On Mon, Nov 23, 2015 at 05:56:29PM +0100, Ard Biesheuvel wrote:
> The way the v7 MMU code is invoked by the Xen port is somewhat of
> a pathological case, since it describes its physical memory space
> using a single cacheable region that covers the entire addressable
> range. When clipping this
I'll take that as a reviewed-by, thanks :)
Pushed as r18925.
On Mon, Nov 23, 2015 at 05:21:16PM +, Carsey, Jaben wrote:
> Not sure how to RB this ... it's fine with me to commit it.
>
> > -Original Message-
> > From: Ard Biesheuvel [mailto:ard.biesheu...@linaro.org]
> > Sent:
Laszlo,
Comments below.
Mike
> -Original Message-
> From: Laszlo Ersek [mailto:ler...@redhat.com]
> Sent: Monday, November 23, 2015 4:34 AM
> To: Kinney, Michael D ; edk2-de...@ml01.01.org
> Cc: Justen, Jordan L
> Subject: Re:
On 11/23/15 21:13, Kinney, Michael D wrote:
> Laszlo,
>
> Comments below.
>
> Mike
>
>> -Original Message-
>> From: Laszlo Ersek [mailto:ler...@redhat.com]
>> Sent: Monday, November 23, 2015 4:34 AM
>> To: Kinney, Michael D ; edk2-de...@ml01.01.org
>> Cc:
Laszlo,
There are 2 reasons to list all source files used in a module build in the
[Sources] section.
1) Support incremental builds. If a change to the .h file is made, then the
module may not be rebuilt if the .h file is not listed in [Sources]
2) Support of UEFI Distribution Package
If there are any logical processor reporting an APIC ID of 255 or greater. We
should enable x2APIC mode.
Jeff Fan (3):
UefiCpuPkg/CpuMpPei: Get APIC ID from CPUID if x2APIC supported
UefiCpuPkg/CpuMpPei: Set X2APIC flag if one x2APIC ID larger than 254
UefiCpuPkg/CpuMpPei: Enable x2APIC
If there are any logical processor reporting an APIC ID of 255 or greater, set
X2ApicEnable flag.
Cc: Feng Tian
Cc: Michael Kinney
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jeff Fan
---
If x2APIC is supported by processor, get the APIC ID from
CPUID.(EAX=0BH, ECX=0H):EDX instead of legacy APIC ID. It is used to check if
need to enable x2APIC mode.
Cc: Feng Tian
Cc: Michael Kinney
Contributed-under: TianoCore Contribution
On 11/21/15 10:44, Ard Biesheuvel wrote:
> Hello all,
>
> After Drew pointed out that mach-virt now populates the memory region beyond
> DRAM, I am proposing this approach instead. Since KVM limits its IPA space to
> 40 bits, there is simply no point in supporting anything beyond that for
>
On 23 November 2015 at 12:41, Laszlo Ersek wrote:
> On 11/21/15 10:44, Ard Biesheuvel wrote:
>> The ID mapping routines on virtual platforms simply map the entire
>> address space as device memory, and then punch some holes for regions
>> that need to be mapped cacheable. On
TSS segment should use (SIZE - 1) as limit, and do not set G bit (highest bit
of LimitHigh) because limit means byte count.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Yao, Jiewen
Cc: Fan, Jeff
Cc: Kinney, Michael D
On 11/21/15 07:36, Kinney, Michael D wrote:
> Laszlo,
>
> Minor comments included below. I know from later items in this thread that
> SMM_COMMUNICATE has already been removed in your local branch.
>
> Reviewed-by: Michael Kinney
Thank you. I've picked this up
On 11/21/15 07:17, Kinney, Michael D wrote:
> Laszlo,
>
> Minor comments included below.
>
> Reviewed-by: Michael Kinney
Thank you. I'll pick up this tag. Answers and further questions below:
>
> Mike
>
>> -Original Message-
>> From: edk2-devel
On 11/21/15 10:44, Ard Biesheuvel wrote:
> The ID mapping routines on virtual platforms simply map the entire
> address space as device memory, and then punch some holes for regions
> that need to be mapped cacheable. On virtual platforms hosted on CPUs
> that support a large physical address
On 11/21/15 07:03, Kinney, Michael D wrote:
> Laszlo,
>
> Minor comments included below.
