Make QDTE usable headless and cross-platform#5
Open
Igor Opaniuk (igoropaniuk) wants to merge 7 commits into
Open
Make QDTE usable headless and cross-platform#5Igor Opaniuk (igoropaniuk) wants to merge 7 commits into
Igor Opaniuk (igoropaniuk) wants to merge 7 commits into
Conversation
The NON-HLOS.bin FAT browser is GUI-only, yet controller.py imported it unconditionally just to resolve the DTGUIController base class. That pulled fs/pyfatfs into every headless run and made --nogui fail to import on minimal sysroots (Yocto native, CI) that have no reason to ship those packages. Import it lazily and stub the base class when absent. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
QDTE imports tkinter at module load across many files and defines widget subclasses at class-definition time, so merely importing the controller requires Tcl/Tk -- even when no window is ever created. That made --nogui unusable on headless hosts (Yocto native sysroots, CI containers) where Tk is not present. Install a permissive tkinter stub before any QDTE module loads when --nogui is requested; GUI mode is untouched. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
flags.global_info only defines the "quts2" install path on Windows, but the startup probe dereferenced it unconditionally. On Linux that raised KeyError before any QDTE logic ran, so the tool could not start at all. Check for the key before using it. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
QDTE was written against an internal pyfdt fork whose to_dtb() takes an out-dict of node byte-offsets for the GUI hex viewer. The only published pyfdt (0.3) defines to_dtb(self), so every consumer outside Qualcomm's toolchain hit a TypeError on the first DTB save. The offsets are a GUI-only convenience and unused headless, so fall back to the zero-arg call when the dict argument is rejected. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
FdtPropertyWords edits cast each token with int(val), which only accepts decimal. Callers naturally pass 0x.. literals when mirroring fdtdump output or packing big-endian uint32 blobs (e.g. certificate injection), and those raised ValueError. int(val, 0) keeps decimal working while auto-detecting a prefixed base. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
GUI startup hid the console window through ctypes.windll, which exists only on Windows. On other platforms the attribute access raised AttributeError and the window never opened, so the GUI was effectively Windows-only. Skip the call when windll is unavailable. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
The --nogui command-line flow is what downstream automation (Yocto image builds, signing pipelines) relies on, yet nothing guarded it, and the planned GUI/headless split is intrusive enough to break it silently. Exercise run.py --nogui end to end on real xbl_config.elf and uefi_dtbs.elf containers and read the edited property back from the rebuilt ELF with fdtdump, so a broken contract fails CI before it reaches consumers. Binaries come from public Qualcomm Software Center archives and QA artifactory. Signed-off-by: Igor Opaniuk <igor.opaniuk@oss.qualcomm.com>
Author
|
Btw, I don't know what commit message convention this downstream project follows, so I went with Conventional Commits (the If that doesn't match your house style, just let me know and I'll reword the commits. |
Author
|
Yeping-Song Neeraj Jetha (@njjetha) let me know if you have any comments/objections. thanks! |
|
Igor Opaniuk (@igoropaniuk) commit-msg-check doesn't follow Conventional Commits it just make sure that commit has |
Author
|
Neeraj Jetha (@njjetha) current commit messages already comply with it. I was just curious whether the team working on QDTE has any preferences. |
|
Igor Opaniuk (@igoropaniuk) I am not sure if they have any standardized format for the commit-msg, but basic commit-msg check are already met. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This series makes QDTE's
--noguiflow run reliably on Linux/macOS hosts, minimal Yocto native sysroots, and CI containers - which is what downstream automation (image builds, signing pipelines) actually depends on.The headless fixes drop mandatory GUI-only dependencies (fs/pyfatfs, Tcl/Tk) and stop the startup path from dereferencing Windows-only state (the QUTS install path), so importing and running the tool no longer requires a desktop environment or the internal pyfdt fork.
The remaining fixes restore correctness for real-world use: zero-arg
to_dtb()compatibility with public pyfdt 0.3, hex/oct/bin literals in --modify word values, and launching the GUI on non-Windows platforms.Finally, a Github workflow definition added in ci: add headless --nogui smoke tests exercises the full --nogui flow end to end against real xbl_config.elf/uefi_dtbs.elf containers, mitigating the risk of silent regressions - especially as more intrusive changes (e.g. the planned GUI/headless split) are on the roadmap.