linux_dsm_epyc7002/drivers/platform/mellanox/mlxreg-hotplug.c

693 lines
19 KiB
C
Raw Normal View History

/*
* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved.
* Copyright (c) 2016-2018 Vadim Pasternak <vadimp@mellanox.com>
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the names of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* Alternatively, this software may be distributed under the terms of the
* GNU General Public License ("GPL") version 2 as published by the Free
* Software Foundation.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <linux/bitops.h>
#include <linux/device.h>
#include <linux/hwmon.h>
#include <linux/hwmon-sysfs.h>
#include <linux/i2c.h>
#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/of_device.h>
#include <linux/platform_data/mlxreg.h>
#include <linux/platform_device.h>
#include <linux/spinlock.h>
#include <linux/regmap.h>
#include <linux/workqueue.h>
/* Offset of event and mask registers from status register. */
#define MLXREG_HOTPLUG_EVENT_OFF 1
#define MLXREG_HOTPLUG_MASK_OFF 2
#define MLXREG_HOTPLUG_AGGR_MASK_OFF 1
/* ASIC good health mask. */
#define MLXREG_HOTPLUG_GOOD_HEALTH_MASK 0x02
#define MLXREG_HOTPLUG_ATTRS_MAX 24
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
#define MLXREG_HOTPLUG_NOT_ASSERT 3
/**
* struct mlxreg_hotplug_priv_data - platform private data:
* @irq: platform device interrupt number;
* @dev: basic device;
* @pdev: platform device;
* @plat: platform data;
* @regmap: register map handle;
* @dwork_irq: delayed work template;
* @lock: spin lock;
* @hwmon: hwmon device;
* @mlxreg_hotplug_attr: sysfs attributes array;
* @mlxreg_hotplug_dev_attr: sysfs sensor device attribute array;
* @group: sysfs attribute group;
* @groups: list of sysfs attribute group for hwmon registration;
* @cell: location of top aggregation interrupt register;
* @mask: top aggregation interrupt common mask;
* @aggr_cache: last value of aggregation register status;
* @after_probe: flag indication probing completion;
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
* @not_asserted: number of entries in workqueue with no signal assertion;
*/
struct mlxreg_hotplug_priv_data {
int irq;
struct device *dev;
struct platform_device *pdev;
struct mlxreg_hotplug_platform_data *plat;
struct regmap *regmap;
struct delayed_work dwork_irq;
spinlock_t lock; /* sync with interrupt */
struct device *hwmon;
struct attribute *mlxreg_hotplug_attr[MLXREG_HOTPLUG_ATTRS_MAX + 1];
struct sensor_device_attribute_2
mlxreg_hotplug_dev_attr[MLXREG_HOTPLUG_ATTRS_MAX];
struct attribute_group group;
const struct attribute_group *groups[2];
u32 cell;
u32 mask;
u32 aggr_cache;
bool after_probe;
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
u8 not_asserted;
};
static int mlxreg_hotplug_device_create(struct mlxreg_hotplug_priv_data *priv,
struct mlxreg_core_data *data)
{
struct mlxreg_core_hotplug_platform_data *pdata;
/* Notify user by sending hwmon uevent. */
kobject_uevent(&priv->hwmon->kobj, KOBJ_CHANGE);
/*
* Return if adapter number is negative. It could be in case hotplug
* event is not associated with hotplug device.
