This is not a regression and has been this way since the dawn of 5.8 riscv, as we have CONFIG_STRICT_KERNEL_RWX set.
When investigating, I see many fixes that involve the linker script, so I assume this is due to some sections, like init sections carrying both text and data, which could cause this.
I tried picking a lot of commits, and needed to either do extensive backports or carry a lot of commits, and still without success, hitting a linker problem that would require carrying even more fixes.
So, contrary to what has been said about 5.8 bringing up STRICT_KERNEL_RWX to riscv, this is really not ready for prime time, and should be left for future versions.
TL;DR: We should just disable CONFIG_STRICT_KERNEL_RWX on linux-riscv 5.8.
This is not a regression and has been this way since the dawn of 5.8 riscv, as we have CONFIG_ STRICT_ KERNEL_ RWX set.
When investigating, I see many fixes that involve the linker script, so I assume this is due to some sections, like init sections carrying both text and data, which could cause this.
I tried picking a lot of commits, and needed to either do extensive backports or carry a lot of commits, and still without success, hitting a linker problem that would require carrying even more fixes.
So, contrary to what has been said about 5.8 bringing up STRICT_KERNEL_RWX to riscv, this is really not ready for prime time, and should be left for future versions.
TL;DR: We should just disable CONFIG_ STRICT_ KERNEL_ RWX on linux-riscv 5.8.
Cascardo.