>
> Reviewed-by: Michael Kinney
Thank you. I'll pick up your R-b, but I'll also respond to your comments
below:
>> -Original Message-
>> From: edk2-devel
Load File protocol requires remaining device path rather than whole
device path. For PXE, it actually requires end node device path only,
or else invalid parameter will be returned directly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Heyi Guo
Cc:
Hi everyone,
I'd like to integrate a few applications into the shell so they are directly
available when booting the shell vie PXE. One of the applications I wanted to
include in the shell is made using EADK, specifically the socket library. I got
it integrated and can run it from the shell,
Correct one typo (SingerChainCerts --> SignerChainCerts) in the comments
for Pkcs7GetCertificatesList() API.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Qin Long
CC: Shumin Qiu
---
CryptoPkg/Include/Library/BaseCryptLib.h
Original code refers FreePoolHdr without check Status. It is obvious wrong and
has risk.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Zeng, Star
Reviewed-by: Yao, Jiewen
Cc: Fan, Jeff
Cc: Kinney, Michael
Original code does not uninstall LoadedImage protocol if SMM driver returns
error and is unloaded.
It causes a wrong LoadedImage protocol existing in system.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Yao, Jiewen
Cc: Zeng, Star
PiSmmCore installs LoadedImage for each SMM driver. However ECP SMM driver is
missing.
Since SmmBaseHelper loads ECP SMM driver, we let SmmBaseHelper installs
LoadedImage protocol for SMM driver.
So that the SMM image information is complete.
Contributed-under: TianoCore Contribution Agreement
On 11/23/15 02:44, Star Zeng wrote:
> Cc: Michael D Kinney
> Cc: Liming Gao
> Cc: Jordan Justen
> Cc: Laszlo Ersek
> Contributed-under: TianoCore Contribution Agreement 1.0
> Signed-off-by: Star
Star,
On 11/23/15 02:44, Star Zeng wrote:
> Beyond just changing the directly related lines in the FDF and DSC files,
> we have to adapt the EarlyFdtPL011SerialPortLib and FdtPL011SerialPortLib
> instances as well, in the same patch. This is because the EmbeddedPkg
> driver expects the
Move Gdt initialization from InitializeMpServiceData() to CPU Arch specific
function.
We create SmmFuncsArch.c for hold CPU specific function, so that
EFI_IMAGE_MACHINE_TYPE_SUPPORTED(EFI_IMAGE_MACHINE_X64) can be removed.
For IA32 version, we always allocate new page for GDT entry, for easy
In this way, we can centralize the silicon configuration in
PerformRemainingTasks() function.
If there are more features need to be configured, they can put in
PerformRemainingTasks() only.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Yao, Jiewen
The TSS is already fixed in PiSmmCpuDxeSmm/X64/SmmFuncsArch.c, InitGdt().
There is no need to fix it again.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Yao, Jiewen
Cc: Fan, Jeff
Cc: Kinney, Michael D
Hi Andrew,
I looked at the ArmVirtPkg and did the same before I sent the previous
email. I am missing some step that is causing the null package to be
picked up. Here are my changes, I think I covered most but still
missing something.
Thanks for your help,
> -Original Message-
> From: Ard Biesheuvel [mailto:ard.biesheu...@linaro.org]
> Sent: Wednesday, November 18, 2015 11:32 PM
> To: Vladimir Olovyannikov
> Cc: Mark Rutland; edk2-devel@lists.01.org
> Subject: Re: [edk2] Armv8 64bit: System error booting linux from the UEFI
>
> On 19
Qiu Shumin (2):
PerformancePkg\Dp_App: Add NULL check to pointer returned from
'AllocateZeroPool'.
ShellPkg: Add NULL check to pointer returned from 'AllocateZeroPool'.
PerformancePkg/Dp_App/Dp.c | 3 ++-
ShellPkg/Library/UefiDpLib/Dp.c | 3 ++-
2 files changed, 4 insertions(+), 2
When CustomCumulativeToken is not NULL, the CustomCumulativeData is expected
non-NULL.
Add 'ASSERT' statement to ensure this.
Cc: Star Zeng
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Qiu Shumin
---
When CustomCumulativeToken is not NULL, the CustomCumulativeData is expected
non-NULL.
