openwrt/target/linux/mvebu/patches-3.10/0085-irqchip-armada-370-xp-fix-IPI-race-condition.patch
Luka Perkov 3af779eb17 mvebu: backport mainline patches from kernel 3.12
This is a backport of the patches accepted to the Linux mainline related to
mvebu SoC (Armada XP and Armada 370) between Linux v3.11, and Linux v3.12.
This work mainly covers:

* Ground work for sharing the pxa nand driver(drivers/mtd/nand/pxa3xx_nand.c)
  between the PXA family,and the Armada family.
* Further updates to the mvebu MBus.
* Work and ground work for enabling MSI on the Armada family.
* some phy / mdio bus initialization related work.
* Device tree binding documentation update.

Signed-off-by: Seif Mazareeb <seif.mazareeb@gmail.com>
CC: Luka Perkov <luka@openwrt.org>

SVN-Revision: 39565
2014-02-11 02:07:41 +00:00

47 lines
2.0 KiB
Diff

From 60b78d28057f9f1ecce39955755d26e4cd3130de Mon Sep 17 00:00:00 2001
From: Lior Amsalem <alior@marvell.com>
Date: Mon, 25 Nov 2013 17:26:44 +0100
Subject: [PATCH 085/203] irqchip: armada-370-xp: fix IPI race condition
In the Armada 370/XP driver, when we receive an IRQ 0, we read the
list of doorbells that caused the interrupt from register
ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS. This gives the list of IPIs that
were generated. However, instead of acknowledging only the IPIs that
were generated, we acknowledge *all* the IPIs, by writing
~IPI_DOORBELL_MASK in the ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS register.
This creates a race condition: if a new IPI that isn't part of the
ones read into the temporary "ipimask" variable is fired before we
acknowledge all IPIs, then we will simply loose it. This is causing
scheduling hangs on SMP intensive workloads.
It is important to mention that this ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS
register has the following behavior: "A CPU write of 0 clears the bits
in this field. A CPU write of 1 has no effect". This is what allows us
to simply write ~ipimask to acknoledge the handled IPIs.
Notice that the same problem is present in the MSI implementation, but
it will be fixed as a separate patch, so that this IPI fix can be
pushed to older stable versions as appropriate (all the way to 3.8),
while the MSI code only appeared in 3.13.
Signed-off-by: Lior Amsalem <alior@marvell.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: stable@vger.kernel.org
Cc: Thomas Gleixner <tglx@linutronix.de>
---
drivers/irqchip/irq-armada-370-xp.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/drivers/irqchip/irq-armada-370-xp.c
+++ b/drivers/irqchip/irq-armada-370-xp.c
@@ -407,7 +407,7 @@ armada_370_xp_handle_irq(struct pt_regs
ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS)
& IPI_DOORBELL_MASK;
- writel(~IPI_DOORBELL_MASK, per_cpu_int_base +
+ writel(~ipimask, per_cpu_int_base +
ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS);
/* Handle all pending doorbells */