Clear EFLAGS.IF before executing Stepping command and save original
value in DEBUG_AGENT_FLAG. It could avoid pending interrupt issued
during executing Stepping command. Restore EFLAGS.IF after Stepping
command execution finished.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Jeff Fan <jeff.fan@intel.com> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17204 6f19259b-4bc3-4df7-8a09-765794883524
ShellPkg: Refine the logic about allocating memory for variable name and data.
The run time service 'QueryVariableInfo' is not proper to be used to get the variable name size. This patch refine the logic about allocating memory for variable name and data.
-- Add BootLoaderTolumSize support
-- Extend FspApiCallingCheck with ApiParam for BootLoaderTolumSize
-- Rename all Bootloader to BootLoader as official name
-- Rename Ucode to Microcode
-- Remove FspSelfCheck API, because it is merged into SecPlatformInit
-- Add GetFspVpdDataPointer() in FspCommonLib.h
-- Document FspSecPlatformLib.h
-- Reorg FSP_PLAT_DATA data structure to let it match FSP spec.
-- Move helper function in FspSecCore to reduce platform enabling effort
-- Fix LibraryClasses declaration in DEC file.
-- Enhance PatchFv to check if it is valid FSP bin.
MdeModulePkg: PCD/Pei: eliminate unused but set variable
- SVN r14866:
MdePkg and MdeModulePkg Pcd: Add the new EFI_GET_PCD_INFO_PROTOCOL and
EFI_GET_PCD_INFO_PPI support for PI 1.2.1 compliance.
added the "DataBase" local variable to PcdPeimInit(), and both set it and
used it.
- SVN r14869:
MdeModulePkg and Nt32Pkg Pcd: Add the new EFI_GET_PCD_INFO_PROTOCOL and
EFI_GET_PCD_INFO_PPI support for PI 1.2.1 compliance.
changed the PcdPeimInit() function, but "DataBase" remained both set and
used.
- SVN r17173:
MdeModulePkg Pcd: Check the input SkuId in SetSku()
changed the function again; and this time "DataBase" became
set-but-unused. It triggers the following build error, when building
ArmVirtualizationQemu.dsc with gcc-4.8:
MdeModulePkg/Universal/PCD/Pei/Pcd.c:150:21: error: variable 'DataBase'
set but not used [-Werror=unused-but-set-variable]
PEI_PCD_DATABASE *DataBase;
^
cc1: all warnings being treated as errors
Fix the error by removing the DataBase variable, restoring the pre-r14866
state locally, when the BuildPcdDatabase() function was called, but its
return value was thrown away.
Cc: Star Zeng <star.zeng@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Laszlo Ersek <lersek@redhat.com> Reviewed-by: Star Zeng <star.zeng@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17179 6f19259b-4bc3-4df7-8a09-765794883524
ArmVirtualizationPkg: Xen: shuffle init order to deal with incoherency
In order to prevent memory corruption issues caused by the fact that,
under virtualization, the guest is incoherent with the hypervisor's view
of memory until it enables its caches and MMU, this patch reshuffles the
init sequence so that the Xen shared memory regions are not touched
before the caches and MMU are enabled.
In addition, the loaded image itself is invalidated by virtual address,
to ensure that any runtime changes (such as the applied relocations) will
not suddenly become invisible once we turn the caches on.
ArmVirtualizationPkg: invalidate PEI memory region by VA
This updates ArmVirtualizationMemoryInitPeiLib so that the PEI memory
region, i.e., the region that is used both before and after the MMU
and caches are enabled, is invalidated by virtual address before
enabling the MMU.
This prevents issues where data we modified with the caches and MMU
off may be shadowed by clean cachelines in system caches or in lower
level caches on other CPUs, resulting in the this data to become
invisible once we turn the MMU and caches on.
ArmPkg: remove cache maintenance by VA operation range size threshold
This removes the range size threshold for virtual address based cache
maintenance instructions that operate on VA ranges to be 'promoted' to
use set/way instructions.