Add 'ASSERT' statement to ensure this.
Cc: Jaben Carsey
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Qiu Shumin
---
Provide a more detailed description of each field of the
ACPI_CPU_DATA and CPU_REGISTER_TABLE structures.
Cc: Laszlo Ersek
Cc: "Yao, Jiewen"
Cc: "Fan, Jeff"
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by:
This module initializes the ACPI_CPU_DATA structure and registers the
address of this structure in the PcdCpuS3DataAddress PCD. This is
simplest version of this module. It does not provide a machine check
handler or CPU register initialization tables for ACPI S3 resume.
This module can be copied
This patch series adds a generic/simple module that allocates,
initializes, and registers the ACPI_CPU_DATA structure required
to support ACPI S3 resume. This patch series is in response to
the OvmfPkg patch series from Laszlo Ersek that enables SMM on
OVMF. The v4 version of that patch
when multiple Dynamic PCD have different token space guid but same PCD
name, it is difficult for user to check why the generated autogen.c and
autogen.h are not consistent. so we add a check before generating
autogen.c and report error directly that user can know what happened
immediately.
Laszlo,
I other modules where we want to halt no matter what, the following 2
statements are used together. ASSERT() could be removed by PCD settings, so
dead loop catches that case. If ASSERT() is enabled, then the ASSERT()
behavior of BP or dead loop is controlled by PCD.
ASSERT
Reviewed-by: Michael Kinney
> -Original Message-
> From: Yao, Jiewen
> Sent: Monday, November 23, 2015 5:13 AM
> To: edk2-de...@ml01.01.org
> Cc: Yao, Jiewen ; Fan, Jeff ;
> Kinney, Michael D
> On Nov 23, 2015, at 2:03 PM, Narinder Dhillon wrote:
>
> Hi Andrew,
>
> I looked at the ArmVirtPkg and did the same before I sent the previous
> email. I am missing some step that is causing the null package to be
> picked up. Here are my changes, I think I covered most
> -Original Message-
> From: Ard Biesheuvel [mailto:ard.biesheu...@linaro.org]
> Sent: Tuesday, November 10, 2015 10:04 AM
> To: Vladimir Olovyannikov
> Cc: Cohen, Eugene; edk2-devel@lists.01.org
> Subject: Re: [edk2] Strange behavior of the DS-5 debugger on AARCH64 with
> step-by-step
Hi Laszlo
For 3), I see your point.
I think 2) again, and realize there might be a corner case.
A) There is SMI handler issue which might call-out SMRAM.
B) Malicious software may construct such environment in OS. (But not trigger
it, because it will be blocked by code access)
C) SMM CPU driver
Reviewed-by: Qiu Shumin
-Original Message-
From: Long, Qin
Sent: Monday, November 23, 2015 4:39 PM
To: edk2-devel@lists.01.org
Cc: Qiu, Shumin
Subject: [Patch] [CryptoPkg] Correct one typo in the API comments.
Correct one typo (SingerChainCerts -->
More comments below.
On 2015/11/24 9:01, Zeng, Star wrote:
Laszlo,
Explain more below.
On 2015/11/23 22:01, Laszlo Ersek wrote:
Star,
On 11/23/15 02:44, Star Zeng wrote:
Beyond just changing the directly related lines in the FDF and DSC
files,
we have to adapt the
I'm trying to make one DXE driver dependent upon another. From the INF spec, it
appears that I can put "AFTER path/to/dependency.inf" in my driver's Depex
section. If I try this I get the error below.
(Python 2.7.3 on win32) Traceback (most recent call last):
File "build.py", line 2032, in
David,
When using BEFORE and AFTER, you have to provide a GUID C name for the FFS file
GUID of the driver you want to run before or after
The supported syntax is
BEFORE
AFTER
This means you have to declare a GUD in a DEC file that is mapped to the same
GUID value of the
On 2015/11/23 21:29, jiewen yao wrote:
Original code does not uninstall LoadedImage protocol if SMM driver returns
error and is unloaded.
It causes a wrong LoadedImage protocol existing in system.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Yao, Jiewen
Jiewen,
Could you combine this patch with [patch] UefiCpuPkg/PiSmmCpu: Eliminate
EFI_IMAGE_MACHINE_TYPE_SUPPORTED(EFI_IMAGE_MACHINE_X64)?
