Hi Jisheng, 在 2021/3/31 22:22, Jisheng Zhang 写道: > On Tue, 30 Mar 2021 18:33:16 +0900 > Masami Hiramatsu <[email protected]> wrote: > >> Hi Jisheng, > > Hi Masami, > >> >> On Tue, 30 Mar 2021 02:16:24 +0800 >> Jisheng Zhang <[email protected]> wrote: >> >>> From: Jisheng Zhang <[email protected]> >>> >>> Current riscv's kprobe handlers are run with both preemption and >>> interrupt enabled, this violates kprobe requirements. Fix this issue >>> by keeping interrupts disabled for BREAKPOINT exception. >> >> Not only while the breakpoint exception but also until the end of >> the single step (maybe you are using __BUG_INSN_32 ??) need to be >> disable interrupts. Can this do that? >> > > interrupt is disabled during "single step" by kprobes_save_local_irqflag() > and kprobes_restore_local_irqflag(). The code flow looks like: > > do_trap_break() // for bp > kprobe_breakpoint_handler() > setup_singlestep() > kprobes_restore_local_irqflag() > > do_trap_break() // for ss > kprobe_single_step_handler() > kprobes_restore_local_irqflag()
Recently, kernel hit BUG_ON() on QEMU after I install a probe at "sys_read" via kprobe, accoriding to my debugging and analysis it looks like caused by the "irq disable" operation for single-stepping. I present a detailed description about this problem in the mail with title "[PATCH] riscv/kprobe: fix kernel panic when invoking sys_read traced by kprobe". Looking forward to some feedback,Thanks. BR, Liao Chang > > Thanks > > > _______________________________________________ > linux-riscv mailing list > [email protected] > http://lists.infradead.org/mailman/listinfo/linux-riscv > . >