*/
if (data->hpdev.nr < 0)
return 0;
pdata = dev_get_platdata(&priv->pdev->dev);
data->hpdev.adapter = i2c_get_adapter(data->hpdev.nr +
pdata->shift_nr);
if (!data->hpdev.adapter) {
dev_err(priv->dev, "Failed to get adapter for bus %d\n",
data->hpdev.nr + pdata->shift_nr);
return -EFAULT;
}
data->hpdev.client = i2c_new_device(data->hpdev.adapter,
data->hpdev.brdinfo);
if (!data->hpdev.client) {
dev_err(priv->dev, "Failed to create client %s at bus %d at addr 0x%02x\n",
data->hpdev.brdinfo->type, data->hpdev.nr +
pdata->shift_nr, data->hpdev.brdinfo->addr);
i2c_put_adapter(data->hpdev.adapter);
data->hpdev.adapter = NULL;
return -EFAULT;
}
return 0;
}
static void
mlxreg_hotplug_device_destroy(struct mlxreg_hotplug_priv_data *priv,
struct mlxreg_core_data *data)
{
/* Notify user by sending hwmon uevent. */
kobject_uevent(&priv->hwmon->kobj, KOBJ_CHANGE);
if (data->hpdev.client) {
i2c_unregister_device(data->hpdev.client);
data->hpdev.client = NULL;
}
if (data->hpdev.adapter) {
i2c_put_adapter(data->hpdev.adapter);
data->hpdev.adapter = NULL;
}
}
static ssize_t mlxreg_hotplug_attr_show(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct mlxreg_hotplug_priv_data *priv = dev_get_drvdata(dev);
struct mlxreg_core_hotplug_platform_data *pdata;
int index = to_sensor_dev_attr_2(attr)->index;
int nr = to_sensor_dev_attr_2(attr)->nr;
struct mlxreg_core_item *item;
struct mlxreg_core_data *data;
u32 regval;
int ret;
pdata = dev_get_platdata(&priv->pdev->dev);
item = pdata->items + nr;
data = item->data + index;
ret = regmap_read(priv->regmap, data->reg, &regval);
if (ret)
return ret;
if (item->health) {
regval &= data->mask;
} else {
/* Bit = 0 : functional if item->inversed is true. */
if (item->inversed)
regval = !(regval & data->mask);
else
regval = !!(regval & data->mask);
}
return sprintf(buf, "%u\n", regval);
}
#define PRIV_ATTR(i) priv->mlxreg_hotplug_attr[i]
#define PRIV_DEV_ATTR(i) priv->mlxreg_hotplug_dev_attr[i]
static int mlxreg_hotplug_attr_init(struct mlxreg_hotplug_priv_data *priv)
{
struct mlxreg_core_hotplug_platform_data *pdata;
struct mlxreg_core_item *item;
struct mlxreg_core_data *data;
int num_attrs = 0, id = 0, i, j;
pdata = dev_get_platdata(&priv->pdev->dev);
item = pdata->items;
/* Go over all kinds of items - psu, pwr, fan. */
for (i = 0; i < pdata->counter; i++, item++) {
num_attrs += item->count;
data = item->data;
/* Go over all units within the item. */
for (j = 0; j < item->count; j++, data++, id++) {
PRIV_ATTR(id) = &PRIV_DEV_ATTR(id).dev_attr.attr;
PRIV_ATTR(id)->name = devm_kasprintf(&priv->pdev->dev,
GFP_KERNEL,
data->label);
if (!PRIV_ATTR(id)->name) {
dev_err(priv->dev, "Memory allocation failed for attr %d.\n",
id);
return -ENOMEM;
}
PRIV_DEV_ATTR(id).dev_attr.attr.name =
PRIV_ATTR(id)->name;
PRIV_DEV_ATTR(id).dev_attr.attr.mode = 0444;
PRIV_DEV_ATTR(id).dev_attr.show =
mlxreg_hotplug_attr_show;
PRIV_DEV_ATTR(id).nr = i;
PRIV_DEV_ATTR(id).index = j;
sysfs_attr_init(&PRIV_DEV_ATTR(id).dev_attr.attr);
}
}
treewide: devm_kzalloc() -> devm_kcalloc() The devm_kzalloc() function has a 2-factor argument form, devm_kcalloc(). This patch replaces cases of: devm_kzalloc(handle, a * b, gfp) with: devm_kcalloc(handle, a * b, gfp) as well as handling cases of: devm_kzalloc(handle, a * b * c, gfp) with: devm_kzalloc(handle, array3_size(a, b, c), gfp) as it's slightly less ugly than: devm_kcalloc(handle, array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: devm_kzalloc(handle, 4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. Some