hugemmap15: Support RISC-V to do __cache_flush. (#79)
PATCH 1/1 hugemmap15: Support RISC-V to do __cache_flush. (#79)
AndesCore CPU test fails due to an illegal instruction at the first jumpfunc(), caused by executing incorrect instructions after self-modifying code.
On RISC-V CPUs, a FENCE.i synchronizes instruction and data streams, ensuring that subsequent instruction fetch on a RISC-V hart will see any previous data stores already visible to the same RISC-V hart.
Thus, cacheflush() in the test should also use __clear_cache for RISC-V. In SMP systems, it invokes the kernel syscall riscv_flush_icache to inform each hart to flush their local i-caches.
--- testcases/kernel/mem/hugetlb/hugemmap/hugemmap15.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
testcases/kernel/mem/hugetlb/hugemmap/hugemmap15.c +1 -1
@@ -47,7 +47,7 @@ static void cacheflush(void *p) {#if defined(__powerpc__) asmvolatile ("dcbst 0,%0; sync; icbi 0,%0; isync" : :"r" (p)); -#elif defined(__arm__) || defined(__aarch64__) +#elif defined(__arm__) || defined(__aarch64__) || defined(__riscv)__clear_cache (p, p +COPY_SIZE );#else (void )p;
Signature · 5 lines
2.34.1 -- Mailing list info: https://lists.linux.it/listinfo/ltp
Hi Hui,
Quoted 11 lines
AndesCore CPU test fails due to an illegal instruction at the first
jumpfunc(), caused by executing incorrect instructions after self-modifying code.On RISC-V CPUs, a FENCE.i synchronizes instruction and data streams, ensuring that subsequent instruction fetch on a RISC-V hart will see any previous data stores already visible to the same RISC-V hart.
Thus,
cacheflush()in the test should also use__clear_cachefor RISC-V. In SMP systems, it invokes the kernel syscallriscv_flush_icacheto inform each hart to flush their local i-caches.
BTW what "#79" in the git message subject means?
Kind regards,
Petr
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
Re: [LTP] [PATCH] hugemmap15: Support RISC-V to do __cache_flush.
Quoted 16 lines · Petr Vorel
[EXTERNAL MAIL]
Hi Hui,
Quoted 11 lines
AndesCore CPU test fails due to an illegal instruction at the first
jumpfunc(), caused by executing incorrect instructions after self-modifying code.On RISC-V CPUs, a FENCE.i synchronizes instruction and data streams, ensuring that subsequent instruction fetch on a RISC-V hart will see any previous data stores already visible to the same RISC-V hart.
Thus,
cacheflush()in the test should also use__clear_cachefor RISC-V. In SMP systems, it invokes the kernel syscallriscv_flush_icacheto inform each hart to flush their local i-caches.
Quoted 6 lines
Reviewed-by Petr Vorel <[email protected]>BTW what "#79" in the git message subject means?
Kind regards,
Petr
Hi Petr,
Sorry, that was my mistake. I'll remove it and send another one.
Thanks,
Mina
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp