lkml.org 
[lkml]   [2012]   [Jul]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] uprobes: don't enable/disable signle step if the user did it
Well. I agree, this needs changes. To begin with, uprobe should avoid
user_enable_single_step() which does access_process_vm(). And I suspect
uprobes have the problems with TIF_FORCED_TF logic.

But I am not sure about this patch...

On 07/26, Sebastian Andrzej Siewior wrote:
>
> @@ -1528,7 +1528,10 @@ static void handle_swbp(struct pt_regs *regs)
>
> utask->state = UTASK_SSTEP;
> if (!pre_ssout(uprobe, regs, bp_vaddr)) {
> - user_enable_single_step(current);
> + if (test_tsk_thread_flag(current, TIF_SINGLESTEP))
> + uprobe->flags |= UPROBE_USER_SSTEP;
> + else
> + user_enable_single_step(current);

This is x86 specific, TIF_SINGLESTEP is not defined on every arch.

> @@ -1569,7 +1572,10 @@ static void handle_singlestep(struct uprobe_task *utask, struct pt_regs *regs)
> put_uprobe(uprobe);
> utask->active_uprobe = NULL;
> utask->state = UTASK_RUNNING;
> - user_disable_single_step(current);
> + if (uprobe->flags & UPROBE_USER_SSTEP)
> + uprobe->flags &= ~UPROBE_USER_SSTEP;
> + else
> + user_disable_single_step(current);

This is not enough (and I am not sure this is portable).

If SINGLESTEP was set, we should send SIGTRAP here. With this patch
we return with X86_EFLAGS_TF set, gdb will be notified only after the
next insn. And if we notify gdb, there is no need to keep X86_EFLAGS_TF.

I'm afraid this needs more thinking and new arch-dependant helpers.

Oleg.



\
 
 \ /
  Last update: 2012-07-26 21:01    [W:0.195 / U:0.020 seconds]
©2003-2014 Jasper Spaans. Advertise on this site