linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
From: Michael Ellerman <mpe@ellerman.id.au>
To: Oleg Nesterov <oleg@redhat.com>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Madhavan Srinivasan <maddy@linux.vnet.ibm.com>,
	Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org,
	Jan Kratochvil <jan.kratochvil@redhat.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH?] powerpc: Hard wire PT_SOFTE value to 1 in gpr_get() too
Date: Thu, 19 Sep 2019 17:52:39 +1000	[thread overview]
Message-ID: <87ftksvi2g.fsf@mpe.ellerman.id.au> (raw)
In-Reply-To: <20190917121256.GA8659@redhat.com>

Hi Oleg,

Thanks for the patch.

Oleg Nesterov <oleg@redhat.com> writes:
> I don't have a ppc machine, this patch wasn't even compile tested,
> could you please review?
>
> The commit a8a4b03ab95f ("powerpc: Hard wire PT_SOFTE value to 1 in
> ptrace & signals") changed ptrace_get_reg(PT_SOFTE) to report 0x1,
> but PTRACE_GETREGS still copies pt_regs->softe as is.

Ugh, that certainly seems broken. I guess we forgot/didn't-know that
there were two paths through ptrace to get the one register.

> This is not consistent and this breaks
> http://sourceware.org/systemtap/wiki/utrace/tests/user-regs-peekpoke

That's a 404 for me?

Is it this: https://sourceware.org/systemtap/wiki/utrace/tests/

That seems to point me to a CVS repo? Which then didn't build. But now I
have that one test built, and you're right it fails with:

$ ./user-regs-peekpoke 
mismatch at offset 0x138: poked 0 but peeked 1


> Reported-by: Jan Kratochvil <jan.kratochvil@redhat.com>
> Signed-off-by: Oleg Nesterov <oleg@redhat.com>
> ---
>  arch/powerpc/kernel/ptrace.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
>
> diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
> index 8c92feb..9e9342c 100644
> --- a/arch/powerpc/kernel/ptrace.c
> +++ b/arch/powerpc/kernel/ptrace.c
> @@ -363,11 +363,36 @@ static int gpr_get(struct task_struct *target, const struct user_regset *regset,
>  	BUILD_BUG_ON(offsetof(struct pt_regs, orig_gpr3) !=
>  		     offsetof(struct pt_regs, msr) + sizeof(long));
>  
> +#ifdef CONFIG_PPC64
> +	if (!ret)
> +		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
> +					  &target->thread.regs->orig_gpr3,
> +					  offsetof(struct pt_regs, orig_gpr3),
> +					  offsetof(struct pt_regs, softe));
> +
> +	if (!ret) {
> +		unsigned long softe = 0x1;
> +		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, &msr,
> +					  offsetof(struct pt_regs, softe),
> +					  offsetof(struct pt_regs, softe) +
> +					  sizeof(softe));
> +	}
> +
> +	BUILD_BUG_ON(offsetof(struct pt_regs, trap) !=
> +		     offsetof(struct pt_regs, softe) + sizeof(long));
> +
> +	if (!ret)
> +		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
> +					  &target->thread.regs->trap,
> +					  offsetof(struct pt_regs, trap),
> +					  sizeof(struct user_pt_regs));
> +#else
>  	if (!ret)
>  		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
>  					  &target->thread.regs->orig_gpr3,
>  					  offsetof(struct pt_regs, orig_gpr3),
>  					  sizeof(struct user_pt_regs));
> +#endif
>  	if (!ret)
>  		ret = user_regset_copyout_zero(&pos, &count, &kbuf, &ubuf,
>  					       sizeof(struct user_pt_regs), -1);

It would be nice if we could isolate the special logic in once place,
ie. ptrace_get_reg().

We could do it like below. I'm 50/50 though on whether it's worth it, or
if we should just go with the big ifdef like in your patch.

cheers


diff --git a/arch/powerpc/kernel/ptrace.c b/arch/powerpc/kernel/ptrace.c
index 8c92febf5f44..55510f1a7ec1 100644
--- a/arch/powerpc/kernel/ptrace.c
+++ b/arch/powerpc/kernel/ptrace.c
@@ -334,6 +334,11 @@ int ptrace_put_reg(struct task_struct *task, int regno, unsigned long data)
 	return -EIO;
 }
 
+#ifndef __powerpc64__
+/* Needed on 32-bit to make the SOFTE logic below work without ifdefs */
+#define PT_SOFTE	PT_MQ
+#endif
+
 static int gpr_get(struct task_struct *target, const struct user_regset *regset,
 		   unsigned int pos, unsigned int count,
 		   void *kbuf, void __user *ubuf)
@@ -367,6 +372,24 @@ static int gpr_get(struct task_struct *target, const struct user_regset *regset,
 		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
 					  &target->thread.regs->orig_gpr3,
 					  offsetof(struct pt_regs, orig_gpr3),
+					  PT_SOFTE * sizeof(long));
+
+	/* SOFTE is special on 64-bit, the logic is in ptrace_get_reg() */
+	if (!ret) {
+		unsigned long val = 0;
+		ptrace_get_reg(target, PT_SOFTE, &val);
+		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, &val,
+					  PT_SOFTE * sizeof(long),
+					  offsetof(struct pt_regs, trap));
+	}
+
+	BUILD_BUG_ON(offsetof(struct pt_regs, trap) !=
+		     (PT_SOFTE * sizeof(long)) + sizeof(long));
+
+	if (!ret)
+		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
+					  &target->thread.regs->trap,
+					  offsetof(struct pt_regs, trap),
 					  sizeof(struct user_pt_regs));
 	if (!ret)
 		ret = user_regset_copyout_zero(&pos, &count, &kbuf, &ubuf,
@@ -3384,9 +3407,13 @@ void __init pt_regs_check(void)
 #ifdef __powerpc64__
 	BUILD_BUG_ON(offsetof(struct pt_regs, softe) !=
 		     offsetof(struct user_pt_regs, softe));
+	BUILD_BUG_ON(offsetof(struct pt_regs, softe) !=
+		     PT_SOFTE * sizeof(long));
 #else
 	BUILD_BUG_ON(offsetof(struct pt_regs, mq) !=
 		     offsetof(struct user_pt_regs, mq));
+	BUILD_BUG_ON(offsetof(struct pt_regs, mq) !=
+		     PT_MQ * sizeof(long));
 #endif
 	BUILD_BUG_ON(offsetof(struct pt_regs, trap) !=
 		     offsetof(struct user_pt_regs, trap));

      parent reply	other threads:[~2019-09-19  7:55 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-17 12:12 [PATCH?] powerpc: Hard wire PT_SOFTE value to 1 in gpr_get() too Oleg Nesterov
2019-09-17 14:09 ` kbuild test robot
2019-09-17 14:37 ` [PATCH? v2] " Oleg Nesterov
2020-06-10 15:07   ` Oleg Nesterov
2020-06-11  8:52     ` Madhavan Srinivasan
2020-06-11 10:58       ` Oleg Nesterov
2020-06-11 11:11         ` Jan Kratochvil
2019-09-19  7:52 ` Michael Ellerman [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87ftksvi2g.fsf@mpe.ellerman.id.au \
    --to=mpe@ellerman.id.au \
    --cc=benh@kernel.crashing.org \
    --cc=jan.kratochvil@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=maddy@linux.vnet.ibm.com \
    --cc=oleg@redhat.com \
    --cc=paulus@samba.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).