Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1610371 > unrolled thread
| Started by | Ingo Molnar <mingo@kernel.org> |
|---|---|
| First post | 2017-03-28 09:10 +0200 |
| Last post | 2017-03-28 17:30 +0200 |
| Articles | 2 — 2 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
Re: [RFC PATCH tip/master V2 4/8] kprobes/x86: Do not modify singlestep buffer while resuming Ingo Molnar <mingo@kernel.org> - 2017-03-28 09:10 +0200
Re: [RFC PATCH tip/master V2 4/8] kprobes/x86: Do not modify singlestep buffer while resuming Masami Hiramatsu <mhiramat@kernel.org> - 2017-03-28 17:30 +0200
| From | Ingo Molnar <mingo@kernel.org> |
|---|---|
| Date | 2017-03-28 09:10 +0200 |
| Subject | Re: [RFC PATCH tip/master V2 4/8] kprobes/x86: Do not modify singlestep buffer while resuming |
| Message-ID | <tpTqq-41Y-7@gated-at.bofh.it> |
* Masami Hiramatsu <mhiramat@kernel.org> wrote:
> Do not modify singlestep execution buffer (kprobe.ainsn.insn)
> while resuming from single-stepping, instead, modifies
> the buffer to add a jump back instruction at preparing
> buffer.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
> ---
> arch/x86/kernel/kprobes/core.c | 42 +++++++++++++++++++---------------------
> 1 file changed, 20 insertions(+), 22 deletions(-)
>
> diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c
> index 6327f95..ea3b8e5 100644
> --- a/arch/x86/kernel/kprobes/core.c
> +++ b/arch/x86/kernel/kprobes/core.c
> @@ -399,23 +399,36 @@ int __copy_instruction(u8 *dest, u8 *src)
> return length;
> }
>
> +/* Prepare reljump right after instruction to boost */
> +static void prepare_boost(struct kprobe *p, int length)
> +{
> + if (can_boost(p->ainsn.insn, p->addr) &&
> + MAX_INSN_SIZE - length >= RELATIVEJUMP_SIZE) {
> + /*
> + * These instructions can be executed directly if it
> + * jumps back to correct address.
> + */
> + synthesize_reljump((void *)p->ainsn.insn + length,
> + (void *)p->addr + length);
> + p->ainsn.boostable = 1;
> + } else
> + p->ainsn.boostable = -1;
Those imbalanced curly braces are not proper kernel style.
Also, is the (void *) cast required? arch.insns ought to be void * already, right?
(I haven't checking whether that's true for all architectures.)
Btw., the original code had the curly braces right:
> - if (p->ainsn.boostable == 0) {
> - if ((regs->ip > copy_ip) &&
> - (regs->ip - copy_ip) + 5 < MAX_INSN_SIZE) {
> - /*
> - * These instructions can be executed directly if it
> - * jumps back to correct address.
> - */
> - synthesize_reljump((void *)regs->ip,
> - (void *)orig_ip + (regs->ip - copy_ip));
> - p->ainsn.boostable = 1;
> - } else {
> - p->ainsn.boostable = -1;
> - }
> - }
Thanks,
Ingo
[toc] | [next] | [standalone]
| From | Masami Hiramatsu <mhiramat@kernel.org> |
|---|---|
| Date | 2017-03-28 17:30 +0200 |
| Message-ID | <tq1ei-1a1-3@gated-at.bofh.it> |
| In reply to | #1610371 |
On Tue, 28 Mar 2017 09:04:19 +0200
Ingo Molnar <mingo@kernel.org> wrote:
>
> * Masami Hiramatsu <mhiramat@kernel.org> wrote:
>
> > Do not modify singlestep execution buffer (kprobe.ainsn.insn)
> > while resuming from single-stepping, instead, modifies
> > the buffer to add a jump back instruction at preparing
> > buffer.
> >
> > Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
> > ---
> > arch/x86/kernel/kprobes/core.c | 42 +++++++++++++++++++---------------------
> > 1 file changed, 20 insertions(+), 22 deletions(-)
> >
> > diff --git a/arch/x86/kernel/kprobes/core.c b/arch/x86/kernel/kprobes/core.c
> > index 6327f95..ea3b8e5 100644
> > --- a/arch/x86/kernel/kprobes/core.c
> > +++ b/arch/x86/kernel/kprobes/core.c
> > @@ -399,23 +399,36 @@ int __copy_instruction(u8 *dest, u8 *src)
> > return length;
> > }
> >
> > +/* Prepare reljump right after instruction to boost */
> > +static void prepare_boost(struct kprobe *p, int length)
> > +{
> > + if (can_boost(p->ainsn.insn, p->addr) &&
> > + MAX_INSN_SIZE - length >= RELATIVEJUMP_SIZE) {
> > + /*
> > + * These instructions can be executed directly if it
> > + * jumps back to correct address.
> > + */
> > + synthesize_reljump((void *)p->ainsn.insn + length,
> > + (void *)p->addr + length);
> > + p->ainsn.boostable = 1;
> > + } else
> > + p->ainsn.boostable = -1;
>
> Those imbalanced curly braces are not proper kernel style.
Ah, OK, I'll fix that.
>
> Also, is the (void *) cast required? arch.insns ought to be void * already, right?
Hmm, it is kprobe_opcode_t *, but anyway, on x86 it is same,, since kprobe_opcode_t
on x86 is u8.
> (I haven't checking whether that's true for all architectures.)
It is OK because this is x86 very specific code.
Thank you,
>
> Btw., the original code had the curly braces right:
>
> > - if (p->ainsn.boostable == 0) {
> > - if ((regs->ip > copy_ip) &&
> > - (regs->ip - copy_ip) + 5 < MAX_INSN_SIZE) {
> > - /*
> > - * These instructions can be executed directly if it
> > - * jumps back to correct address.
> > - */
> > - synthesize_reljump((void *)regs->ip,
> > - (void *)orig_ip + (regs->ip - copy_ip));
> > - p->ainsn.boostable = 1;
> > - } else {
> > - p->ainsn.boostable = -1;
> > - }
> > - }
>
> Thanks,
>
> Ingo
--
Masami Hiramatsu <mhiramat@kernel.org>
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web