Revision 63cae12bce9861cec309798d34701cf3da20bc71 authored by Peter Zijlstra on 09 December 2016, 13:59:00 UTC, committed by Ingo Molnar on 14 January 2017, 09:56:10 UTC
There is problem with installing an event in a task that is 'stuck' on
an offline CPU.
Blocked tasks are not dis-assosciated from offlined CPUs, after all, a
blocked task doesn't run and doesn't require a CPU etc.. Only on
wakeup do we ammend the situation and place the task on a available
CPU.
If we hit such a task with perf_install_in_context() we'll loop until
either that task wakes up or the CPU comes back online, if the task
waking depends on the event being installed, we're stuck.
While looking into this issue, I also spotted another problem, if we
hit a task with perf_install_in_context() that is in the middle of
being migrated, that is we observe the old CPU before sending the IPI,
but run the IPI (on the old CPU) while the task is already running on
the new CPU, things also go sideways.
Rework things to rely on task_curr() -- outside of rq->lock -- which
is rather tricky. Imagine the following scenario where we're trying to
install the first event into our task 't':
CPU0 CPU1 CPU2
(current == t)
t->perf_event_ctxp[] = ctx;
smp_mb();
cpu = task_cpu(t);
switch(t, n);
migrate(t, 2);
switch(p, t);
ctx = t->perf_event_ctxp[]; // must not be NULL
smp_function_call(cpu, ..);
generic_exec_single()
func();
spin_lock(ctx->lock);
if (task_curr(t)) // false
add_event_to_ctx();
spin_unlock(ctx->lock);
perf_event_context_sched_in();
spin_lock(ctx->lock);
// sees event
So its CPU0's store of t->perf_event_ctxp[] that must not go 'missing'.
Because if CPU2's load of that variable were to observe NULL, it would
not try to schedule the ctx and we'd have a task running without its
counter, which would be 'bad'.
As long as we observe !NULL, we'll acquire ctx->lock. If we acquire it
first and not see the event yet, then CPU0 must observe task_curr()
and retry. If the install happens first, then we must see the event on
sched-in and all is well.
I think we can translate the first part (until the 'must not be NULL')
of the scenario to a litmus test like:
C C-peterz
{
}
P0(int *x, int *y)
{
int r1;
WRITE_ONCE(*x, 1);
smp_mb();
r1 = READ_ONCE(*y);
}
P1(int *y, int *z)
{
WRITE_ONCE(*y, 1);
smp_store_release(z, 1);
}
P2(int *x, int *z)
{
int r1;
int r2;
r1 = smp_load_acquire(z);
smp_mb();
r2 = READ_ONCE(*x);
}
exists
(0:r1=0 /\ 2:r1=1 /\ 2:r2=0)
Where:
x is perf_event_ctxp[],
y is our tasks's CPU, and
z is our task being placed on the rq of CPU2.
The P0 smp_mb() is the one added by this patch, ordering the store to
perf_event_ctxp[] from find_get_context() and the load of task_cpu()
in task_function_call().
The smp_store_release/smp_load_acquire model the RCpc locking of the
rq->lock and the smp_mb() of P2 is the context switch switching from
whatever CPU2 was running to our task 't'.
This litmus test evaluates into:
Test C-peterz Allowed
States 7
0:r1=0; 2:r1=0; 2:r2=0;
0:r1=0; 2:r1=0; 2:r2=1;
0:r1=0; 2:r1=1; 2:r2=1;
0:r1=1; 2:r1=0; 2:r2=0;
0:r1=1; 2:r1=0; 2:r2=1;
0:r1=1; 2:r1=1; 2:r2=0;
0:r1=1; 2:r1=1; 2:r2=1;
No
Witnesses
Positive: 0 Negative: 7
Condition exists (0:r1=0 /\ 2:r1=1 /\ 2:r2=0)
Observation C-peterz Never 0 7
Hash=e427f41d9146b2a5445101d3e2fcaa34
And the strong and weak model agree.
Reported-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Will Deacon <will.deacon@arm.com>
Cc: jeremy.linton@arm.com
Link: http://lkml.kernel.org/r/20161209135900.GU3174@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
1 parent ad5013d
electra_cf.c
/*
* Copyright (C) 2007 PA Semi, Inc
*
* Maintained by: Olof Johansson <olof@lixom.net>
*
* Based on drivers/pcmcia/omap_cf.c
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/platform_device.h>
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/interrupt.h>
#include <linux/mm.h>
#include <linux/vmalloc.h>
#include <linux/of_address.h>
#include <linux/of_irq.h>
#include <linux/of_platform.h>
#include <linux/slab.h>
#include <pcmcia/ss.h>
static const char driver_name[] = "electra-cf";
struct electra_cf_socket {
struct pcmcia_socket socket;
struct timer_list timer;
unsigned present:1;
unsigned active:1;
struct platform_device *ofdev;
unsigned long mem_phys;
void __iomem *mem_base;
unsigned long mem_size;
void __iomem *io_virt;
unsigned int io_base;
unsigned int io_size;
u_int irq;
struct resource iomem;
void __iomem *gpio_base;
int gpio_detect;
int gpio_vsense;
int gpio_3v;
int gpio_5v;
};
#define POLL_INTERVAL (2 * HZ)
static int electra_cf_present(struct electra_cf_socket *cf)
{
unsigned int gpio;
gpio = in_le32(cf->gpio_base+0x40);
return !(gpio & (1 << cf->gpio_detect));
}
static int electra_cf_ss_init(struct pcmcia_socket *s)
{
return 0;
}
/* the timer is primarily to kick this socket's pccardd */
static void electra_cf_timer(unsigned long _cf)
{
struct electra_cf_socket *cf = (void *) _cf;
int present = electra_cf_present(cf);
if (present != cf->present) {
cf->present = present;
pcmcia_parse_events(&cf->socket, SS_DETECT);
}
if (cf->active)
mod_timer(&cf->timer, jiffies + POLL_INTERVAL);
}
static irqreturn_t electra_cf_irq(int irq, void *_cf)
{
electra_cf_timer((unsigned long)_cf);
return IRQ_HANDLED;
}
static int electra_cf_get_status(struct pcmcia_socket *s, u_int *sp)
{
struct electra_cf_socket *cf;
if (!sp)
return -EINVAL;
cf = container_of(s, struct electra_cf_socket, socket);
/* NOTE CF is always 3VCARD */
if (electra_cf_present(cf)) {
*sp = SS_READY | SS_DETECT | SS_POWERON | SS_3VCARD;
s->pci_irq = cf->irq;
} else
*sp = 0;
return 0;
}
static int electra_cf_set_socket(struct pcmcia_socket *sock,
struct socket_state_t *s)
{
unsigned int gpio;
unsigned int vcc;
struct electra_cf_socket *cf;
cf = container_of(sock, struct electra_cf_socket, socket);
/* "reset" means no power in our case */
vcc = (s->flags & SS_RESET) ? 0 : s->Vcc;
switch (vcc) {
case 0:
gpio = 0;
break;
case 33:
gpio = (1 << cf->gpio_3v);
break;
case 5:
gpio = (1 << cf->gpio_5v);
break;
default:
return -EINVAL;
}
gpio |= 1 << (cf->gpio_3v + 16); /* enwr */
gpio |= 1 << (cf->gpio_5v + 16); /* enwr */
out_le32(cf->gpio_base+0x90, gpio);
pr_debug("%s: Vcc %d, io_irq %d, flags %04x csc %04x\n",
driver_name, s->Vcc, s->io_irq, s->flags, s->csc_mask);
return 0;
}
static int electra_cf_set_io_map(struct pcmcia_socket *s,
struct pccard_io_map *io)
{
return 0;
}
static int electra_cf_set_mem_map(struct pcmcia_socket *s,
struct pccard_mem_map *map)
{
struct electra_cf_socket *cf;
if (map->card_start)
return -EINVAL;
cf = container_of(s, struct electra_cf_socket, socket);
map->static_start = cf->mem_phys;
map->flags &= MAP_ACTIVE|MAP_ATTRIB;
if (!(map->flags & MAP_ATTRIB))
map->static_start += 0x800;
return 0;
}
static struct pccard_operations electra_cf_ops = {
.init = electra_cf_ss_init,
.get_status = electra_cf_get_status,
.set_socket = electra_cf_set_socket,
.set_io_map = electra_cf_set_io_map,
.set_mem_map = electra_cf_set_mem_map,
};
static int electra_cf_probe(struct platform_device *ofdev)
{
struct device *device = &ofdev->dev;
struct device_node *np = ofdev->dev.of_node;
struct electra_cf_socket *cf;
struct resource mem, io;
int status;
const unsigned int *prop;
int err;
struct vm_struct *area;
err = of_address_to_resource(np, 0, &mem);
if (err)
