use when storing the
question’s value. A value of zero indicates
no associated variable store."
Instead of hitting an assert just skip this
question as there is no value to return.
Signed-off-by: Jeff Brasen <jbrasen@nvidia.com>
Add dsc + fdf include files to add the MorLock drivers to the build.
Add the include files to OVMF build configurations.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Adding definitions for Ipmi Net Sensor Get/Set Thresholds commands and
structures as found in Ipmi specification v2.0
Signed-off-by: Aaron Pop <aaronpop@microsoft.com>
Enable the non-hardcode version of FdtNorFlashQemuLib in ArmVirtQemu.dsc
and ArmVirtQemuKernel.dsc, and it can work rightly after enabling it.
BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=4770
Build-tested (with "ArmVirtQemu.dsc" and "ArmVirtQemuKernel.dsc").
Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>
Cc: Leif Lindholm <quic_llindhol@quicinc.com>
Cc: Sami Mujawar <sami.mujawar@arm.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Signed-off-by: Chao Li <lichao@loongson.cn>
Co-authored-by: Xianglai Li <lixianglai@loongson.cn>
Tested-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
This library is copied from ArmVirtPkg, in the Arm version, the value of
PcdFlashNvStorageVariableBase, PcdFlashNvStorageFtwWorkingBase and
PcdFlashNvStorageFtwSpareBase are hardcoded in INC file.
This version will calculate them from FDT resource and using the set PCD
to store when the NorFlashInitialise is called. By default, the first
available flash(not used for storage UEFI code) as NV variable storage
medium.
In this way, UEFI can better handle the change of flash base address,
which is suitable for different cpu architecture board implementation.
BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=4770
Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>
Cc: Leif Lindholm <quic_llindhol@quicinc.com>
Cc: Sami Mujawar <sami.mujawar@arm.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Signed-off-by: Chao Li <lichao@loongson.cn>
Co-authored-by: Xianglai Li <lixianglai@loongson.cn>
Tested-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Struct assignments are not permitted in EDK2, as they may be converted
by the compiler into calls to the 'memcpy' intrinsic, which is not
guaranteed to be available in EDK2.
So replace the assignment with a call to CopyMem (), and -while at it-
replace the loop with a single CopyMem () call, as the loop operates on
items that are contiguous in memory.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
This patch is to fix issue when splitting leaf paging
entry in CpuPageTableLib code.
In previous code, before we assign the new child paging
structure address to the content of splitted paging entry,
PageTableLibSetPnle() is called to make sure the bit7 is
set to 0, which indicate the previous leaf entry is
changed to non-leaf entry now. There is a gap between
we change the bit7 and we assign the new child paging
structure address to the content of the splitted paging
entry. If the address of code execution or data access
happens to be in the range covered by the splitted paging
entry, this gap may cause issue.
In this patch, we prepare the new paging entry content
value in a local variable and assign the value to the
splitted paging entry at once. The volatile keyword
is used to ensure that no optimization will occur in
compilation.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Cc: Zhou Jianfeng <jianfeng.zhou@intel.com>
Remove GetAcpiCpuData() in CpuS3.c. The mAcpiCpuData
is not needed in S3 boot anymore.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
This MpService2Ppi field in SMM_S3_RESUME_STATE is used to
wakeup AP to do the CPU initialization during smm s3 boot when
the execution mode of PEI and DXE are the same.
Currently, in CpuS3.c of smm cpu driver, BSP doesn't need to
wakeup AP anymore. The initialization for AP will be done in
S3Resume.c before transfer to CpuS3.c of smm cpu driver.
So we can remove the MpService2Ppi field in SMM_S3_RESUME_STATE.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Reviewed-by: Liming Gao <gaoliming@byosoft.com.cn>
Remove the unneeded assignment of MpService2Ppi field
in SmmS3ResumeState. Previously, when the execution
combination of PEI and DXE are the same, the pointer
of mpservice ppi will be passed to CpuS3.c in smm cpu
driver to wakeup all APs, instead of init-sipi-sipi.