manual whitespace fixes were needed in this patch, as Coccinelle really liked to write "=devm_kcalloc..." instead of "= devm_kcalloc...". The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ expression HANDLE; type TYPE; expression THING, E; @@ ( devm_kzalloc(HANDLE, - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | devm_kzalloc(HANDLE, - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression HANDLE; expression COUNT; typedef u8; typedef __u8; @@ ( devm_kzalloc(HANDLE, - sizeof(u8) * (COUNT) + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(__u8) * (COUNT) + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(char) * (COUNT) + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(unsigned char) * (COUNT) + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(u8) * COUNT + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(__u8) * COUNT + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(char) * COUNT + COUNT , ...) | devm_kzalloc(HANDLE, - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ expression HANDLE; type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ expression HANDLE; identifier SIZE, COUNT; @@ - devm_kzalloc + devm_kcalloc (HANDLE, - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression HANDLE; expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( devm_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression HANDLE; expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( devm_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | devm_kzalloc(HANDLE, - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | devm_kzalloc(HANDLE, - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ expression HANDLE; identifier STRIDE, SIZE, COUNT; @@ ( devm_kzalloc(HANDLE, - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | devm_kzalloc(HANDLE, - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression HANDLE; expression E1, E2, E3; constant C1, C2, C3; @@ ( devm_kzalloc(HANDLE, C1 * C2 * C3, ...) | devm_kzalloc(HANDLE, - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | devm_kzalloc(HANDLE, - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | devm_kzalloc(HANDLE, - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | devm_kzalloc(HANDLE, - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression HANDLE; expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( devm_kzalloc(HANDLE, sizeof(THING) * C2, ...) | devm_kzalloc(HANDLE, sizeof(TYPE) * C2, ...) | devm_kzalloc(HANDLE, C1 * C2 * C3, ...) | devm_kzalloc(HANDLE, C1 * C2, ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - (E1) * E2 + E1, E2 , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - (E1) * (E2) + E1, E2 , ...) | - devm_kzalloc + devm_kcalloc (HANDLE, - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-13 04:07:58 +07:00
priv->group.attrs = devm_kcalloc(&priv->pdev->dev,
num_attrs,
sizeof(struct attribute *),
GFP_KERNEL);
if (!priv->group.attrs)
return -ENOMEM;
priv->group.attrs = priv->mlxreg_hotplug_attr;
priv->groups[0] = &priv->group;
priv->groups[1] = NULL;
return 0;
}
static void
mlxreg_hotplug_work_helper(struct mlxreg_hotplug_priv_data *priv,
struct mlxreg_core_item *item)
{
struct mlxreg_core_data *data;
u32 asserted, regval, bit;
int ret;
/*
* Validate if item related to received signal type is valid.
* It should never happen, excepted the situation when some
* piece of hardware is broken. In such situation just produce
* error message and return. Caller must continue to handle the
* signals from other devices if any.