Reviewed-by: Jeff Fan
Jeff
-Original Message-
From: Yao, Jiewen
Sent: Monday, November 23, 2015 10:15 PM
To:
Agree. Thanks!
-Original Message-
From: Fan, Jeff
Sent: Tuesday, November 24, 2015 11:33 AM
To: Yao, Jiewen; edk2-de...@ml01.01.org
Cc: Kinney, Michael D
Subject: RE: [patch] UefiCpuPkg/PiSmmCpu: Eliminate
EFI_IMAGE_MACHINE_TYPE_SUPPORTED(EFI_IMAGE_MACHINE_X64).
This commit title is
Sorry, please ignore this patch.
I just realized that mLoadPe32Image->LoadPeImage() can install LoadedImage
protocol.
-Original Message-
From: edk2-devel [mailto:edk2-devel-boun...@lists.01.org] On Behalf Of jiewen
yao
Sent: Monday, November 23, 2015 9:41 PM
To: edk2-de...@ml01.01.org
Reviewed-by: Michael Kinney
> -Original Message-
> From: Yao, Jiewen
> Sent: Monday, November 23, 2015 5:29 AM
> To: edk2-de...@ml01.01.org
> Cc: Yao, Jiewen ; Zeng, Star ;
> Fan, Jeff ; Kinney,
HI
I just discussed with Star.
We decide to let PiSmmCore install LoadedImage protocol - same way as DxeCore.
If so, we can assign correct FilePath there. Then we do not need update
PiSmmIpl.
I will send another patch. This update can be discard.
-Original Message-
From: Yao, Jiewen
Jiewen,
I agree making the behavior the same as DXE Core makes more sense.
Thanks,
Mike
> -Original Message-
> From: Yao, Jiewen
> Sent: Monday, November 23, 2015 11:22 PM
> To: edk2-de...@ml01.01.org
> Cc: Zeng, Star ; Fan, Jeff ; Kinney,
>
Mike,
OK. So, if 0xB leaf is not supported, then ASSERT() in GetX2ApicIdFromCpuId();
//
// Get the maximum index of basic CPUID
//
AsmCpuid (CPUID_SIGNATURE, , NULL, NULL, NULL);
ASSERT (MaxCpuIdIndex >= CPUID_EXTENDED_TOPOLOGY);
Jeff
-Original Message-
From: Kinney,
On 23 November 2015 at 21:55, Wei Huang wrote:
>
>
> On 11/23/2015 11:10 AM, Ard Biesheuvel wrote:
>> Hello all,
>>
>> After Drew pointed out that mach-virt now populates the memory region beyond
>> DRAM, I am proposing this approach instead. Since KVM limits its IPA space to
>>
Jeff,
Yes. That is ASSERT() location I was suggesting.
Does is also makes sense to fall back to GetInitialApicId() if CPUID leaf is
not supported? Or is this really a condition that should never occur?
Mike
> -Original Message-
> From: Fan, Jeff
> Sent: Monday, November 23, 2015
Using RVCT 4, I get the following error building openssl-1.0.2d:
1> Building ... c:\bios_24s\edk2\CryptoPkg\Library\OpensslLib\OpensslLib.inf
[ARM]
[snip]
c:\bios_24s\edk2\CryptoPkg\Library\OpensslLib\openssl-1.0.2d\crypto\cryptlib.c
R_ARM_REL32 are relative relocations, so we don't need to do anything
special when performing the ELF to PE/COFF conversion, since our memory
layout is identical between the two binary formats. So just allow them.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard
By special request, this implements ARM support to the ArmVirtXen
platform. This is probably still rough around the edges, since I have
not tested it myself under Xen yet. I did test all the code changes in
this series under QEMU. (This is possible since the early code only depends
on the Linux
The way the v7 MMU code is invoked by the Xen port is somewhat of
a pathological case, since it describes its physical memory space
using a single cacheable region that covers the entire addressable
range. When clipping this region to the part that is 1:1 addressable,
we end up with a region of
Parsing the DTB early on using a handcoded assembly routine is a pointless
waste of brain cycles, since the UEFI firmware always executes from RAM
under Xen. So instead, set up a temporary stack in the memory region at the
beginning of the image, and use the libfdt C library.
