On Wed, Nov 15, 2023 at 01:01:24PM -0800, Randy Dunlap wrote: > > > On 11/14/23 17:27, Stephen Rothwell wrote: > > Hi all, > > > > Changes since 20231114: > > > > (using gcc 13.2.0 from kernel.org) > > on riscv 32-bit: > > arch/riscv/kernel/tests/module_test/test_uleb128.S: Assembler messages: > arch/riscv/kernel/tests/module_test/test_uleb128.S:9: Error: unrecognized opcode `ld a0,second' > arch/riscv/kernel/tests/module_test/test_uleb128.S:15: Error: unrecognized opcode `ld a0,fourth' > riscv32-linux-ld: cannot find arch/riscv/kernel/tests/module_test/test_uleb128.o: No such file or directory > > > Full randconfig file is attached. > > Is this a toolchain problem? > > thanks. > -- > ~Randy I had an assumption when writing this code that uleb128 was only supported on rv64 toolchains. Since it appears that there do exist rv32 toolchains with uleb128 support I will modify this test to be supported on rv32 platforms. The config AS_HAS_ULEB128 in arch/riscv/kernel/tests/Kconfig.debug should be auto-set to N if the toolchain does not support ULEB128 even though the randconfig has CONFIG_AS_HAS_ULEB128=y. There are a couple of sparse warnings in this series that I will fix up and send out in a new series with this fix. - Charlie
