-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Controller/BUILD: Cannot add controller tests for certain platforms (Darwin, clang) #9173
Comments
@yunhanw-google Why do we have these undefined symbols? |
You need to add these files to XCode project. |
Not sure if I understood. The build failure happens for clang as well (if I include the controller tests for all platforms), not just Darwin. Which files/libs need to be added to the XCode project? Can you explain why that would fix this problem? |
Seems I misunderstood your need, can you try adding |
Tried that on a branch with this change. Build on Darwin succeeds, but then all-clusters-app build starts to fail on esp32, nrfconenct, qemu with the following type of error. Full logs: https://github.com/sharadb-amazon/connectedhomeip/runs/3413744436 This happens only when I make the above change.
|
Hold on, why would we need to add src/controller/data_model ? |
It looks like #8824 merged. @sharadb-amazon is there still a problem here? If so, what are the steps to reproduce? |
It looks like #8824 merged by just not building these tests on various OSes, and #9374 disabled them on still others where they were failing. Fixing the build on Darwin here is pretty simple: just use the right spelling of
where |
1) Modify our CHIPClientCallbacks template to suppress the "possibly unbounded stack" warning from a VLA we know we want to remove. 2) Don't try to link in BLELayer in the TestDevice test if BLE networking is disabled (so that BLELayer symbols don't actually exist). 3) Heap-allocate the huge FabricTable so we don't run into compile errors due to a stack frame being too big in TestDevice. 4) Fix the fact that SystemLayerImplSelect did not default-initialize mDispatchQueue. This matters because TestDevice does weird things with a separate SystemLayer that the PlatformManager does not know about, nor does it call InitChipStack in any case. The tests are still disabled on nrfconnect because I could not get them to link there so far. Fixes project-chip#9173
1) Modify our CHIPClientCallbacks template to suppress the "possibly unbounded stack" warning from a VLA we know we want to remove. 2) Don't try to link in BLELayer in the TestDevice test if BLE networking is disabled (so that BLELayer symbols don't actually exist). 3) Heap-allocate the huge FabricTable so we don't run into compile errors due to a stack frame being too big in TestDevice. 4) Fix the fact that SystemLayerImplSelect did not default-initialize mDispatchQueue. This matters because TestDevice does weird things with a separate SystemLayer that the PlatformManager does not know about, nor does it call InitChipStack in any case. The tests are still disabled on nrfconnect because I could not get them to link there so far. Fixes project-chip#9173
1) Modify our CHIPClientCallbacks template to suppress the "possibly unbounded stack" warning from a VLA we know we want to remove. 2) Don't try to link in BLELayer in the TestDevice test if BLE networking is disabled (so that BLELayer symbols don't actually exist). 3) Heap-allocate the huge FabricTable so we don't run into compile errors due to a stack frame being too big in TestDevice. 4) Fix the fact that SystemLayerImplSelect did not default-initialize mDispatchQueue. This matters because TestDevice does weird things with a separate SystemLayer that the PlatformManager does not know about, nor does it call InitChipStack in any case. The tests are still disabled on nrfconnect because I could not get them to link there so far. Fixes project-chip#9173
1) Modify our CHIPClientCallbacks template to suppress the "possibly unbounded stack" warning from a VLA we know we want to remove. 2) Don't try to link in BLELayer in the TestDevice test if BLE networking is disabled (so that BLELayer symbols don't actually exist). 3) Heap-allocate the huge FabricTable so we don't run into compile errors due to a stack frame being too big in TestDevice. 4) Fix the fact that SystemLayerImplSelect did not default-initialize mDispatchQueue. This matters because TestDevice does weird things with a separate SystemLayer that the PlatformManager does not know about, nor does it call InitChipStack in any case. The tests are still disabled on nrfconnect because I could not get them to link there so far. Fixes project-chip#9173
…9638) 1) Modify our CHIPClientCallbacks template to suppress the "possibly unbounded stack" warning from a VLA we know we want to remove. 2) Don't try to link in BLELayer in the TestDevice test if BLE networking is disabled (so that BLELayer symbols don't actually exist). 3) Heap-allocate the huge FabricTable so we don't run into compile errors due to a stack frame being too big in TestDevice. 4) Fix the fact that SystemLayerImplSelect did not default-initialize mDispatchQueue. This matters because TestDevice does weird things with a separate SystemLayer that the PlatformManager does not know about, nor does it call InitChipStack in any case. The tests are still disabled on nrfconnect because I could not get them to link there so far. Fixes #9173
Problem
This PR (#8824) attempts to add tests for src/controller: https://github.com/project-chip/connectedhomeip/pull/8824/files#diff-8b9a0a48a2fb362c82aef9724b90a5476c4b80358995f69c83324546c0c96648
This builds with gcc, but fails for clang, Darwin with the following errors:
Expectation is that src/controller/tests can be built and run for all platforms.
Proposed Solution
TBD
The text was updated successfully, but these errors were encountered: