[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1453926929-17663-185-git-send-email-kamal@canonical.com>
Date: Wed, 27 Jan 2016 12:34:05 -0800
From: Kamal Mostafa <kamal@...onical.com>
To: linux-kernel@...r.kernel.org, stable@...r.kernel.org,
kernel-team@...ts.ubuntu.com
Cc: Huacai Chen <chenhc@...ote.com>,
Aurelien Jarno <aurelien@...el32.net>,
"Steven J. Hill" <Steven.Hill@...tec.com>,
Fuxin Zhang <zhangfx@...ote.com>,
Zhangjin Wu <wuzhangjin@...il.com>, linux-mips@...ux-mips.org,
Ralf Baechle <ralf@...ux-mips.org>,
Kamal Mostafa <kamal@...onical.com>
Subject: [PATCH 4.2.y-ckt 184/268] MIPS: Loongson-3: Fix SMP_ASK_C0COUNT IPI handler
4.2.8-ckt3 -stable review patch. If anyone has any objections, please let me know.
---8<------------------------------------------------------------
From: Huacai Chen <chenhc@...ote.com>
commit 5754843225f78ac7cbe142a6899890a9733a5a5d upstream.
When Core-0 handle SMP_ASK_C0COUNT IPI, we should make other cores to
see the result as soon as possible (especially when Store-Fill-Buffer
is enabled). Otherwise, C0_Count syncronization makes no sense.
BTW, array is more suitable than per-cpu variable for syncronization,
and there is a corner case should be avoid: C0_Count of Core-0 can be
really 0.
Signed-off-by: Huacai Chen <chenhc@...ote.com>
Cc: Aurelien Jarno <aurelien@...el32.net>
Cc: Steven J. Hill <Steven.Hill@...tec.com>
Cc: Fuxin Zhang <zhangfx@...ote.com>
Cc: Zhangjin Wu <wuzhangjin@...il.com>
Cc: Huacai Chen <chenhc@...ote.com>
Cc: linux-mips@...ux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/12160/
Signed-off-by: Ralf Baechle <ralf@...ux-mips.org>
Signed-off-by: Kamal Mostafa <kamal@...onical.com>
---
arch/mips/loongson64/loongson-3/smp.c | 20 +++++++++++++-------
1 file changed, 13 insertions(+), 7 deletions(-)
diff --git a/arch/mips/loongson64/loongson-3/smp.c b/arch/mips/loongson64/loongson-3/smp.c
index 1a4738a..509832a9 100644
--- a/arch/mips/loongson64/loongson-3/smp.c
+++ b/arch/mips/loongson64/loongson-3/smp.c
@@ -30,13 +30,13 @@
#include "smp.h"
DEFINE_PER_CPU(int, cpu_state);
-DEFINE_PER_CPU(uint32_t, core0_c0count);
static void *ipi_set0_regs[16];
static void *ipi_clear0_regs[16];
static void *ipi_status0_regs[16];
static void *ipi_en0_regs[16];
static void *ipi_mailbox_buf[16];
+static uint32_t core0_c0count[NR_CPUS];
/* read a 32bit value from ipi register */
#define loongson3_ipi_read32(addr) readl(addr)
@@ -275,12 +275,14 @@ void loongson3_ipi_interrupt(struct pt_regs *regs)
if (action & SMP_ASK_C0COUNT) {
BUG_ON(cpu != 0);
c0count = read_c0_count();
- for (i = 1; i < num_possible_cpus(); i++)
- per_cpu(core0_c0count, i) = c0count;
+ c0count = c0count ? c0count : 1;
+ for (i = 1; i < nr_cpu_ids; i++)
+ core0_c0count[i] = c0count;
+ __wbflush(); /* Let others see the result ASAP */
}
}
-#define MAX_LOOPS 1111
+#define MAX_LOOPS 800
/*
* SMP init and finish on secondary CPUs
*/
@@ -305,16 +307,20 @@ static void loongson3_init_secondary(void)
cpu_logical_map(cpu) / loongson_sysconf.cores_per_package;
i = 0;
- __this_cpu_write(core0_c0count, 0);
+ core0_c0count[cpu] = 0;
loongson3_send_ipi_single(0, SMP_ASK_C0COUNT);
- while (!__this_cpu_read(core0_c0count)) {
+ while (!core0_c0count[cpu]) {
i++;
cpu_relax();
}
if (i > MAX_LOOPS)
i = MAX_LOOPS;
- initcount = __this_cpu_read(core0_c0count) + i;
+ if (cpu_data[cpu].package)
+ initcount = core0_c0count[cpu] + i;
+ else /* Local access is faster for loops */
+ initcount = core0_c0count[cpu] + i/2;
+
write_c0_count(initcount);
}
--
1.9.1
Powered by blists - more mailing lists