On Thu, Dec 20, 2018 at 04:40:30PM +0100, Greg Kroah-Hartman wrote:
> On Thu, Dec 20, 2018 at 12:14:00PM +0000, Sudip Mukherjee wrote:
> > Hi Greg,
> >
> > On Thu, Dec 20, 2018 at 9:28 AM Greg Kroah-Hartman
> > <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > 4.14-stable review patch. If anyone has any objections, please let me know.
> > >
> > > ------------------
> > >
> > > commit 7aa54be2976550f17c11a1c3e3630002dea39303 upstream.
> >
> > Another upstream commit fixes this.
> > b987ffc18fb3 ("x86/qspinlock: Fix compile error")
>
> Maybe, but that commit doesn't apply to any of these stable trees :(
>
> Care to provide a backport?
Attached now.
--
Regards
Sudip
>From 35c4feb16f5204faa40e9b0451cd86549819f375 Mon Sep 17 00:00:00 2001
From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Date: Fri, 2 Nov 2018 14:26:53 +0100
Subject: [PATCH] x86/qspinlock: Fix compile error
commit b987ffc18fb3b3b76b059aa9e372dbee26f7c4f2 upstream
With a compiler that has asm-goto but not asm-cc-output and
CONFIG_PROFILE_ALL_BRANCHES=y we get a compiler error:
arch/x86/include/asm/rmwcc.h:23:17: error: jump into statement expression
Fix this by writing the if() as a boolean multiplication instead.
Reported-by: kbuild test robot <lkp@xxxxxxxxx>
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Will Deacon <will.deacon@xxxxxxx>
Cc: linux-kernel@xxxxxxxxxxxxxxx
Fixes: 7aa54be29765 ("locking/qspinlock, x86: Provide liveness guarantee")
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
Signed-off-by: Sudip Mukherjee <sudipm.mukherjee@xxxxxxxxx>
---
arch/x86/include/asm/qspinlock.h | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/arch/x86/include/asm/qspinlock.h b/arch/x86/include/asm/qspinlock.h
index f784b95e44df..f0f2dd74b8c9 100644
--- a/arch/x86/include/asm/qspinlock.h
+++ b/arch/x86/include/asm/qspinlock.h
@@ -19,10 +19,14 @@ static __always_inline bool __queued_RMW_btsl(struct qspinlock *lock)
static __always_inline u32 queued_fetch_set_pending_acquire(struct qspinlock *lock)
{
- u32 val = 0;
+ u32 val;
- if (__queued_RMW_btsl(lock))
- val |= _Q_PENDING_VAL;
+ /*
+ * We can't use GEN_BINARY_RMWcc() inside an if() stmt because asm goto
+ * and CONFIG_PROFILE_ALL_BRANCHES=y results in a label inside a
+ * statement expression, which GCC doesn't like.
+ */
+ val = __queued_RMW_btsl(lock) * _Q_PENDING_VAL;
val |= atomic_read(&lock->val) & ~_Q_PENDING_MASK;
--
2.11.0