Contributed-under:
This adds ARM support to the ArmVirtXen platform. As is the case for
AARCH64, the ARM port adheres to the ARM Linux boot protocol, i.e.,
it expects the address of a DTB describing the platform to be passed
in r2, and relocates itself at runtime to the actual load time memory
offset.
This is a port of the AARCH64 low level init routines to ARM. This
mainly covers the platform boot code that extracts the system base
and size from the DTB, copies it and updates the FD and FV base
addresses according to the load time offset.
Contributed-under: TianoCore Contribution Agreement
This adds support to the self relocating PrePi instance that is built
as a PIE ET_DYN executable. It primarily involves porting the relocation
routine to use ELF32 REL entries instead of ELF64 RELA entries which is
what AArch64 uses.
Contributed-under: TianoCore Contribution Agreement 1.0
On 11/23/15 17:56, Ard Biesheuvel wrote:
> By special request, this implements ARM support to the ArmVirtXen
> platform. This is probably still rough around the edges, since I have
> not tested it myself under Xen yet. I did test all the code changes in
> this series under QEMU. (This is possible
KVM uses a fixed size of 40 bits for its intermediate physical address
space, so there is no need to support anything beyond that even if the
host hardware does.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel
Reviewed-by: Laszlo
Hello all,
After Drew pointed out that mach-virt now populates the memory region beyond
DRAM, I am proposing this approach instead. Since KVM limits its IPA space to
40 bits, there is simply no point in supporting anything beyond that for
ArmVirtQemu.
I am cc'ing the Xen guys since they may run
On 11/23/15 18:10, Ard Biesheuvel wrote:
> The ID mapping routines on virtual platforms simply map the entire
> address space as device memory, and then punch some holes for regions
> that need to be mapped cacheable. On virtual platforms hosted on CPUs
> that support a large physical address
Hi, Eugene,
Sorry, I have no RVCT environment for more investigations.
Could you help to double-check if this is caused by any CryptoPkg updates? Or
It is one OpenSSL-native issue?
Could we silence this simply with any build flag, such as " --diag_suppress"?
Moreover, some old RVCT release
Not sure how to RB this ... it's fine with me to commit it.
> -Original Message-
> From: Ard Biesheuvel [mailto:ard.biesheu...@linaro.org]
> Sent: Monday, November 23, 2015 5:22 AM
> To: Leif Lindholm
> Cc: edk2-devel@lists.01.org; Carsey, Jaben
Comments below:
-Original Message-
From: Laszlo Ersek [mailto:ler...@redhat.com]
Sent: Monday, November 23, 2015 10:39 PM
To: Yao, Jiewen; edk2-de...@ml01.01.org
Cc: Kinney, Michael D; Fan, Jeff
Subject: Re: [edk2] [patch] UefiCpuPkg/PiSmmCpu: Move
RestoreSmmConfigurationInS3 from
On 23 November 2015 at 17:56, Ard Biesheuvel wrote:
> By special request, this implements ARM support to the ArmVirtXen
> platform. This is probably still rough around the edges, since I have
> not tested it myself under Xen yet. I did test all the code changes in
>
Laszlo,
Based on size analysis of compiler output, (3) is not the smallest size. That
is why we prefer pre-init global.
Using your current patch style (1) in OvmfPkg specific modules is ok until you
can find a solution to your ctags issue.
Mike
> -Original Message-
> From:
On 11/23/15 22:01, Kinney, Michael D wrote:
> Laszlo,
>
> Based on size analysis of compiler output, (3) is not the smallest size.
> That is why we prefer pre-init global.
Ugh, I completely forgot about physical flash, where size is a premium...
>
> Using your current patch style (1) in
On 11/23/15 21:28, Kinney, Michael D wrote:
> Laszlo,
>
> There are 2 reasons to list all source files used in a module build in the
> [Sources] section.
>
> 1) Support incremental builds. If a change to the .h file is made,
> then the module may not be rebuilt if the .h file is not listed in
Laszlo,
The section [UserExtensions.TianoCore."ExtraFiles"] can be used to list files
not required to build and support incremental builds, but need to be included
in a UDP package.
Mike
> -Original Message-
> From: Laszlo Ersek [mailto:ler...@redhat.com]
> Sent: Monday, November 23,
74 matches
Mail list logo