For SEA exception, kernel require take some action to recover from memory error, such as isolate poison page adn kill failure thread, which are done in memory_failure().
During our test, the failure thread cannot be killed due to this issue[1], Here, I temporarily workaround this issue by sending signals to user processes in do_sea(). After [1] is merged, this patch can be rolled back or the SIGBUS will be sent repeated. [1]https://lore.kernel.org/lkml/20240204080144.7977-1-xuesh...@linux.alibaba.com/ Signed-off-by: Tong Tiangen <tongtian...@huawei.com> --- arch/arm64/mm/fault.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c index 2dc65f99d389..37d7e74d9aee 100644 --- a/arch/arm64/mm/fault.c +++ b/arch/arm64/mm/fault.c @@ -730,9 +730,6 @@ static int do_sea(unsigned long far, unsigned long esr, struct pt_regs *regs) const struct fault_info *inf; unsigned long siaddr; - if (do_apei_claim_sea(regs)) - return 0; - inf = esr_to_fault_info(esr); if (esr & ESR_ELx_FnV) { siaddr = 0; @@ -744,6 +741,17 @@ static int do_sea(unsigned long far, unsigned long esr, struct pt_regs *regs) */ siaddr = untagged_addr(far); } + + if (do_apei_claim_sea(regs)) { + if (current->mm) { + set_thread_esr(0, esr); + arm64_force_sig_fault(inf->sig, inf->code, siaddr, + "Uncorrected memory error on access \ + to poison memory\n"); + } + return 0; + } + arm64_notify_die(inf->name, regs, inf->sig, inf->code, siaddr, esr); return 0; -- 2.25.1
