When using InOsEmuPkg/Unix/build64.sh -D USE_NEW_SHELL,
the UEFI shell binary will now be included in the firmware
image.
Signed-off-by: jljusten
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11889 6f19259b-4bc3-4df7-8a09-765794883524
Images that can't be loaded with dlopen are
added to SecMain.gdb.
Whenever gdb stops, the hook-stop macro will automatically
load the symbol files.
Signed-off-by: jljusten
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11888 6f19259b-4bc3-4df7-8a09-765794883524
I've got this working. The tricky part was a driver/application can mix malloc forms. If the driver links against this library all the allocations come from the OS service, but if a protocol or EFI boot service return an allocated buffer it will still use the EFI allocation. The gasket code can detect this and do the right thing, as the emulator is passing up the memory map to EFI.
You can now use the OS based malloc debugging tools in an EFI driver or appliaction. On OS X this is libgmalloc, also called guard malloc.
Signed-off-by: andrewfish.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11886 6f19259b-4bc3-4df7-8a09-765794883524
This is a fix for XCODE/XCLANG. I'm still seeing an issue, but the stat info for the filesystem is now correct.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11885 6f19259b-4bc3-4df7-8a09-765794883524
Added OS malloc and free so we can make MemoryAllocationLib instance that uses OS guard malloc. This will allow all the debug support built into the OS for finding malloc bugs to be used with a driver in the emulator.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11883 6f19259b-4bc3-4df7-8a09-765794883524
For IA32 builds, if /lib32 and /usr/lib32 exist,
they will be used instead of /lib and /usr/lib.
For X64 builds, if /lib64 and /usr/lib64 exist,
they will be used instead of /lib and /usr/lib.
Signed-off-by: jljusten
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11879 6f19259b-4bc3-4df7-8a09-765794883524
Start using MdeModulePkg MetronomeDxe driver. Add PEI and DXE_CORE TimerLibs that just call back to the emulator. Add a DXE/UEFI TimerLib that also does a gBS->WaitForEvent() if the stall is for a long period of time. Change the Thunk API for Sleep to nanoseconds.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11872 6f19259b-4bc3-4df7-8a09-765794883524
This implementation only covers a few BaseMemoryLib functions.
The implementation leverates malloc/free in stdlib.
Signed-off-by: jljusten
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11866 6f19259b-4bc3-4df7-8a09-765794883524
Turning on serial prints for early drivers, turned off debug prints for DXE/UEFI drivers. Add them back in via report status code.
signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11857 6f19259b-4bc3-4df7-8a09-765794883524
At this point IA-32 emulator boots the shell with an X11 GOP window. Block IO seems to have an issue that needs to be debugged.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11854 6f19259b-4bc3-4df7-8a09-765794883524
There are still issues with the IA-32 gasket code. Mostly having to do with yank put bugs not accounting for UINT64 returns and arguments. Currently the IA-32 build crashes in the timer callback since it is UINT64, function ptr but the gasket is UINTN UINTN.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11853 6f19259b-4bc3-4df7-8a09-765794883524
Wrote IA-32 (align stack to 16-bytes) gaskets for emulator and got it to compile for IA-32. TempRam switch code is not ported to IA-32 in Sec. Code crashes in Sec.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11849 6f19259b-4bc3-4df7-8a09-765794883524
Added a CpuSleep () API to the Emulator Thunk. We needed to do this as the Stall() works hard to not get broken by the timer tic (POSIX signal). nanosleep() gets interrupted by the timer signal so it is a good emulator of a CpuSleep(); I was also able to remove some stalls in the X11 keyboard and mouse checking events, now that the gIdleLoopEventGuid was added.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11846 6f19259b-4bc3-4df7-8a09-765794883524
Devices get the block size via ioctl, but for a file the block size needs to be set. Default to 512, but optionally allow other values, like 2048/0x800 for ISO CD-ROM images. Also updated the comments in .DSC and .DEC files.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11843 6f19259b-4bc3-4df7-8a09-765794883524
BerkelyPacketFilter seems to work. Some one else will need to use the BerkeleyPacketFilter as a template and implement the LinuxPacketFilter. I don't think it will be that hard, as they are simular.
Signed-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11828 6f19259b-4bc3-4df7-8a09-765794883524
This only adds the EFI parts and not the OS specific implementation. Since the EFI parts no longer use system include files this is safe to do. Also fix two minor typos.
Signed-off-by:andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11781 6f19259b-4bc3-4df7-8a09-765794883524
Update the InOsEmuPkg to properly function with XIP. Make the Recovery FV read only. Remove the use of global variable writes from XIP code. Add a new global page that can be used in place of writting to the FD by XIP code. Think of this global page as a system SRAM.
igned-off-by: andrewfish
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11771 6f19259b-4bc3-4df7-8a09-765794883524
The current scheme maps a file into a memory buffer. You can write to the memory buffer, but if you restart the emulator the data is not written back to the FD. Since the build, rebuilds the FD it will blow away any changes. You can now set variables and restart the emulator and they are still set. There is a bug if the FV is set to write protect that still needs to be tracked down.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11761 6f19259b-4bc3-4df7-8a09-765794883524
The following can be defined on the build command line:
* BUILD_NEW_SHELL: Build & run the UefiShell
* USE_NEW_SHELL: Use the pre-build UefiShell binary
* BUILD_OLD_SHELL: Use the GCC shell binary previously built
For the FAT driver:
* BUILD_FAT: Build the fat driver from source
Signed-off-by: jljusten
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@11652 6f19259b-4bc3-4df7-8a09-765794883524
jljusten
andrewfish