Currently, CpuS3.c doesn't need to wakeup Aps anymore.
So remove the duplicated mpservice locate and assignment
to MpService2Ppi field in SmmS3ResumeState.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
After the code to load mtrr setting, set register table,
handle APIC setting and Interrupt after INIT-SIPI-SIPI
is moved, the InitializeCpuProcedure() only contains
following code logic:
1.Bsp runs ExecuteFirstSmiInit().
2.Bsp transfers AP to safe hlt-loop
During S3 boot, since APs will be relocated to new safe
buffer by the callback of gEdkiiEndOfS3ResumeGuid in
PeiMpLib, Bsp doesn't need to transfer AP to safe hlt-loop
any more. SmmRestoreCpu() in CpuS3 only needs to runs the
ExecuteFirstSmiInit() on BSP. So remove code to wakeup
AP by INIT-SIPI-SIPI and remove code to relocate ap to
safe hlt-loop.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Remove ProgramVirtualWireMode()/DisableLvtInterrupts()
since APs won't be waken by INIT-SIPI-SIPI in CpuS3.c
any more. The two functions has been executed in
MpInitLibInitialize() in PeiMplib.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
In this commit, change PeiMpLib to install callback
of gEdkiiEndOfS3ResumeGuid to relocate AP to new safe
buffer. The gEdkiiEndOfS3ResumeGuid is installed in
S3Resume.c before jmping to OS waking vector.
Previously, code in CpuS3.c of PiSmmCpuDxe driver will
prepare the new safe buffer for AP and place AP in hlt
loop state. With this code change, we can remove the
Machine Instructions of mApHltLoopCode in PiSmmCpuDxe.
Also we can reuse the related code in DxeMpLib for
PeiMpLib.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Install gEdkiiEndOfS3ResumeGuid in S3Resume to trigger
callback registered by PeiMpLib. The callback is to
relocate Ap to new safe memory before jump to OS waking
vector in S3 boot flow.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Move some code in DxeMpLib.C to common MpLib.c.
The related code is to relocate Ap to new safe buffer
before booting into OS. In next commits, these code
also will be used by PeiMpLib. This commit doesn't
change any code functionality.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Abstract some DxeMpLib code to function in this commit.
Some of these internal functions will be moved to common
MpLib.c in following commits. Then PeiMpLib can reuse
the code.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Disable paging in IA32 RelocateApLoop assembly
code to fix the issue that the AP page table is
unavailiable after boot OS under IA32 execution mode.
This issue exist in IA32 PEI + IA32 DXE normal boot
(also S3 boot with IA32 PEI after previous three commits
are accepted). In current MpLib code, the IA32 execution
mode code did not create page table in reserved memory
like what X64 code did. If PcdCpuStackGuard is TRUE, the
PG is enabled for AP in current RelocateApLoop assembly
code. And the page table for AP is unavailiable after
boot OS. This might cause potential issue. So disable PG
in IA32 RelocateApLoop.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Remove code to set register table in CpuS3.c.
In previous commit, PcdCpuFeaturesInitOnS3Resume
has been set to TRUE. So that CpuFeaturesPei PEIM
will initialize the CPU registers and perform CPU
features initialization.
Signed-off-by: Dun Tan <dun.tan@intel.com>
Reviewed-by: Ray Ni <ray.ni@intel.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiaxin Wu <jiaxin.wu@intel.com>
Pierre Gondois
Leif Lindholm
Gerd Hoffmann
Jiaxin Wu
Pedro Falcato
Ray Ni
Dhaval
Sebastian Witt
Ard Biesheuvel
Nhi Pham
Rebecca Cran
HoraceX Lien
Oliver Steffen
Michael D Kinney
Wenxing Hou
Jeff Brasen
Chao Li
Neo Hsueh
Aaron
Dun Tan