eric_tian [Mon, 30 Jun 2008 15:01:15 +0000 (15:01 +0000)]
provide a temporary method to generate SecMain executable file.
[Background]
The SecMain's generation is difference than other modules. It depends on some standard lib. At present, we only have extended build option mechanism in module build. It can not support this special case. Then provide this script to temporarily build SecMain
eric_tian [Mon, 30 Jun 2008 06:25:43 +0000 (06:25 +0000)]
Provide source level debug ability in *unix environment.
[Background]
The UnixPkg provide an simulator in *unix. To support source level debug, we need force all modules be built as ELF shard objects and dynamically load all of modules using dlopen(). For peicore, peim and dxecore, we provide a wrapper to provide PeCoffLoader interface
eric_tian [Mon, 30 Jun 2008 05:50:49 +0000 (05:50 +0000)]
make change to support UnixPkg build. The changes are listed as follows:
1. change ASM_PFX() macro, the underscore preceding to function symbol don't be added in Linux.
2. move some illegal characters in .S file.
3. change ELFGCC flag in tools_def.template to support Linux&ELFGCC build
eric_tian [Mon, 30 Jun 2008 05:08:49 +0000 (05:08 +0000)]
Port EdkUnixPkg to UnixPkg. The changes are listed as follows:
1. change *.msa to *.inf, and create platform configuration files .dec&.dsc&.fdf to comply with Edk2 build process
2. using PCD mechanism to replace macro.
3. change Sec code to cowork with PI1.0 Pei Core and produce temparory memory ppi.
vanjeff [Wed, 25 Jun 2008 05:50:41 +0000 (05:50 +0000)]
[Description]:
Sync 3 trackers.
1. A bug that may cause the low/full speed device behind 2.0 hub will not work.
2. MajorRevision and MinorRevision need updating.
3. Misuse UhciConvertPollRate in UHCI driver.
[Impaction]:
EhicDxe, UhciDxe and UsbBusDxe.
[Reference Info]:
EDK tracker 1122 - A bug that may cause the low/full speed device behind 2.0 hub will not work .
EDK tracker 1101 - Phoenix found issue in USB and Misuse UhciConvertPollRate in UHCI driver.
qhuang8 [Wed, 25 Jun 2008 02:52:43 +0000 (02:52 +0000)]
Fix an issue in the implementation of GetDriverName() in Con splitter driver. We need to compare "This" paramter with all 5 protocol instances to decide whether it stands for CN or CN2.
vanjeff [Tue, 24 Jun 2008 08:36:11 +0000 (08:36 +0000)]
[Description]:
Misuse of retransmisson of DHCPDISCOVER in Dhcp4 drive, the root cause is that DHCP doesn't retransmit discover message when timeout.
[Solution]:
Correct the mechanism for discover message.
[Impaction]:
Dhcp4Dxe.
[Reference Info]:
EDK tracker 1150 - Misuse of retransmisson of DHCPDISCOVER in Dhcp4 drive.
vanjeff [Tue, 24 Jun 2008 08:13:58 +0000 (08:13 +0000)]
[Description]:
The Usb1.1 flash disk will hang for Bootable Image Support Test -> Bloack IO protocol Test -> Reset_Func.
[Solution]:
Add a stall 20ms after set address when port reset for Usb hub.
[Impaction]:
UsbBusDxe.
[Reference Info]:
EDK tracker 1120 - USB flash disk hang or fail in the new SCT blockIo test.
klu2 [Tue, 24 Jun 2008 07:14:18 +0000 (07:14 +0000)]
Fix doxygen issue:
1) comment for structure should be in doxygen format, such as /// , /**, /*!
2) comment for data field of structure should in doxygen format.
3) should not exist blank between ") (" when declaring a type of function point.
klu2 [Tue, 24 Jun 2008 04:37:24 +0000 (04:37 +0000)]
Fix doxygen issue:
1) comment for structure should be in doxygen format, such as /// , /**, /*!
2) comment for data field of structure should in doxygen format.
