Reverted ULONG to be 4 bytes long, fixing missaligment issue in the current port#534
Conversation
|
@akifejaz Please review. |
Sure, I'll review shortly. Hi @cpdpls, there is PR which enables regression for risc-v ports. It has bunch of tests that runs on both RV32 and RV64. It would be nice if you can try running those. here is how you can run tests.. you would need to merge the upstream/dev branch into yours before running tests. |
|
Hi @akifejaz That was literally the question I was about to ask, so your comment comes in super handy ! I'll surely run the test. Thank you for your input ! |
|
Hi @fdesbiens, Sorry I’ve been a tiny bit busy for this week/week-end. I’ll run a test by this week for sure and confirm back to you if the regression tests are passing. Sorry for the delay ! Have a nice starting week |
|
@fdesbiens FYI, I'm currently running into an issue where the command to build threadx as a standalone library does not work anymore because THREADX_ARCH is not defined. I guess this is an aftermath of removing the project directive at the top of the CMakeLists.txt in a recent commit. |
|
My mistake, I will think about it next time ! :) |
This PR should fix #520. After the recent internal meeting, here is the solution we came up with. The riscv64 port now follows the same type size as other 64-bit architectures in tx_port.h. This also fixes some misalignment issues reported in #520.
This also makes other ThreadX components such as NetX working again (ULONG was assumed to be 4 bytes and not 8 bytes as the current official port).
PR checklist