*/
if (unlikely(!item)) {
dev_err(priv->dev, "False signal: at offset:mask 0x%02x:0x%02x.\n",
item->reg, item->mask);
return;
}
/* Mask event. */
ret = regmap_write(priv->regmap, item->reg + MLXREG_HOTPLUG_MASK_OFF,
0);
if (ret)
goto out;
/* Read status. */
ret = regmap_read(priv->regmap, item->reg, &regval);
if (ret)
goto out;
/* Set asserted bits and save last status. */
regval &= item->mask;
asserted = item->cache ^ regval;
item->cache = regval;
for_each_set_bit(bit, (unsigned long *)&asserted, 8) {
data = item->data + bit;
if (regval & BIT(bit)) {
if (item->inversed)
mlxreg_hotplug_device_destroy(priv, data);
else
mlxreg_hotplug_device_create(priv, data);
} else {
if (item->inversed)
mlxreg_hotplug_device_create(priv, data);
else
mlxreg_hotplug_device_destroy(priv, data);
}
}
/* Acknowledge event. */
ret = regmap_write(priv->regmap, item->reg + MLXREG_HOTPLUG_EVENT_OFF,
0);
if (ret)
goto out;
/* Unmask event. */
ret = regmap_write(priv->regmap, item->reg + MLXREG_HOTPLUG_MASK_OFF,
item->mask);
out:
if (ret)
dev_err(priv->dev, "Failed to complete workqueue.\n");
}
static void
mlxreg_hotplug_health_work_helper(struct mlxreg_hotplug_priv_data *priv,
struct mlxreg_core_item *item)
{
struct mlxreg_core_data *data = item->data;
u32 regval;
int i, ret = 0;
for (i = 0; i < item->count; i++, data++) {
/* Mask event. */
ret = regmap_write(priv->regmap, data->reg +
MLXREG_HOTPLUG_MASK_OFF, 0);
if (ret)
goto out;
/* Read status. */
ret = regmap_read(priv->regmap, data->reg, &regval);
if (ret)
goto out;
regval &= data->mask;
if (item->cache == regval)
goto ack_event;
/*
* ASIC health indication is provided through two bits. Bits
* value 0x2 indicates that ASIC reached the good health, value
* 0x0 indicates ASIC the bad health or dormant state and value
* 0x3 indicates the booting state. During ASIC reset it should
* pass the following states: dormant -> booting -> good.
*/
if (regval == MLXREG_HOTPLUG_GOOD_HEALTH_MASK) {
if (!data->attached) {
/*
* ASIC is in steady state. Connect associated
* device, if configured.
*/
mlxreg_hotplug_device_create(priv, data);
data->attached = true;
}
} else {
if (data->attached) {
/*
* ASIC health is failed after ASIC has been
* in steady state. Disconnect associated
* device, if it has been connected.
*/
mlxreg_hotplug_device_destroy(priv, data);
data->attached = false;
data->health_cntr = 0;
}
}
item->cache = regval;
ack_event:
/* Acknowledge event. */
ret = regmap_write(priv->regmap, data->reg +
MLXREG_HOTPLUG_EVENT_OFF, 0);
if (ret)
goto out;
/* Unmask event. */
ret = regmap_write(priv->regmap, data->reg +
MLXREG_HOTPLUG_MASK_OFF, data->mask);
if (ret)
goto out;
}
out:
if (ret)
dev_err(priv->dev, "Failed to complete workqueue.\n");
}
/*
* mlxreg_hotplug_work_handler - performs traversing of device interrupt
* registers according to the below hierarchy schema:
*
* Aggregation registers (status/mask)
* PSU registers: *---*
* *-----------------* | |
* |status/event/mask|-----> | * |
* *-----------------* | |
* Power registers: | |
* *-----------------* | |
* |status/event/mask|-----> | * |
* *-----------------* | |
* FAN registers: | |--> CPU
* *-----------------* | |
* |status/event/mask|-----> | * |
* *-----------------* | |
* ASIC registers: | |
* *-----------------* | |
* |status/event/mask|-----> | * |
* *-----------------* | |
* *---*
*
* In case some system changed are detected: FAN in/out, PSU in/out, power
* cable attached/detached, ASIC health good/bad, relevant device is created
* or destroyed.
