mdkinney [Sat, 18 Dec 2010 00:22:08 +0000 (00:22 +0000)]
Remove CONSTRUCTOR from DxePcdLib aligning the DxePcdLib design with the PeiPcdLib and reduces the number of CONSTRUCTORs that have to be run for most DXE modules.
mdkinney [Fri, 17 Dec 2010 22:47:06 +0000 (22:47 +0000)]
Remove UefiDriverEntryPoint from [LibraryClasses] section. Library implementations should never link against an entry point lib. Only modules implementation should link against an entry point lib.
niruiyu [Fri, 17 Dec 2010 05:55:47 +0000 (05:55 +0000)]
Removes the assumption in Supported() and Start() routine that the EFI_SIO_PROTOCOL is installed in the immediate child handle of the PCI controller. Multiple SIO platform could have an additional layer between the PCI controller and the EFI_SIO_PROTOCOL instance.
rsun3 [Fri, 17 Dec 2010 02:38:26 +0000 (02:38 +0000)]
Fix a bug that the size of a gate descriptor in the IDT is 8 bytes. The size is 8 bytes in 32-bit mode, while it is 16 bytes in 64-bit mode. Replace the hard-coded size with a sizeof.
andrewfish [Thu, 16 Dec 2010 21:44:49 +0000 (21:44 +0000)]
Add support for ARM MOVW/MOVT instructions that were added in the latest PE/COFF specification. Currently they are not hooked in as we need to wait for the tools to get updated. Tools are needed to convert ELF to PE/COFF and to refixup FVs.
andrewfish [Wed, 15 Dec 2010 00:58:34 +0000 (00:58 +0000)]
Fix issue with CodeSourcery gcc requiring assembly functions to have a .type define inorder to support interworking (calling from thumb(C code) to ARM code (hadn written assembly). We had to add a macro as the .type directive is not supported, or needed, by Xcode. All ARM .S files need to be updated.
rsun3 [Tue, 14 Dec 2010 09:52:38 +0000 (09:52 +0000)]
Add special treatment for EFI_PCI_IO_ATTRIBUTE_EMBEDDED_DEVICE and EFI_PCI_IO_ATTRIBUTE_EMBEDDED_ROM, because these attributes are only for PCI IO, while not supported by PCI Root Bridge IO.
niruiyu [Tue, 14 Dec 2010 07:17:55 +0000 (07:17 +0000)]
Fix the IsaIo driver to not create ISA resource item when the ACPI resource returned from EFI_SIO_PROTOCOL::GetResources() is 0.
This could avoid creating the ISA resource item with StartRange = 0 and EndRange = 0xffffffff when the ACPI resource Base and Len both are 0 because we set StartRange = Base and EndRange = Base + End - 1.
erictian [Tue, 14 Dec 2010 02:50:56 +0000 (02:50 +0000)]
update USB_PORT_STAT_OWNER from 2^11 to 2^13 which is added as a possible value of EFI_USB_PORT_STATUS and is the equivalent of EfiUsbPortOwner of EFI_USB_PORT_FEATURE
erictian [Tue, 14 Dec 2010 02:50:25 +0000 (02:50 +0000)]
update USB_PORT_STAT_OWNER from 2^11 to 2^13 which is added as a possible value of EFI_USB_PORT_STATUS and is the equivalent of EfiUsbPortOwner of EFI_USB_PORT_FEATURE
rsun3 [Tue, 14 Dec 2010 02:33:15 +0000 (02:33 +0000)]
Update the PCI bus driver to correctly set the EFI_PCI_IO_ATTRIBUTE_EMBEDDED_ROM attribute of PCI IO. Per UEFI spec, this attribute is used to determine if the ROM image is from the ROM BAR or from a platform specific location.
rsun3 [Thu, 9 Dec 2010 08:02:34 +0000 (08:02 +0000)]
when SMM Communication Protocol.Communicate() is called from within SMM, SMRAM is accessible so it is not necessary to check the SMRAM open/lock state via SMM ACCESS Protocol.
ydong10 [Thu, 9 Dec 2010 02:18:44 +0000 (02:18 +0000)]
Co-work with patch V11137 to refine the Hii menu.
1. Move the Mac address string from the form title to form title help for the driver Ipv4 and VLAN.
2. Change the menu show in the device manager from "Network Controller Information" to "ABC Information Sample".
ydong10 [Thu, 9 Dec 2010 01:44:07 +0000 (01:44 +0000)]
[BDS] Device Manager was requested to add more sub-level menu about network device.