Doing so is unsafe: set/way operations are fundamentally different
from VA operations, and really only suitable for cleaning or invalidating
a cache when turning it on or off.
To quote the ARM ARM (DDI0487A_d G3.4):
"""
Since the set/way instructions are performed only locally, there is no
guarantee of the atomicity of cache maintenance between different PEs,
even if those different PEs are each performing the same cache maintenance
instructions at the same time. Since any cacheable line can be allocated
into the cache at any time, it is possible for [a] cache line to migrate
from an entry in the cache of one PE to the cache of a different PE in a
manner that the cache line avoids being affected by set/way based cache
maintenance. Therefore, ARM strongly discourages the use of set/way
instructions to manage coherency in coherent systems.
"""
ArmPlatformPkg: do not fulfil MemoryInitPeiLib dependency directly via .c file
MemoryInitPeim short-circuits its MemoryInitPeiLib dependency by including
the .c file directly. This prevents us from having a special implementation
for ArmVirtualizationPkg that performs additional cache maintenance before
enabling the MMU. So instead, make it depend on the library class.
ArmVirtualizationPkg: make ArmVirtualizationMemoryInitPeiLib the default
This updates ArmVirtualization.dsc.inc to use the MemoryInitPeiLib
implementation for virt targets. The only difference between that one
and the original one is that the original one removes memory from the
available list if it overlaps the FD region (which may be the case when
shadowing NOR flash with system RAM). This is never the case for the
Qemu target, which is the only platform affected by this change, since
the Xen target already uses this library explicitly.
This allows us to remove the Xen-specific declaration of this library
dependency. For the Qemu target, this change will not take effect until
after the MemoryInitPeim<->MemoryInitPeiLib dependency resolution is
fixed in a subsequent patch.
Save initial TSVal from TCP connection initiation packets.
RFC1323 says the most recently received TSVal must be echoed in TSecr in ACK packets
which was not done at first connection because the code saving the TSVal from the peer
was skipped at this point. This resulted in sending an ACK reply with a 0 TSecr that
was rejected at least by FreeBSD. This patch fixes this by updating the saved TSVal
also for connection initiation packets.
Thanks to Laszlo Ersek for analysis and help in debugging.
Maurice Ma [Fri, 10 Apr 2015 22:08:57 +0000 (22:08 +0000)]
CorebootPayloadPkg: Remove empty folder reference in DEC file
In current CorebootPayloadPkg.dec an empty 'Include' directory is
listed in [includes] section. However, this empty directory will
not be mirrored into git repo. If the source tree is pulled from git,
the 'Include' empty folder will not exist and it will cause build
failure. The fix is to remove the whole [Includes] section in the
DEC file.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Maurice Ma <maurice.ma@intel.com> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17164 6f19259b-4bc3-4df7-8a09-765794883524
Bob Feng [Fri, 10 Apr 2015 06:59:47 +0000 (06:59 +0000)]
BaseTools/Build: Add SDL support
1.BaseTool add ATTRIBUTE (+/-RT, RO) support in PCD declaration in DSC file
2.BaseTool collect valid PCD value in DEC file and generate data base for runtime sanity check
3.BaseTool support SetPcd error.
Scott Duplichan [Fri, 10 Apr 2015 03:00:40 +0000 (03:00 +0000)]
MdePkg: Fix build fail when DDK3790 tool chain is used
Modify the preprocessor check for variadic macro support. The existing
check assumes all Microsoft compilers with major version of 14 and greater
support variadic macros. This is almost correct. An exception is the X64
compiler from DDK3790. This compiler has version 14.00.40310.41, yet does
not support variadic macros. Both VS2005 compilers have version number
14.00.50727.762 and do support variadic macros. For simplicity, this change
suppresses the use of variadic macros for all Microsoft compilers with
version 14 or smaller. If the variadic macro support needs to be restored
for VS2005, that could be done using new logic invloving _MSC_FULL_VER.
Enable use of 1GB pages to avoid page table memory allocation failure.
