[PATCH] riscv: Fix TRACE_IRQFLAGS call points in entry.S

Guo Ren guoren at kernel.org
Sat Sep 10 17:45:58 PDT 2022


I think it would be fixed by GENERIC_ENTRY. You could have a try.

https://lore.kernel.org/linux-riscv/20220908022506.1275799-1-guoren@kernel.org/

On Fri, Sep 9, 2022 at 5:46 PM Yipeng Zou <zouyipeng at huawei.com> wrote:
>
> TRACE_IRQFLAGS need trace all {Enable,Disable} hard interrupts points
> in entry.S.
> We miss a path(work_notifysig) that cause an WARN as shown below.
>
> [   50.362431] ------------[ cut here ]------------
> [   50.363607] DEBUG_LOCKS_WARN_ON(!lockdep_hardirqs_enabled())
> [   50.367229] WARNING: CPU: 3 PID: 74 at kernel/locking/lockdep.c:5510 check_flags+0x10a/0x1e0
> [   50.368905] Modules linked in:
> [   50.369440] CPU: 3 PID: 74 Comm: rcS Tainted: G
> [   50.370225] Hardware name: riscv-virtio,qemu (DT)
> [   50.370632] epc : check_flags+0x10a/0x1e0
> [   50.371015]  ra : check_flags+0x10a/0x1e0
> [   50.371751] epc : ffffffff808e2670 ra : ffffffff808e2670 sp : ff20000000c3bca0
> [   50.372721]  gp : ffffffff817e2188 tp : ff60000002fe1800 t0 : ffffffff8162a2a8
> [   50.373321]  t1 : 0000000000000001 t2 : 2d2d2d2d2d2d2d2d s0 : ff20000000c3bcc0
> [   50.373903]  s1 : ffffffff817e3340 a0 : 0000000000000030 a1 : ffffffff81693fa8
> [   50.374476]  a2 : 0000000000000010 a3 : fffffffffffffffe a4 : 083151a422b8df00
> [   50.375039]  a5 : 083151a422b8df00 a6 : ffffffff8006891e a7 : 083151a422b8df00
> [   50.376028]  s2 : ffffffff817e54e0 s3 : ffffffff81695070 s4 : ff600000027d7800
> [   50.376978]  s5 : ffffffff80c22770 s6 : 0000000200000022 s7 : ffffffffffffffff
> [   50.377779]  s8 : ffffffff82301ed0 s9 : 0000000000001000 s10: 00000000001a9905
> [   50.378335]  s11: 000000000014f3e8 t3 : ffffffff822d714f t4 : ffffffff822d714f
> [   50.378955]  t5 : ffffffff822d7150 t6 : ff20000000c3ba98
> [   50.379783] status: 0000000200000100 badaddr: 0000000000000000 cause: 0000000000000003
> [   50.380810] [<ffffffff808e27be>] lock_is_held_type+0x78/0x14e
> [   50.381477] [<ffffffff8003e850>] __might_resched+0x26/0x22c
> [   50.382006] [<ffffffff8003ea92>] __might_sleep+0x3c/0x66
> [   50.382477] [<ffffffff8002324e>] get_signal+0xaa/0x912
> [   50.382947] [<ffffffff80005368>] do_notify_resume+0x82/0x434
> [   50.383839] [<ffffffff80003bec>] ret_from_exception+0x0/0x10
>
> Fixes: 3c4697982982 ("riscv: Enable LOCKDEP_SUPPORT & fixup TRACE_IRQFLAGS_SUPPORT")
> Signed-off-by: Yipeng Zou <zouyipeng at huawei.com>
> ---
>  arch/riscv/kernel/entry.S | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/arch/riscv/kernel/entry.S b/arch/riscv/kernel/entry.S
> index b9eda3fcbd6d..f4d23995a4f1 100644
> --- a/arch/riscv/kernel/entry.S
> +++ b/arch/riscv/kernel/entry.S
> @@ -370,16 +370,20 @@ resume_kernel:
>
>  work_pending:
>         /* Enter slow path for supplementary processing */
> -       la ra, ret_from_exception
>         andi s1, s0, _TIF_NEED_RESCHED
>         bnez s1, work_resched
>  work_notifysig:
> +#ifdef CONFIG_TRACE_IRQFLAGS
> +       call __trace_hardirqs_on
> +#endif
> +       la ra, ret_from_exception
>         /* Handle pending signals and notify-resume requests */
>         csrs CSR_STATUS, SR_IE /* Enable interrupts for do_notify_resume() */
>         move a0, sp /* pt_regs */
>         move a1, s0 /* current_thread_info->flags */
>         tail do_notify_resume
>  work_resched:
> +       la ra, ret_from_exception
>         tail schedule
>
>  /* Slow paths for ptrace. */
> --
> 2.17.1
>


-- 
Best Regards
 Guo Ren



More information about the linux-riscv mailing list