From: Oleg Nesterov <oleg@redhat.com>
To: Al Viro <viro@ZenIV.linux.org.uk>,
Andrew Morton <akpm@linux-foundation.org>,
Fenghua Yu <fenghua.yu@intel.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Bean Anderson <bean@azulsystems.com>,
"H. Peter Anvin" <hpa@zytor.com>, Ingo Molnar <mingo@redhat.com>,
Thomas Gleixner <tglx@linutronix.de>,
x86@kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH 2/5] x86, fpu: don't drop_fpu() in __restore_xstate_sig() if use_eager_fpu()
Date: Sun, 24 Aug 2014 21:47:36 +0200 [thread overview]
Message-ID: <20140824194736.GA27441@redhat.com> (raw)
In-Reply-To: <20140824194700.GA27281@redhat.com>
__restore_xstate_sig() calls math_state_restore() with preemption
enabled, not good. But this is minor, the main problem is that this
drop_fpu/set_used_math/math_state_restore sequence creates the nasty
"use_eager_fpu() && !used_math()" special case which complicates other
FPU paths.
Change __restore_xstate_sig() to switch to swapper's fpu state, copy
the user state to the thread's fpu state, and switch fpu->state back
after sanitize_restored_xstate().
Without use_eager_fpu() fpu->state is null in between but this is fine
because in this case we rely on clear_used_math()/set_used_math(), so
this doesn't differ from !fpu_allocated() case.
Note: with or without this patch, perhaps it makes sense to send SEGV
if __copy_from_user() fails.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
arch/x86/kernel/xsave.c | 36 ++++++++++++++++++++++--------------
1 files changed, 22 insertions(+), 14 deletions(-)
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index 74d4129..51be404 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -325,6 +325,22 @@ static inline int restore_user_xstate(void __user *buf, u64 xbv, int fx_only)
return frstor_user(buf);
}
+static void switch_fpu_xstate(struct task_struct *tsk, union thread_xstate *xstate)
+{
+ preempt_disable();
+ __drop_fpu(tsk);
+ tsk->thread.fpu_counter = 0;
+ tsk->thread.fpu.state = xstate;
+ /* use_eager_fpu() => xstate != NULL */
+ if (use_eager_fpu())
+ math_state_restore();
+ else if (xstate)
+ set_used_math();
+ else
+ clear_used_math();
+ preempt_enable();
+}
+
int __restore_xstate_sig(void __user *buf, void __user *buf_fx, int size)
{
int ia32_fxstate = (buf != buf_fx);
@@ -377,28 +393,20 @@ int __restore_xstate_sig(void __user *buf, void __user *buf_fx, int size)
union thread_xstate *xstate = tsk->thread.fpu.state;
struct user_i387_ia32_struct env;
int err = 0;
-
/*
- * Drop the current fpu which clears used_math(). This ensures
- * that any context-switch during the copy of the new state,
- * avoids the intermediate state from getting restored/saved.
- * Thus avoiding the new restored state from getting corrupted.
- * We will be ready to restore/save the state only after
- * set_used_math() is again set.
+ * Ensure that that any context-switch during the copy of
+ * the new state, avoids the intermediate state from getting
+ * restored/saved.
*/
- drop_fpu(tsk);
-
+ switch_fpu_xstate(tsk, init_task.thread.fpu.state);
if (__copy_from_user(&xstate->xsave, buf_fx, state_size) ||
__copy_from_user(&env, buf, sizeof(env))) {
+ fpu_finit(&tsk->thread.fpu);
err = -1;
} else {
sanitize_restored_xstate(xstate, &env, xstate_bv, fx_only);
- set_used_math();
}
-
- if (use_eager_fpu())
- math_state_restore();
-
+ switch_fpu_xstate(tsk, xstate);
return err;
} else {
/*
--
1.5.5.1
next prev parent reply other threads:[~2014-08-24 19:50 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-08-22 17:11 [PATCH 0/1] x86, fpu: shift drop_init_fpu() from save_xstate_sig() to handle_signal() Oleg Nesterov
2014-08-22 17:12 ` [PATCH 1/1] " Oleg Nesterov
2014-08-24 19:47 ` [PATCH 0/5] x86, fpu: make use_eager_fpu() more eager Oleg Nesterov
2014-08-24 19:47 ` [PATCH 1/5] x86, fpu: change sanitize_restored_xstate() and convert_to_fxsr() to accept thread_xstate Oleg Nesterov
2014-08-24 19:47 ` Oleg Nesterov [this message]
2014-08-24 20:05 ` [PATCH 2/5] x86, fpu: don't drop_fpu() in __restore_xstate_sig() if use_eager_fpu() Linus Torvalds
2014-08-25 14:26 ` Oleg Nesterov
2014-08-25 14:41 ` Oleg Nesterov
2014-08-25 16:27 ` Linus Torvalds
2014-08-25 17:09 ` Oleg Nesterov
2014-08-25 17:26 ` Linus Torvalds
2014-08-25 17:39 ` Oleg Nesterov
2014-08-27 17:02 ` Oleg Nesterov
2014-08-24 19:47 ` [PATCH 3/5] x86, fpu: don't drop_fpu() in exit_thread() " Oleg Nesterov
2014-08-24 19:47 ` [PATCH 4/5] x86, fpu: shift init_fpu() from eager_fpu_init() to eager_fpu_init_bp() Oleg Nesterov
2014-08-24 19:47 ` [PATCH 5/5] x86, fpu: sanitize the usage of use_eager_fpu() in switch_fpu_prepare() Oleg Nesterov
2014-08-25 18:08 ` [PATCH] x86, fpu: __restore_xstate_sig()->math_state_restore() needs preempt_disable() Oleg Nesterov
2014-09-02 5:01 ` Suresh Siddha
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=20140824194736.GA27441@redhat.com \
--to=oleg@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=bean@azulsystems.com \
--cc=fenghua.yu@intel.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=suresh.b.siddha@intel.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=viro@ZenIV.linux.org.uk \
--cc=x86@kernel.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