Processing metadata -> Failed (no information)

[aaronjamt]

I'm trying to compile this project but, after cloning EDK2 (with --recurse-submodules), cloning the project into the EDK2 directory, running source edksetup.sh, and trying to build, all I get is the following:

Build environment: Linux-6.0.0-kali3-amd64-x86_64-with-glibc2.36
Build start time: 17:17:22, Nov.27 2022

WORKSPACE        = .../edk2
EDK_TOOLS_PATH   = .../edk2/BaseTools
CONF_PATH        = .../edk2/Conf
PYTHON_COMMAND   = /usr/bin/python3


Processing meta-data 
Architecture(s)  = X64
Build target     = RELEASE
Toolchain        = VS2015x86

Active Platform          = .../edk2/FrameworkHacksPkg/FrameworkHacksPkg.dsc
... 

- Failed -
Build end time: 17:17:25, Nov.27 2022
Build total time: 00:00:03

There is no information about what's wrong and the closest discussion to this issue I could find (#3629) was on Windows using Python 2 (but I'm using Kali Linux with Python 3.10). Can anyone help?

Edit: I also tried to build https://github.com/lersi/edk2_project_template (linked in #3629) and had the same issue.

[lersi]

first of all, sorry that my project did not help :(

have you built basetools, before running build command?
you can compile basetools by using the command make -C edk2/BaseTools

you can contact me trough my email, for discussing this issue farther.

[lersi]

I have tried to compile this project using my build system, and it currently works on linux, as it can be seen by this workflow run (release 0.0.4)

[aaronjamt]

I just ran make -C BaseTools from the edk2 directory and it definitely compiled some stuff but I ran build again and got the exact same error.

[ajfish]

Pro Tip: If you want prompt help copy the error message into the radar so people don’t have to invest lots of time tracking down the information….. Mac Build Failure: Undefined symbols for architecture x86_64: 55 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:56> "_memset", referenced from: 56 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:57> 57 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:58> _cmd_version in ECTool.lib(ECTool.obj) 58 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:59>build.py... 59 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:60> _cmd_console in ECTool.lib(ECTool.obj) 60 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:61> : error 7000: Failed to execute command 61 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:62> _cmd_reflash in ECTool.lib(FWUpdate.obj) 62 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:63> make tbuild [/Users/runner/work/Compile-FrameworkHacksPkg/Compile-FrameworkHacksPkg/build/Build/FrameworkHacksPkg/RELEASE_XCODE5/X64/FrameworkHacksPkg/Application/ECTool/ECTool] 63 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:64> _flash_read in ECTool.lib(Flash.obj) 64 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:65> _flash_write in ECTool.lib(Flash.obj) 65 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425850#step:8:66>A memset link failure is usually assigning to zero a variable bigger thank 64-bits. If you compiled with optimizations disable (-O0 compiler flag) you always fail like this. With the normal size optimization (-Os) it is possible that the optimizer will optimize away the memset and the code will compile. This exposes you do different behavior between different compilers, and even different compiler versions. Generally how I’d debug this is to: Look at the assembler to see the call to memset and then track that back to the code. $ lldb ECTool.obj (lldb) dis -m -n cmd_version But it is likely just some local variable (like a struct) getting set to zero in the cmd_console() command. Windows error: D:\a\Compile-FrameworkHacksPkg\Compile-FrameworkHacksPkg\uefi_apps\FrameworkHacksPkg\Include\Library/CrosECLib.h(597): error C2143: syntax error: missing ')' before '(' [D:\a\Compile-FrameworkHacksPkg\Compile-FrameworkHacksPkg\build\FrameworkHacksPkg.vcxproj] 45 <https://github.com/lersi/Compile-FrameworkHacksPkg/actions/runs/3565363184/jobs/5990425635#step:6:46>This looks like a bug in some include file? Thanks, Andrew Fish

…

On Nov 28, 2022, at 7:53 AM, Aaronjamt ***@***.***> wrote: I just ran make -C BaseTools from the edk2 directory and it definitely compiled some stuff but I ran build again and got the exact same error. — Reply to this email directly, view it on GitHub <#3666 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD4ZCO27LH5OEZNPBSHFZH3WKTIOFANCNFSM6AAAAAASMZ3NW4>. You are receiving this because you are subscribed to this thread.

[aaronjamt]

There is no error information... that's why I'm asking here. Initially I had some errors and I was able to resolve them myself (well, with Google) but there is 0 information about what's wrong. Is there some debugging/verbose flag(s)? I tried --verbose and it adds some information but nothing resembling an error.

[lersi]

thank you for the help, my solution was a quick and hasty work and the propose was to show it can be compiled on Linux (and easily).

[aaronjamt]

Turns out I had to add -t GCC5 to build with GCC5 as the toolchain instead of VS2017.