2008-03-03 21:09:29

by Suresh Siddha

Subject: [patch] x86, i387: fix ptrace leakage using init_fpu()

This needs to go into 2.6.25. Thanks.
---

This bug is introduced by the recent i387 merge.

Current usage of unlazy_fpu() in ptrace specific routines is wrong. unlazy_fpu()
will not init fpu if the task never used math. So the ptrace calls
can expose the parent tasks FPU data in some cases.

Replace it with the init_fpu() which will init the math state, if the task
never used math before.

Signed-off-by: Suresh Siddha <[email protected]>
---

diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index 763dfc4..60fe801 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -132,7 +132,7 @@ int xfpregs_get(struct task_struct *target, const struct user_regset *regset,
if (!cpu_has_fxsr)
return -ENODEV;

- unlazy_fpu(target);
+ init_fpu(target);

return user_regset_copyout(&pos, &count, &kbuf, &ubuf,
&target->thread.i387.fxsave, 0, -1);
@@ -147,7 +147,7 @@ int xfpregs_set(struct task_struct *target, const struct user_regset *regset,
if (!cpu_has_fxsr)
return -ENODEV;

- unlazy_fpu(target);
+ init_fpu(target);
set_stopped_child_used_math(target);

ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf,
@@ -307,7 +307,7 @@ int fpregs_get(struct task_struct *target, const struct user_regset *regset,
if (!HAVE_HWFP)
return fpregs_soft_get(target, regset, pos, count, kbuf, ubuf);

- unlazy_fpu(target);
+ init_fpu(target);

if (!cpu_has_fxsr)
return user_regset_copyout(&pos, &count, &kbuf, &ubuf,
@@ -332,7 +332,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset,
if (!HAVE_HWFP)
return fpregs_soft_set(target, regset, pos, count, kbuf, ubuf);

- unlazy_fpu(target);
+ init_fpu(target);
set_stopped_child_used_math(target);

if (!cpu_has_fxsr)

2008-03-03 21:12:49

by Roland McGrath

Subject: Re: [patch] x86, i387: fix ptrace leakage using init_fpu()

Looks like a good fix to me.

Thanks,
Roland

2008-03-04 08:13:46

by Ingo Molnar

Subject: Re: [patch] x86, i387: fix ptrace leakage using init_fpu()

* Suresh Siddha <[email protected]> wrote:

> This bug is introduced by the recent i387 merge.
>
> Current usage of unlazy_fpu() in ptrace specific routines is wrong.
> unlazy_fpu() will not init fpu if the task never used math. So the
> ptrace calls can expose the parent tasks FPU data in some cases.
>
> Replace it with the init_fpu() which will init the math state, if the
> task never used math before.

thanks Suresh, good catch! Applied to x86.git.

Ingo