*/
static void mlxreg_hotplug_work_handler(struct work_struct *work)
{
struct mlxreg_core_hotplug_platform_data *pdata;
struct mlxreg_hotplug_priv_data *priv;
struct mlxreg_core_item *item;
u32 regval, aggr_asserted;
unsigned long flags;
int i, ret;
priv = container_of(work, struct mlxreg_hotplug_priv_data,
dwork_irq.work);
pdata = dev_get_platdata(&priv->pdev->dev);
item = pdata->items;
/* Mask aggregation event. */
ret = regmap_write(priv->regmap, pdata->cell +
MLXREG_HOTPLUG_AGGR_MASK_OFF, 0);
if (ret < 0)
goto out;
/* Read aggregation status. */
ret = regmap_read(priv->regmap, pdata->cell, &regval);
if (ret)
goto out;
regval &= pdata->mask;
aggr_asserted = priv->aggr_cache ^ regval;
priv->aggr_cache = regval;
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
/*
* Handler is invoked, but no assertion is detected at top aggregation
* status level. Set aggr_asserted to mask value to allow handler extra
* run over all relevant signals to recover any missed signal.
*/
if (priv->not_asserted == MLXREG_HOTPLUG_NOT_ASSERT) {
priv->not_asserted = 0;
aggr_asserted = pdata->mask;
}
if (!aggr_asserted)
goto unmask_event;
/* Handle topology and health configuration changes. */
for (i = 0; i < pdata->counter; i++, item++) {
if (aggr_asserted & item->aggr_mask) {
if (item->health)
mlxreg_hotplug_health_work_helper(priv, item);
else
mlxreg_hotplug_work_helper(priv, item);
}
}
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
spin_lock_irqsave(&priv->lock, flags);
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
/*
* It is possible, that some signals have been inserted, while
* interrupt has been masked by mlxreg_hotplug_work_handler. In this
* case such signals will be missed. In order to handle these signals
* delayed work is canceled and work task re-scheduled for immediate
* execution. It allows to handle missed signals, if any. In other case
* work handler just validates that no new signals have been received
* during masking.
*/
cancel_delayed_work(&priv->dwork_irq);
schedule_delayed_work(&priv->dwork_irq, 0);
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
spin_unlock_irqrestore(&priv->lock, flags);
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
return;
platform/mellanox: mlxreg-hotplug: add extra cycle for hotplug work queue Add extra cycle for hotplug work queue to handle the case when a signal is It adds missed logic for signal acknowledge, by adding an extra run for received, but no specific signal assertion is detected. Such case theoretically can happen for example in case several units are removed or inserted at the same time. In this situation acknowledge for some signal can be missed at signal top aggreagation status level. The extra run will allow to handler to acknowledge the missed signal. The interrupt handling flow performs the next steps: (1) Enter mlxreg_hotplug_work_handler due to signal assertion. Aggregation status register is changed for example from 0xff to 0xfd (event signal group related to bit 1). (2) Mask aggregation interrupts, read aggregation status register and save it (0xfd) in aggr_cache, then traverse down to handle signal from groups related to the changed bit. (3) Read and mask group related signal. Acknowledge and unmask group related signal (acknowledge should clear aggregation status register from 0xfd back to 0xff). (4) Re-schedule work queue for the immediate execution. (5) Enter mlxreg_hotplug_work_handler due to re-scheduling. Aggregation status is changed from previous 0xfd to 0xff. Go over steps (2) - (5) and in case no new signal assertion is detected - unmask aggregation