Re: [PATCH 1/5] i387: math_state_restore() isn't called from asm

Re: [PATCH 1/5] i387: math_state_restore() isn't called from asm

[George Spelvin]

> + * Must be called with kernel preemption disabled (eg with local
> + * local interrupts as in the case of do_device_not_available).

"eg with local local interrupts"?

As long as I'm quibbling, "exempli gratia" deserves periods to make the
abbreviation, and a parenthetical comment at the end of a sentence is
usually better set off with a comma.

+ * Must be called with kernel preemption disabled, e.g. with local
+ * interrupts disabled as in the case of do_device_not_available.

[PATCH 0/5] i387: stable kernel backport

[Linus Torvalds]

This is a reasonably minimal backport of the i387 state save/restore bug.

A few of the commits are just minimal "make it easier to backport" commits 
that don't necessarily fix anything on their own.  And a few of the others 
are combinations of what is two commits in the development tree, because 
it turned out to be easier and cleaner that way.

The last one is the one that fixes the x86-32 bug, but the preemption 
fixes are real fixes too, although they are probably not something that 
anybody has necessarily ever hit in reality because the race window is so 
small.  Even so, if the fix itself isn't that important, the "make it 
easier to backport the main one" would still be a big argument for it.

I *really* hope that the people who could reproduce this bug will test the 
back-port series too, since I never actually saw the bug personally to 
begin with. And again, big thanks to Raphael who helped pinpoint and debug 
this.

                   Linus

---
Linus Torvalds (5):
  i387: math_state_restore() isn't called from asm
  i387: make irq_fpu_usable() tests more robust
  i387: fix x86-64 preemption-unsafe user stack save/restore
  i387: move TS_USEDFPU clearing out of __save_init_fpu and into callers
  i387: move TS_USEDFPU flag from thread_info to task_struct

 arch/x86/include/asm/i387.h        |  158 +++++++++++++++++++++++++++++++-----
 arch/x86/include/asm/processor.h   |    1 +
 arch/x86/include/asm/thread_info.h |    2 -
 arch/x86/kernel/traps.c            |   14 ++--
 arch/x86/kernel/xsave.c            |   12 +--
 arch/x86/kvm/vmx.c                 |    2 +-
 6 files changed, 150 insertions(+), 39 deletions(-)

-- 
1.7.9.188.g12766.dirty

[PATCH 1/5] i387: math_state_restore() isn't called from asm

[Linus Torvalds]

From: Linus Torvalds <torvalds@linux-foundation.org>
Date: Mon, 13 Feb 2012 13:47:25 -0800

[ Upstream commit be98c2cdb15ba26148cd2bd58a857d4f7759ed38 ]

It was marked asmlinkage for some really old and stale legacy reasons.
Fix that and the equally stale comment.

Noticed when debugging the irq_fpu_usable() bugs.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: stable@kernel.org	# Helps back-porting
---

Ok, this is trivial, but correct, and avoids the silly clashes with the 
other patches.

 arch/x86/include/asm/i387.h |    2 +-
 arch/x86/kernel/traps.c     |    6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/x86/include/asm/i387.h b/arch/x86/include/asm/i387.h
index c9e09ea05644..cba143210780 100644
--- a/arch/x86/include/asm/i387.h
+++ b/arch/x86/include/asm/i387.h
@@ -29,7 +29,7 @@ extern unsigned int sig_xstate_size;
 extern void fpu_init(void);
 extern void mxcsr_feature_mask_init(void);
 extern int init_fpu(struct task_struct *child);
-extern asmlinkage void math_state_restore(void);
+extern void math_state_restore(void);
 extern void __math_state_restore(void);
 extern int dump_fpu(struct pt_regs *, struct user_i387_struct *);
 
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index a8e3eb83466c..727e6c16f294 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -590,10 +590,10 @@ void __math_state_restore(void)
  * Careful.. There are problems with IBM-designed IRQ13 behaviour.
  * Don't touch unless you *really* know how it works.
  *
- * Must be called with kernel preemption disabled (in this case,
- * local interrupts are disabled at the call-site in entry.S).
+ * Must be called with kernel preemption disabled (eg with local
+ * local interrupts as in the case of do_device_not_available).
  */
-asmlinkage void math_state_restore(void)
+void math_state_restore(void)
 {
 	struct thread_info *thread = current_thread_info();
 	struct task_struct *tsk = thread->task;
-- 
1.7.9.188.g12766.dirty