Now show menu like:
Network Device List -> Mac:XX:XX:XX... -> IPv4 Network Config
Mac:XX:XX:XX... VLAN Configuration
andrewfish [Wed, 8 Dec 2010 19:58:48 +0000 (19:58 +0000)]
Move away from 0x40 block size (worked around a DXE core bug assuming the FV header fit in a block) of the FV to 1 byte. This makes more sense as the FV is shadowed from FLASH into RAM by HW so it is never on top of a real block device.
vanjeff [Wed, 8 Dec 2010 06:58:33 +0000 (06:58 +0000)]
add note for EFI_SW_PEI_EC_INVALID_CAPSULE_DESCRIPTOR and EFI_SW_PEI_PC_RECOVERY_BEGIN to express they are different with the current PI specification 1.2.
andrewfish [Wed, 8 Dec 2010 00:14:23 +0000 (00:14 +0000)]
Add vpush/vpop to the exception handler as we added CopyMem/SetMem that can use NEON registers to speed things up. This means we need to save state so we don't blow up a CopyMem.
andrewfish [Tue, 7 Dec 2010 22:44:08 +0000 (22:44 +0000)]
Older tools rounded down the size of the FV, and the new tools round up. Our main FV region is not a multiple of the block size, but it is set to fill the space. Probably need to move to a block size of 1 at some point.
mdkinney [Mon, 6 Dec 2010 21:55:14 +0000 (21:55 +0000)]
1) Change location of DEBUG() macro that prints the name of the FFS file being evaluated by the DXE Dispatcher to eliminate a duplicate print.
2) Change location of DEBUG() macro that prints the name of the FFS file being evaluated by the SMM Dispatcher to eliminate a duplicate print.
3) Remove all references and processing of SOR from the PiSmmCore. Treat use of SOR in the dependency expression of an SMM Driver as an invalid depex opcode.
4) Remove all references for an SMM Driver being in the "Untrusted" state from the PiSmmCore.
jljusten [Mon, 6 Dec 2010 06:06:35 +0000 (06:06 +0000)]
MdePkg: Fix GCC Thunk support for IA32
In r11047 Ia32/Thunk.S was modified. However, the current
implemention of Ia32/Thunk.S has contant values in the code
which must be updated if the code is changed.
This change updates those constants for the r11047 change.
jljusten [Mon, 6 Dec 2010 06:04:38 +0000 (06:04 +0000)]
DuetPkg, MdeModulePkg: Fix variable services hang with GCC44 X64
The C ellipses parameters are passed to functions differently
by default with GCC 4.4. To make sure they are properly sent to
VariableGetBestLanguage, we add 'EFIAPI' to this function.
andrewfish [Tue, 30 Nov 2010 23:38:40 +0000 (23:38 +0000)]
Adding Simple Pointer, GOP, SimpleTextInEx, and Networking protocols to the emulator. Cleaned up POSIX include situation by centralizing it in a single file, like NT32. Fixed TPL issue with TPL High not being emulated correctly, it was possible to take a timer tick when the locks in the DXE core should have prevented this. Remove some unused files to make things easier to maintain.
rsun3 [Mon, 29 Nov 2010 03:30:38 +0000 (03:30 +0000)]
Add a new Timer Library instance SecPeiDxeTimerLibUefiCpu into UefiCpuPkg. This library differs with the SecPeiDxeTimerLibCpu library in the MdePkg in that it uses the local APIC library so that it supports x2APIC mode.
niruiyu [Wed, 24 Nov 2010 03:23:30 +0000 (03:23 +0000)]
Fix AutoUpdateLangVariable() logic to handle the case PlatformLang/Lang is set before PlatformLangCodes/LangCodes.
Pre-allocate pool for runtime phase.
niruiyu [Wed, 24 Nov 2010 03:04:20 +0000 (03:04 +0000)]
Fix AutoUpdateLangVariable() logic to handle the case PlatformLang/Lang is set before PlatformLangCodes/LangCodes.
Pre-allocate pool for runtime phase.
vanjeff [Tue, 23 Nov 2010 07:51:00 +0000 (07:51 +0000)]
1. Rollback the changing on replacing MAX_EXTENDED_DATA_SIZE by EFI_STATUS_CODE_DATA_MAX_SIZE,
use MAX_EXTENDED_DATA_SIZE as before.
2. Use DEBUG error message instead of ASSERT(FASLE) when extended data is too large
3. Expand 1 for buffer array size to avoid potential issue.
vanjeff [Tue, 23 Nov 2010 07:50:31 +0000 (07:50 +0000)]
1. Rollback the changing on replacing MAX_EXTENDED_DATA_SIZE by EFI_STATUS_CODE_DATA_MAX_SIZE,
use MAX_EXTENDED_DATA_SIZE as before.
2. Use DEBUG error message instead of ASSERT(FASLE) when extended data is too large
3. Expand 1 for buffer array size to avoid potential issue.
rsun3 [Mon, 22 Nov 2010 07:16:02 +0000 (07:16 +0000)]
The PCI HotPlug Init Protocol returns a list of Root HPC, but it is possible that some HPCs in the list are not found during PCI enumeration. In this case, the PCI HotPlug Init Protocol.InitializeRootHpc() will not be called for the un-found HPCs. However, in AllRootHPCInitialized(), all reported HPCs are checked to see if it has been initialized, so TIME_OUT happens.
This check-in fix the issue by adding a flag to indicate if a HPC is found during PCI enumeration or not. This check-in also adds a debug message if any found HPC failed with initialization.
niruiyu [Mon, 22 Nov 2010 05:25:03 +0000 (05:25 +0000)]
Use PcdResetOnMemoryTypeInformationChange in MdeModulePkg to control whether to reset system when memory type information changes.
Check memory type information after signaling the ReadyToBoot event.
projects / mirror_edk2.git / log