IRC channel logs

2025-02-11.log

back to list of logs

<stikonas>gtker: hmm, so I've ran live-bootstrap on x86 and amd64, there are no new issues
<stikonas>on riscv64 I think there is an issue...
<stikonas> +> ./configurator /steps/configurator
<stikonas>./configurator abnormal termination, signal number = 11
<stikonas>gtker, ok, ungz also crashes
<stikonas>we'll have to double check assembly instructions in multiply logic
<stikonas>hmm, the crash might be somewhere else, not in pointer arithmtic
<oriansj>well, I have cleared out a dozen segfaults in M2-Planet but I somehow managed to find a segfault in M2-Planet that crashed the host system 0_0
<stikonas>well, my ungz crash was in the binary compiled by M2-Planet...
<oriansj>risc-v instructions appear to be missing: rd_sp rs1_sp !-12 addi
<oriansj>as there is a push onto the stack without a matching allocation
<oriansj>x86 and amd64 (and armv7l and aarch64) all update the stack pointer with every push and pop
<oriansj>and my segfault fixes have been merged
<stikonas>oh, perhaps we indeed can't use stack withotu allocating it
<oriansj>good news is RISC-V can avoid that entirely for the pointer arithmetic by just dedicating a register for the immediate value