The failure occurs when allocating enough 4KB pages to map a large
virtual address space, such as the 48-bit virtual address space of AMD
family 10h and family 15h processors. The change is needed only for the
X64 build because page tables are not used by the IA32 build.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Prince Agyeman <prince.agyeman@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17151 6f19259b-4bc3-4df7-8a09-765794883524
Scott Duplichan [Fri, 10 Apr 2015 02:41:36 +0000 (02:41 +0000)]
CorebootPayloadPkg: Add NOOPT build to accommodate source level debugging
Add NOOPT build to accommodate source level debugging. The NOOPT build
avoids the use of compiler optimization so that every local variable is
accessible by a source level debugger.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Prince Agyeman <prince.agyeman@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17149 6f19259b-4bc3-4df7-8a09-765794883524
Scott Duplichan [Fri, 10 Apr 2015 02:41:01 +0000 (02:41 +0000)]
CorebootModulePkg: Change CbParseAcpiTable prototype to avoid gcc fail
Use of void** as a generic pointer to pointer is a Microsoft extension
to the C language and is not supported by gcc. Without this change, gcc
compile fails with error:
passing argument 1 of 'CbParseAcpiTable' from incompatible pointer type
note: expected 'void **' but argument is of type
'struct EFI_ACPI_3_0_ROOT_SYSTEM_DESCRIPTION_POINTER **'
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Maurice Ma <maurice.ma@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17144 6f19259b-4bc3-4df7-8a09-765794883524
Scott Duplichan [Fri, 10 Apr 2015 02:05:48 +0000 (02:05 +0000)]
CorebootModulePkg: Fix build failure with 32-bit NOOPT target
Fix build failure with 32-bit NOOPT target by replacing direct shift
of 64-bit integer with a function call. Otherwise Microsoft tool chains
will generate a call to function __allshl and fail to link.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Maurice Ma <maurice.ma@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17142 6f19259b-4bc3-4df7-8a09-765794883524
Elvin Li [Fri, 10 Apr 2015 01:37:41 +0000 (01:37 +0000)]
MdeModulePkg: Roll back report status code change in RuntimeDriverSetVirtualAddressMap.
Roll back report status code to original place in RuntimeDriverSetVirtualAddressMap.
Per UEFI spec, the call to SetVirtualAddressMap() must be done with the physical mappings.
We can not assume virtual address could work in SetVirtualAddressMap (), so we can not call
report status code interface with virtual address.
Scott Duplichan [Fri, 10 Apr 2015 00:42:30 +0000 (00:42 +0000)]
CorebootModulePkg: Add EFIAPI to OnReadyToBoot to fix gcc compile fail
Make OnReadyToBoot function match the prototype in UefiLib.h. The change
only affects gcc builds because EFIAPI ABI differs from the default ABI
when building with gcc.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com> Reviewed-by: Maurice Ma <maurice.ma@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17137 6f19259b-4bc3-4df7-8a09-765794883524
Scott Duplichan [Thu, 9 Apr 2015 03:09:17 +0000 (03:09 +0000)]
UefiCpuPkg: Avoid "error A2085" when DDK3790 tool chain is used
The DDK3790 tool chain fails when the PAUSE instruction is assembled:
error A2085: instruction or register not accepted in current CPU mode The solution is to use the .686 directive along with the .xmm directive.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Scott Duplichan <scott@notabs.org> Reviewed-by: Jeff Fan <jeff.fan@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17134 6f19259b-4bc3-4df7-8a09-765794883524
Tapan Shah [Tue, 7 Apr 2015 20:39:22 +0000 (20:39 +0000)]
ShellPkg: fix mv and cp command related issues
1.mv deletes file/directory when trying to move it to non-existing file system.
2.mv causes RSOD in system when trying to move same file at the same location.
3.Refactor mv and cp command with command name passed-in.
remove redundant move status error message when file failed to move across file system.