interrupts. The possible race could happen in case new signal from the same group is asserted after step (3) and prior step (5). In such case aggregation status will change back from 0xff to 0xfd and the value read from the aggregation status register will be the same as a value saved in aggr_cache. As a result the handler will not traverse down and signal will stay unhandled. Example of faulty flow: The signal routing flow is as following (f.e. for of FANi removing): - FAN status and event registers related bit is changed; -- intermediate aggregation status register is changed; --- top aggregation status register is changed; ---- interrupt routed to CPU and interrupt handler is invoked. When interrupt handler is invoked it follows the next simple logic (f.e FAN3 is removed): (a1) mask top aggregation interrupt mask register; (a2) read top aggregation interrupt status register and test to which underling group belongs a signal (FANs in this case and is changed from 0xff to 0xfb and 0xfb is saved as a last status value); (b1) mask FANs interrupt mask register; (b2) read FANs status register and test which FAN has been changed FAN3 in this example); (c1) perform relevant action; <--------------- (FAN2 is removed at this point) (b3) clear FANs interrupt event register to acknowledge FAN3 signal; (b4) unmask FANs interrupt mask register (a3) unmask top aggregation interrupt mask register; An interrupt handler is invoked, since FAN2 interrupt is not acknowledge. It should set top aggregation interrupt status register bit 6 (0xfb). In step (a2) (a2) read top aggregation interrupt and comparing it with saved value does not show change (same 0xfb) and after (a2) execution jumps to (a3) and signal leaved unhandled The fix will enforce handler to traverse down in case the signal is received, but signal assertion is not detected. Fixes: 304887041d95 ("platform/x86: Introduce support for Mellanox hotplug driver") Signed-off-by: Vadim Pasternak <vadimp@mellanox.com> Signed-off-by: Darren Hart (VMware) <dvhart@infradead.org>
2018-05-07 13:48:51 +07:00
unmask_event:
priv->not_asserted++;
/* Unmask aggregation event (no need acknowledge). */
ret = regmap_write(priv->regmap, pdata->cell +
MLXREG_HOTPLUG_AGGR_MASK_OFF, pdata->mask);
out:
if (ret)
dev_err(priv->dev, "Failed to complete workqueue.\n");
}
static int mlxreg_hotplug_set_irq(struct mlxreg_hotplug_priv_data *priv)
{
struct mlxreg_core_hotplug_platform_data *pdata;
struct mlxreg_core_item *item;
int i, ret;
pdata = dev_get_platdata(&priv->pdev->dev);
item = pdata->items;
for (i = 0; i < pdata->counter; i++, item++) {
/* Clear group presense event. */
ret = regmap_write(priv->regmap, item->reg +
MLXREG_HOTPLUG_EVENT_OFF, 0);
if (ret)
goto out;
/* Set group initial status as mask and unmask group event. */
if (item->inversed) {
item->cache = item->mask;
ret = regmap_write(priv->regmap, item->reg +
MLXREG_HOTPLUG_MASK_OFF,
item->mask);
if (ret)
goto out;
}
}
/* Keep aggregation initial status as zero and unmask events. */
ret = regmap_write(priv->regmap, pdata->cell +
MLXREG_HOTPLUG_AGGR_MASK_OFF, pdata->mask);
if (ret)
goto out;
/* Keep low aggregation initial status as zero and unmask events. */
if (pdata->cell_low) {
ret = regmap_write(priv->regmap, pdata->cell_low +
MLXREG_HOTPLUG_AGGR_MASK_OFF,
pdata->mask_low);
if (ret)
goto out;
}
/* Invoke work handler for initializing hot plug devices setting. */
mlxreg_hotplug_work_handler(&priv->dwork_irq.work);
out:
if (ret)
dev_err(priv->dev, "Failed to set interrupts.\n");
enable_irq(priv->irq);
return ret;
}
static void mlxreg_hotplug_unset_irq(struct mlxreg_hotplug_priv_data *priv)
