| Message ID | 20201027225454.3492351-5-bigeasy@linutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | in_interrupt() cleanup, part 2 | expand |
On Tue, 27 Oct 2020 23:54:43 +0100 Sebastian Andrzej Siewior wrote: > mlx5_eq_async_int() uses in_irq() to decide whether eq::lock needs to be > acquired and released with spin_[un]lock() or the irq saving/restoring > variants. > > The usage of in_*() in drivers is phased out and Linus clearly requested > that code which changes behaviour depending on context should either be > seperated or the context be conveyed in an argument passed by the caller, > which usually knows the context. > > mlx5_eq_async_int() knows the context via the action argument already so > using it for the lock variant decision is a straight forward replacement > for in_irq(). > > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> > Cc: Saeed Mahameed <saeedm@nvidia.com> > Cc: Leon Romanovsky <leon@kernel.org> > Cc: "David S. Miller" <davem@davemloft.net> > Cc: Jakub Kicinski <kuba@kernel.org> > Cc: linux-rdma@vger.kernel.org Saeed, please pick this up into your tree.
On Sat, 2020-10-31 at 09:59 -0700, Jakub Kicinski wrote: > On Tue, 27 Oct 2020 23:54:43 +0100 Sebastian Andrzej Siewior wrote: > > mlx5_eq_async_int() uses in_irq() to decide whether eq::lock needs > > to be > > acquired and released with spin_[un]lock() or the irq > > saving/restoring > > variants. > > > > The usage of in_*() in drivers is phased out and Linus clearly > > requested > > that code which changes behaviour depending on context should > > either be > > seperated or the context be conveyed in an argument passed by the > > caller, > > which usually knows the context. > > > > mlx5_eq_async_int() knows the context via the action argument > > already so > > using it for the lock variant decision is a straight forward > > replacement > > for in_irq(). > > > > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> > > Cc: Saeed Mahameed <saeedm@nvidia.com> > > Cc: Leon Romanovsky <leon@kernel.org> > > Cc: "David S. Miller" <davem@davemloft.net> > > Cc: Jakub Kicinski <kuba@kernel.org> > > Cc: linux-rdma@vger.kernel.org > > Saeed, please pick this up into your tree. Ack
On Sat, 2020-10-31 at 09:59 -0700, Jakub Kicinski wrote: > On Tue, 27 Oct 2020 23:54:43 +0100 Sebastian Andrzej Siewior wrote: > > mlx5_eq_async_int() uses in_irq() to decide whether eq::lock needs > > to be > > acquired and released with spin_[un]lock() or the irq > > saving/restoring > > variants. > > > > The usage of in_*() in drivers is phased out and Linus clearly > > requested > > that code which changes behaviour depending on context should > > either be > > seperated or the context be conveyed in an argument passed by the > > caller, > > which usually knows the context. > > > > mlx5_eq_async_int() knows the context via the action argument > > already so > > using it for the lock variant decision is a straight forward > > replacement > > for in_irq(). > > > > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> > > Cc: Saeed Mahameed <saeedm@nvidia.com> > > Cc: Leon Romanovsky <leon@kernel.org> > > Cc: "David S. Miller" <davem@davemloft.net> > > Cc: Jakub Kicinski <kuba@kernel.org> > > Cc: linux-rdma@vger.kernel.org > > Saeed, please pick this up into your tree. Applied to net-next-mlx5 will submit to net-next shortly.
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c index 8ebfe782f95e5..3800e9415158b 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c @@ -189,19 +189,21 @@ u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq_comp *eq) return count_eqe; } -static void mlx5_eq_async_int_lock(struct mlx5_eq_async *eq, unsigned long *flags) +static void mlx5_eq_async_int_lock(struct mlx5_eq_async *eq, bool recovery, + unsigned long *flags) __acquires(&eq->lock) { - if (in_irq()) + if (!recovery) spin_lock(&eq->lock); else spin_lock_irqsave(&eq->lock, *flags); } -static void mlx5_eq_async_int_unlock(struct mlx5_eq_async *eq, unsigned long *flags) +static void mlx5_eq_async_int_unlock(struct mlx5_eq_async *eq, bool recovery, + unsigned long *flags) __releases(&eq->lock) { - if (in_irq()) + if (!recovery) spin_unlock(&eq->lock); else spin_unlock_irqrestore(&eq->lock, *flags); @@ -222,12 +224,14 @@ static int mlx5_eq_async_int(struct notifier_block *nb, struct mlx5_core_dev *dev; struct mlx5_eqe *eqe; unsigned long flags; + bool recovery; int num_eqes = 0; dev = eq->dev; eqt = dev->priv.eq_table; - mlx5_eq_async_int_lock(eq_async, &flags); + recovery = action == ASYNC_EQ_RECOVER; + mlx5_eq_async_int_lock(eq_async, recovery, &flags); eqe = next_eqe_sw(eq); if (!eqe) @@ -249,9 +253,9 @@ static int mlx5_eq_async_int(struct notifier_block *nb, out: eq_update_ci(eq, 1); - mlx5_eq_async_int_unlock(eq_async, &flags); + mlx5_eq_async_int_unlock(eq_async, recovery, &flags); - return unlikely(action == ASYNC_EQ_RECOVER) ? num_eqes : 0; + return unlikely(recovery) ? num_eqes : 0; } void mlx5_cmd_eq_recover(struct mlx5_core_dev *dev)
mlx5_eq_async_int() uses in_irq() to decide whether eq::lock needs to be acquired and released with spin_[un]lock() or the irq saving/restoring variants. The usage of in_*() in drivers is phased out and Linus clearly requested that code which changes behaviour depending on context should either be seperated or the context be conveyed in an argument passed by the caller, which usually knows the context. mlx5_eq_async_int() knows the context via the action argument already so using it for the lock variant decision is a straight forward replacement for in_irq(). Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Saeed Mahameed <saeedm@nvidia.com> Cc: Leon Romanovsky <leon@kernel.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Jakub Kicinski <kuba@kernel.org> Cc: linux-rdma@vger.kernel.org --- drivers/net/ethernet/mellanox/mlx5/core/eq.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)