ShellPkg: UefiShellDebug1CommandsLib: fix hex string parsing in SETVAR
The ShellCommandRunSetVar() function calls the ShellIsHexOrDecimalNumber()
UefiShellLib function to determine if the data string in the SETVAR
command is a string of hexadecimal bytes.
However, ShellIsHexOrDecimalNumber() calls ShellConvertStringToUint64()
for validation. Therefore SETVAR rejects hex strings that cannot be
interpreted as UINT64 values due to range errors, despite the fact that
the hexadecimal data string of the SETVAR command is supposed to describe
an arbitrary array of bytes, rather than UINT64 values.
The internal library function InternalShellIsHexOrDecimalNumber() comes
close, as the first idea for the fix, however it is not quite right
either; it removes a leading minus sign, plus it allows 0x / 0X prefixes.
This would not be correct for the SETVAR command.
Instead, add a trivial utility function that is specific to the SETVAR
implementation. IsStringOfHexNibbles() accepts empty strings for
simplicity, but where we call it we have already ensured that the data
string is not empty (because that is handled earlier by the "delete
variable" branch of SETVAR). An even number of nibbles is also enforced
near the call site.
Elvin Li [Tue, 7 Apr 2015 03:33:07 +0000 (03:33 +0000)]
IntelFrameworkModulePkg: Put report status code after event was signaled per PI spec.
For PI spec vol3,
"EFI_SW_DXE_BS_PC_LEGACY_BOOT_EVENT The event with GUID EFI_EVENT_LEGACY_BOOT_GUID was signaled."
"EFI_SW_DXE_BS_PC_READY_TO_BOOT_EVENT The EFI_EVENT_GROUP_READY_TO_BOOT event was signaled."
However, in current code base, they are reported before events were signaled.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Elvin Li <elvin.li@intel.com> Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17124 6f19259b-4bc3-4df7-8a09-765794883524
Elvin Li [Tue, 7 Apr 2015 03:31:17 +0000 (03:31 +0000)]
MdeModulePkg: Put report status code after event was signaled per PI spec.
For PI spec vol3,
"EFI_SW_DXE_BS_PC_VIRTUAL_ADDRESS_CHANGE_EVENT The EVT_SIGNAL_VIRTUAL_ADDRESS_CHANGE event was signaled."
However, in current code base, it is reported before events were signaled.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Elvin Li <elvin.li@intel.com> Reviewed-by: Ruiyu Ni <ruiyu.ni@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17123 6f19259b-4bc3-4df7-8a09-765794883524
Copy below two library instances from IntelFrameworkModulePkg to MdeModulePkg. Then, Platform dsc can
refer to them from MdeModulePkg, and remove the dependency of IntelFrameworkModulePkg. The ones in
IntelFrameworkModulePkg are still kept for compatibility.
1. PeiDxeDebugLibReportStatusCode
2. LzmaCustomDecompressLib
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Liming Gao <liming.gao@intel.com> Reviewed-by: Star Zeng <star.zeng@intel.com> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17112 6f19259b-4bc3-4df7-8a09-765794883524
Ronald Cron [Thu, 2 Apr 2015 13:49:05 +0000 (13:49 +0000)]
EmbeddedPkg/Lan9118Dxe: Fix the reset after a receiver or transmitter error
The Lan9118 driver did not recover after a receiver error as the error
handling code stopped the transmitter but did not restart it. Added the
restart of the transmitter.
Added also the restart of the receiver after a transmitter error and
the reactivation of the LEDs after all resets.
Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Ronald Cron <ronald.cron@arm.com> Reviewed-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17106 6f19259b-4bc3-4df7-8a09-765794883524
Felix Polyudov [Thu, 2 Apr 2015 06:40:02 +0000 (06:40 +0000)]
MdePkg: Missing GUID variable in DiskInfo.h
The DiskInfo.h defines EFI_DISK_INFO_NVME_INTERFACE_GUID, but does not declare a corresponding EFI_GUID variable.
The attached patch adds "extern" statement for the variable in DiskInfo.h. It also adds variable definition to MdePkg.dec.
projects / mirror_edk2.git / log