[PATCH] RISC-V: Fix PTRACE_SETREGSET bug.

[PATCH] RISC-V: Fix PTRACE_SETREGSET bug.

[Jim Wilson]

In riscv_gpr_set, pass regs instead of &regs to user_regset_copyin to fix
gdb segfault.

Signed-off-by: Jim Wilson <jimw@sifive.com>
---
 arch/riscv/kernel/ptrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c
index ba3e80712797..9f82a7e34c64 100644
--- a/arch/riscv/kernel/ptrace.c
+++ b/arch/riscv/kernel/ptrace.c
@@ -50,7 +50,7 @@ static int riscv_gpr_set(struct task_struct *target,
 	struct pt_regs *regs;
 
 	regs = task_pt_regs(target);
-	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &regs, 0, -1);
+	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, regs, 0, -1);
 	return ret;
 }
 
-- 
2.17.1

Re: [PATCH] RISC-V: Fix PTRACE_SETREGSET bug.

[Palmer Dabbelt]

On Mon, 11 Jun 2018 14:48:22 PDT (-0700), Jim Wilson wrote:
> In riscv_gpr_set, pass regs instead of &regs to user_regset_copyin to fix
> gdb segfault.
>
> Signed-off-by: Jim Wilson <jimw@sifive.com>
> ---
>  arch/riscv/kernel/ptrace.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/riscv/kernel/ptrace.c b/arch/riscv/kernel/ptrace.c
> index ba3e80712797..9f82a7e34c64 100644
> --- a/arch/riscv/kernel/ptrace.c
> +++ b/arch/riscv/kernel/ptrace.c
> @@ -50,7 +50,7 @@ static int riscv_gpr_set(struct task_struct *target,
>  	struct pt_regs *regs;
>
>  	regs = task_pt_regs(target);
> -	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &regs, 0, -1);
> +	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, regs, 0, -1);
>  	return ret;
>  }

Whoops, that's embarassing :).  I poked around and didn't see this anywhere 
else, so I'm not sure where we managed to obtain this particular pathology.

Thanks for the patch!

Re: [PATCH] RISC-V: Fix PTRACE_SETREGSET bug.

[Jim Wilson]

On Mon, Jun 11, 2018 at 5:46 PM, Palmer Dabbelt <palmer@sifive.com> wrote:
> Whoops, that's embarassing :).  I poked around and didn't see this anywhere
> else, so I'm not sure where we managed to obtain this particular pathology.

Comment say it was copied from tile which I think was just recently
removed.  But looking at older copy, tile has "struct pt_regs regs"
whereas riscv has "struct pt_regs *regs" and so tile needed the & and
riscv does not, but only one of the two ampersands was removed in the
riscv port.  My patch removes the other one.

Jim