summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: 95cf00f)
raw | patch | inline | side by side (parent: 95cf00f)
author | Oleg Nesterov <oleg@redhat.com> | |
Mon, 3 Sep 2012 13:24:17 +0000 (15:24 +0200) | ||
committer | Oleg Nesterov <oleg@redhat.com> | |
Sat, 15 Sep 2012 15:37:30 +0000 (17:37 +0200) |
user_enable/disable_single_step() was designed for ptrace, it assumes
a single user and does unnecessary and wrong things for uprobes. For
example:
- arch_uprobe_enable_step() can't trust TIF_SINGLESTEP, an
application itself can set X86_EFLAGS_TF which must be
preserved after arch_uprobe_disable_step().
- we do not want to set TIF_SINGLESTEP/TIF_FORCED_TF in
arch_uprobe_enable_step(), this only makes sense for ptrace.
- otoh we leak TIF_SINGLESTEP if arch_uprobe_disable_step()
doesn't do user_disable_single_step(), the application will
be killed after the next syscall.
- arch_uprobe_enable_step() does access_process_vm() we do
not need/want.
Change arch_uprobe_enable/disable_step() to set/clear X86_EFLAGS_TF
directly, this is much simpler and more correct. However, we need to
clear TIF_BLOCKSTEP/DEBUGCTLMSR_BTF before executing the probed insn,
add set_task_blockstep(false).
Note: with or without this patch, there is another (hopefully minor)
problem. A probed "pushf" insn can see the wrong X86_EFLAGS_TF set by
uprobes. Perhaps we should change _disable to update the stack, or
teach arch_uprobe_skip_sstep() to emulate this insn.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
a single user and does unnecessary and wrong things for uprobes. For
example:
- arch_uprobe_enable_step() can't trust TIF_SINGLESTEP, an
application itself can set X86_EFLAGS_TF which must be
preserved after arch_uprobe_disable_step().
- we do not want to set TIF_SINGLESTEP/TIF_FORCED_TF in
arch_uprobe_enable_step(), this only makes sense for ptrace.
- otoh we leak TIF_SINGLESTEP if arch_uprobe_disable_step()
doesn't do user_disable_single_step(), the application will
be killed after the next syscall.
- arch_uprobe_enable_step() does access_process_vm() we do
not need/want.
Change arch_uprobe_enable/disable_step() to set/clear X86_EFLAGS_TF
directly, this is much simpler and more correct. However, we need to
clear TIF_BLOCKSTEP/DEBUGCTLMSR_BTF before executing the probed insn,
add set_task_blockstep(false).
Note: with or without this patch, there is another (hopefully minor)
problem. A probed "pushf" insn can see the wrong X86_EFLAGS_TF set by
uprobes. Perhaps we should change _disable to update the stack, or
teach arch_uprobe_skip_sstep() to emulate this insn.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
arch/x86/include/asm/processor.h | patch | blob | history | |
arch/x86/kernel/step.c | patch | blob | history | |
arch/x86/kernel/uprobes.c | patch | blob | history |
index d048cad9bcad4b8c48c62279d7b73eb0499305d7..433d2e5c98a7976951b134c90e4bf4984c8bfe18 100644 (file)
wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctlmsr);
}
+extern void set_task_blockstep(struct task_struct *task, bool on);
+
/*
* from system description table in BIOS. Mostly for MCA use, but
* others may find it useful:
diff --git a/arch/x86/kernel/step.c b/arch/x86/kernel/step.c
index f89cdc6ccd5b167d5645557cdd3a71d505701dd5..cd3b2438a9800b8568b0d79788ef3fdb438aa1ca 100644 (file)
--- a/arch/x86/kernel/step.c
+++ b/arch/x86/kernel/step.c
return 1;
}
-static void set_task_blockstep(struct task_struct *task, bool on)
+void set_task_blockstep(struct task_struct *task, bool on)
{
unsigned long debugctl;
index 309a0e02b124b9b10e3221e890c5de53c5330b90..3b4aae68efe052f5804f1ede1a831ce759e6e01c 100644 (file)
@@ -683,26 +683,30 @@ bool arch_uprobe_skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs)
void arch_uprobe_enable_step(struct arch_uprobe *auprobe)
{
- struct uprobe_task *utask = current->utask;
- struct arch_uprobe_task *autask = &utask->autask;
+ struct task_struct *task = current;
+ struct arch_uprobe_task *autask = &task->utask->autask;
+ struct pt_regs *regs = task_pt_regs(task);
autask->restore_flags = 0;
- if (!test_tsk_thread_flag(current, TIF_SINGLESTEP) &&
- !(auprobe->fixups & UPROBE_FIX_SETF))
+ if (!(regs->flags & X86_EFLAGS_TF) &&
+ !(auprobe->fixups & UPROBE_FIX_SETF))
autask->restore_flags |= UPROBE_CLEAR_TF;
- /*
- * The state of TIF_BLOCKSTEP is not saved. With the TF flag set we
- * would to examine the opcode and the flags to make it right. Without
- * TF block stepping makes no sense.
- */
- user_enable_single_step(current);
+
+ regs->flags |= X86_EFLAGS_TF;
+ if (test_tsk_thread_flag(task, TIF_BLOCKSTEP))
+ set_task_blockstep(task, false);
}
void arch_uprobe_disable_step(struct arch_uprobe *auprobe)
{
- struct uprobe_task *utask = current->utask;
- struct arch_uprobe_task *autask = &utask->autask;
-
+ struct task_struct *task = current;
+ struct arch_uprobe_task *autask = &task->utask->autask;
+ struct pt_regs *regs = task_pt_regs(task);
+ /*
+ * The state of TIF_BLOCKSTEP was not saved so we can get an extra
+ * SIGTRAP if we do not clear TF. We need to examine the opcode to
+ * make it right.
+ */
if (autask->restore_flags & UPROBE_CLEAR_TF)
- user_disable_single_step(current);
+ regs->flags &= ~X86_EFLAGS_TF;
}