return -EINVAL;
err = of_address_to_resource(np, 1, &io);
if (err)
return -EINVAL;
cf = kzalloc(sizeof(*cf), GFP_KERNEL);
if (!cf)
return -ENOMEM;
setup_timer(&cf->timer, electra_cf_timer, (unsigned long)cf);
cf->irq = NO_IRQ;
cf->ofdev = ofdev;
cf->mem_phys = mem.start;
cf->mem_size = PAGE_ALIGN(resource_size(&mem));
cf->mem_base = ioremap(cf->mem_phys, cf->mem_size);
cf->io_size = PAGE_ALIGN(resource_size(&io));
area = __get_vm_area(cf->io_size, 0, PHB_IO_BASE, PHB_IO_END);
if (area == NULL) {
status = -ENOMEM;
goto fail1;
}
cf->io_virt = (void __iomem *)(area->addr);
cf->gpio_base = ioremap(0xfc103000, 0x1000);
dev_set_drvdata(device, cf);
if (!cf->mem_base || !cf->io_virt || !cf->gpio_base ||
(__ioremap_at(io.start, cf->io_virt, cf->io_size,
pgprot_val(pgprot_noncached(__pgprot(0)))) == NULL)) {
dev_err(device, "can't ioremap ranges\n");
status = -ENOMEM;
goto fail1;
}
cf->io_base = (unsigned long)cf->io_virt - VMALLOC_END;
cf->iomem.start = (unsigned long)cf->mem_base;
cf->iomem.end = (unsigned long)cf->mem_base + (mem.end - mem.start);
cf->iomem.flags = IORESOURCE_MEM;
cf->irq = irq_of_parse_and_map(np, 0);
status = request_irq(cf->irq, electra_cf_irq, IRQF_SHARED,
driver_name, cf);
if (status < 0) {
dev_err(device, "request_irq failed\n");
goto fail1;
}
cf->socket.pci_irq = cf->irq;
prop = of_get_property(np, "card-detect-gpio", NULL);
if (!prop)
goto fail1;
cf->gpio_detect = *prop;
prop = of_get_property(np, "card-vsense-gpio", NULL);
if (!prop)
goto fail1;
cf->gpio_vsense = *prop;
prop = of_get_property(np, "card-3v-gpio", NULL);
if (!prop)
goto fail1;
cf->gpio_3v = *prop;
prop = of_get_property(np, "card-5v-gpio", NULL);
if (!prop)
goto fail1;
cf->gpio_5v = *prop;
cf->socket.io_offset = cf->io_base;
/* reserve chip-select regions */
if (!request_mem_region(cf->mem_phys, cf->mem_size, driver_name)) {
status = -ENXIO;
dev_err(device, "Can't claim memory region\n");
goto fail1;
}
if (!request_region(cf->io_base, cf->io_size, driver_name)) {
status = -ENXIO;
dev_err(device, "Can't claim I/O region\n");
goto fail2;
}
cf->socket.owner = THIS_MODULE;
cf->socket.dev.parent = &ofdev->dev;
cf->socket.ops = &electra_cf_ops;
cf->socket.resource_ops = &pccard_static_ops;
cf->socket.features = SS_CAP_PCCARD | SS_CAP_STATIC_MAP |
SS_CAP_MEM_ALIGN;
cf->socket.map_size = 0x800;
status = pcmcia_register_socket(&cf->socket);
if (status < 0) {
dev_err(device, "pcmcia_register_socket failed\n");
goto fail3;
}
dev_info(device, "at mem 0x%lx io 0x%llx irq %d\n",
cf->mem_phys, io.start, cf->irq);
cf->active = 1;
electra_cf_timer((unsigned long)cf);
return 0;
fail3:
release_region(cf->io_base, cf->io_size);
fail2:
release_mem_region(cf->mem_phys, cf->mem_size);
fail1:
if (cf->irq != NO_IRQ)
free_irq(cf->irq, cf);
if (cf->io_virt)
__iounmap_at(cf->io_virt, cf->io_size);
if (cf->mem_base)
iounmap(cf->mem_base);
if (cf->gpio_base)
iounmap(cf->gpio_base);
if (area)
device_init_wakeup(&ofdev->dev, 0);
kfree(cf);
return status;
}
static int electra_cf_remove(struct platform_device *ofdev)
{
struct device *device = &ofdev->dev;
struct electra_cf_socket *cf;
cf = dev_get_drvdata(device);
cf->active = 0;
pcmcia_unregister_socket(&cf->socket);
free_irq(cf->irq, cf);
del_timer_sync(&cf->timer);
__iounmap_at(cf->io_virt, cf->io_size);
iounmap(cf->mem_base);
iounmap(cf->gpio_base);
release_mem_region(cf->mem_phys, cf->mem_size);
release_region(cf->io_base, cf->io_size);
kfree(cf);
return 0;
}
static const struct of_device_id electra_cf_match[] = {
{
.compatible = "electra-cf",
},
{},
};
MODULE_DEVICE_TABLE(of, electra_cf_match);
static struct platform_driver electra_cf_driver = {
.driver = {
.name = driver_name,
.of_match_table = electra_cf_match,
},
.probe = electra_cf_probe,
.remove = electra_cf_remove,
};
module_platform_driver(electra_cf_driver);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Olof Johansson <olof@lixom.net>");
MODULE_DESCRIPTION("PA Semi Electra CF driver");
Computing file changes ...