3) should not exist blank between ") (" when declaring a type of function point.
klu2 [Tue, 24 Jun 2008 04:01:35 +0000 (04:01 +0000)]
Fix issue broken doxygen parsing. The issue is that when declaring a function pointer type, the "(" for parameter should follows ")" of function type name without any blank, for example:
Bad: "typedef EFI_STATUS (EFIAPI name) (parameter list)"
GOOD: "typedef EFI_STATUS (EFIAPI name)(parameter list)"
qwang12 [Mon, 23 Jun 2008 09:41:33 +0000 (09:41 +0000)]
Merged in the following trackers from EDK:
EDK1145 Cursor mising in shell in some case
EDK1099: Dell - [HII] HiiGetFontInfo() not retrieve the system font by FoFontInfoMask
EDK1127: [UEFI 2.10] Keyboard layout support
EDK1129: [UEFI HII] GUID is represented wrongly in Config String
And some other fixes such as
*[UEFI HII] HiiGetAltCfg is generating "Name=" sub string in the wrong format
*UEFI HII: GetUnicodeStringTextOrSize() doesn't handle NULL StringDest properly
*GetFontInfo() need be updated to avoid iteration
*HIIStringProtocolTest failed on multiple platform
*[Uefi 2.1] Comply with latest Hii ECR
* GetFontInfo() need be updated to avoid iteration
qwang12 [Mon, 23 Jun 2008 09:38:38 +0000 (09:38 +0000)]
Merged in the following trackers from EDK:
EDK1145 Cursor mising in shell in some case
EDK1099: Dell - [HII] HiiGetFontInfo() not retrieve the system font by FoFontInfoMask
EDK1127: [UEFI 2.10] Keyboard layout support
EDK1129: [UEFI HII] GUID is represented wrongly in Config String
And some other fixes such as
*[UEFI HII] HiiGetAltCfg is generating "Name=" sub string in the wrong format
*UEFI HII: GetUnicodeStringTextOrSize() doesn't handle NULL StringDest properly
*GetFontInfo() need be updated to avoid iteration
*HIIStringProtocolTest failed on multiple platform
*[Uefi 2.1] Comply with latest Hii ECR
* GetFontInfo() need be updated to avoid iteration
qwang12 [Thu, 19 Jun 2008 07:05:44 +0000 (07:05 +0000)]
1) Add in EBC build support.
2) Add FvInfoToFvHobThunk in the DSC file.
3) Remove EdkCompatibilityPkg/Foundation/Library/Pei/PeiLib/PeiLib.inf from [Components.Common] as it only support IA32, X64 and IPF for now.
qwang12 [Wed, 11 Jun 2008 05:26:24 +0000 (05:26 +0000)]
remove reference to the obsolete EFI_LIGHT_ISA_IO_PROTOCOL. This protocol is introduced for size reduction. But later this method is obsoleted as EDK II use feature flags to conditional mask off some code. As EDK II core code does not produce this protocol so we remove all reference to this protocol.
qwang12 [Thu, 5 Jun 2008 06:19:12 +0000 (06:19 +0000)]
Install a Framework SetupBrowser protocol interface. But this interface will return EFI_UNSUPPORTED. This is to satisfy that some platform/chipset code may located this protocol but never invoke either of these two interfaces in the protocol.
qwang12 [Thu, 5 Jun 2008 01:57:08 +0000 (01:57 +0000)]
CHAR8 is defined as "1-byte Character" in UEFI spec, but not "unsigned Character". Most compiler treats "char" as signed unless overide by compiler options. Most compiler generate warnings for statement like "CHAR8 *S = "Hello"". In addition, we have defined CHAR8 as char for processor arch such as IA32, X64 and IPF in EDK II.
qwang12 [Wed, 4 Jun 2008 10:03:44 +0000 (10:03 +0000)]
Add a type cast of UINT32 to insure that EBC compiler does not report warning of "Integer conversion resulted in a change of sign" when building with -Za option (accept only ANSI/ISO standard C code). 1 << 31 is consider to be a sign change by compiler.
qwang12 [Thu, 29 May 2008 07:12:06 +0000 (07:12 +0000)]
1) add in DeviceIoToPciRootBridgeIoThunk
2) Change BaseMemoryLib instance to BaseMemoryLib|MdePkg/Library/BaseMemoryLib/BaseMemoryLib.inf so that we can verify the IPF build for Compatibility modules.
vanjeff [Tue, 27 May 2008 03:34:39 +0000 (03:34 +0000)]
[Description]:
Problem with PXE boot to Windows Server 2008 install. The reason is that UdpRead and UdpWrite interfaces cannot work well with the same UDP instance.
[Solution]
Use separate instances for UDP Read and UDP write in UefiPxeBC module.
[Impaction]:
UefiPxeBcDxe and Udp4Dxe module.
[Reference Info]:
EDK tracker 1133 - Problem with PXE boot to Windows Server 2008 install.