{
struct mlxreg_core_hotplug_platform_data *pdata;
struct mlxreg_core_item *item;
struct mlxreg_core_data *data;
int count, i, j;
pdata = dev_get_platdata(&priv->pdev->dev);
item = pdata->items;
disable_irq(priv->irq);
cancel_delayed_work_sync(&priv->dwork_irq);
/* Mask low aggregation event, if defined. */
if (pdata->cell_low)
regmap_write(priv->regmap, pdata->cell_low +
MLXREG_HOTPLUG_AGGR_MASK_OFF, 0);
/* Mask aggregation event. */
regmap_write(priv->regmap, pdata->cell + MLXREG_HOTPLUG_AGGR_MASK_OFF,
0);
/* Clear topology configurations. */
for (i = 0; i < pdata->counter; i++, item++) {
data = item->data;
/* Mask group presense event. */
regmap_write(priv->regmap, data->reg + MLXREG_HOTPLUG_MASK_OFF,
0);
/* Clear group presense event. */
regmap_write(priv->regmap, data->reg +
MLXREG_HOTPLUG_EVENT_OFF, 0);
/* Remove all the attached devices in group. */
count = item->count;
for (j = 0; j < count; j++, data++)
mlxreg_hotplug_device_destroy(priv, data);
}
}
static irqreturn_t mlxreg_hotplug_irq_handler(int irq, void *dev)
{
struct mlxreg_hotplug_priv_data *priv;
priv = (struct mlxreg_hotplug_priv_data *)dev;
/* Schedule work task for immediate execution.*/
schedule_delayed_work(&priv->dwork_irq, 0);
return IRQ_HANDLED;
}
static int mlxreg_hotplug_probe(struct platform_device *pdev)
{
struct mlxreg_core_hotplug_platform_data *pdata;
struct mlxreg_hotplug_priv_data *priv;
struct i2c_adapter *deferred_adap;
int err;
pdata = dev_get_platdata(&pdev->dev);
if (!pdata) {
dev_err(&pdev->dev, "Failed to get platform data.\n");
return -EINVAL;
}
/* Defer probing if the necessary adapter is not configured yet. */
deferred_adap = i2c_get_adapter(pdata->deferred_nr);
if (!deferred_adap)
return -EPROBE_DEFER;
i2c_put_adapter(deferred_adap);
priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
if (!priv)
return -ENOMEM;
if (pdata->irq) {
priv->irq = pdata->irq;
} else {
priv->irq = platform_get_irq(pdev, 0);
if (priv->irq < 0) {
dev_err(&pdev->dev, "Failed to get platform irq: %d\n",
priv->irq);
return priv->irq;
}
}
priv->regmap = pdata->regmap;
priv->dev = pdev->dev.parent;
priv->pdev = pdev;
err = devm_request_irq(&pdev->dev, priv->irq,
mlxreg_hotplug_irq_handler, IRQF_TRIGGER_FALLING
| IRQF_SHARED, "mlxreg-hotplug", priv);
if (err) {
dev_err(&pdev->dev, "Failed to request irq: %d\n", err);
return err;
}
disable_irq(priv->irq);
spin_lock_init(&priv->lock);
INIT_DELAYED_WORK(&priv->dwork_irq, mlxreg_hotplug_work_handler);
dev_set_drvdata(&pdev->dev, priv);
err = mlxreg_hotplug_attr_init(priv);
if (err) {
dev_err(&pdev->dev, "Failed to allocate attributes: %d\n",
err);
return err;
}
priv->hwmon = devm_hwmon_device_register_with_groups(&pdev->dev,
"mlxreg_hotplug", priv, priv->groups);
if (IS_ERR(priv->hwmon)) {
dev_err(&pdev->dev, "Failed to register hwmon device %ld\n",
PTR_ERR(priv->hwmon));
return PTR_ERR(priv->hwmon);
}
/* Perform initial interrupts setup. */
mlxreg_hotplug_set_irq(priv);
priv->after_probe = true;
return 0;
}
static int mlxreg_hotplug_remove(struct platform_device *pdev)
{
struct mlxreg_hotplug_priv_data *priv = dev_get_drvdata(&pdev->dev);
/* Clean interrupts setup. */
mlxreg_hotplug_unset_irq(priv);
return 0;
}
static struct platform_driver mlxreg_hotplug_driver = {
.driver = {
.name = "mlxreg-hotplug",
},
.probe = mlxreg_hotplug_probe,
.remove = mlxreg_hotplug_remove,
};
module_platform_driver(mlxreg_hotplug_driver);
MODULE_AUTHOR("Vadim Pasternak <vadimp@mellanox.com>");
MODULE_DESCRIPTION("Mellanox regmap hotplug platform driver");
MODULE_LICENSE("Dual BSD/GPL");
MODULE_ALIAS("platform:mlxreg-hotplug");