| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20170810202952.GA13007@roeck-us.net>
Date: Thu, 10 Aug 2017 13:29:52 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Eddie James <eajames@...ux.vnet.ibm.com>
Cc: jdelvare@...e.com, linux-hwmon@...r.kernel.org,
linux-kernel@...r.kernel.org, joel@....id.au, jk@...abs.org,
andrew@...id.au, cbostic@...ux.vnet.ibm.com,
"Edward A. James" <eajames@...ibm.com>
Subject: Re: [PATCH 4/4] drivers/hwmon/pmbus: Add debugfs for status registers
On Thu, Aug 10, 2017 at 03:15:57PM -0500, Eddie James wrote:
>
>
> On 08/09/2017 08:15 PM, Guenter Roeck wrote:
> >On Wed, Aug 09, 2017 at 05:19:17PM -0500, Eddie James wrote:
> >>From: "Edward A. James" <eajames@...ibm.com>
> >>
> >>Export all the available status registers through debugfs, if the client
> >>driver wants them.
> >>
> >>Signed-off-by: Edward A. James <eajames@...ibm.com>
> >>---
> >> drivers/hwmon/pmbus/pmbus.c | 24 +++++-
> >> drivers/hwmon/pmbus/pmbus.h | 11 +++
> >> drivers/hwmon/pmbus/pmbus_core.c | 175 +++++++++++++++++++++++++++++++++++++++
> >> 3 files changed, 209 insertions(+), 1 deletion(-)
> >>
> >>diff --git a/drivers/hwmon/pmbus/pmbus.c b/drivers/hwmon/pmbus/pmbus.c
> >>index 7718e58..fc27417 100644
> >>--- a/drivers/hwmon/pmbus/pmbus.c
> >>+++ b/drivers/hwmon/pmbus/pmbus.c
> >That won't work: SENSORS_PMBUS is independent of PMBUS and does not have
> >to be enabled even if PMBUS is enabled. This will have to be in pmbus_core.c.
> >
> >I would suggest to do the same as done with other drivers. Move
> >pmbus_debugfs_dir into pmbus_core.c and create the directory on probe
> >if it has not already been created.
>
> Hmm, I'm having trouble with this. Don't I have to call
> debugfs_recursive_remove() on the pmbus_debugfs_dir when the pmbus driver is
> unloaded? That can't be done from pmbus_core.c, as that is just device
> probe/remove. I agree it makes sense to only create it if a device is going
> to be using it (and is enabled in the kernel) but I think I need to declare
> it here so I can remove it on unload.
>
So if CONFIG_SENSORS_PMBUS is not enabled, you won't have debugfs.
actually, even worse, this effectively mandates that CONFIG_SENSORS_PMBUS
must be enabled since it derclares a variable used by the core.
That doesn't make sense. What makes even less sense is that it defines the
variable in a module, thus _mandating_ that PMBUS_CORE is also built as module
if this driver is built as module. Which you can't even express in Kconfig
right now without adding a lot of complexity.
Guenter
> >
> >>@@ -18,6 +18,7 @@
> >> * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
> >> */
> >>+#include <linux/debugfs.h>
> >> #include <linux/kernel.h>
> >> #include <linux/module.h>
> >> #include <linux/init.h>
> >>@@ -230,7 +231,28 @@ static int pmbus_probe(struct i2c_client *client,
> >> .id_table = pmbus_id,
> >> };
> >>-module_i2c_driver(pmbus_driver);
> >>+struct dentry *pmbus_debugfs_dir;
> >>+EXPORT_SYMBOL_GPL(pmbus_debugfs_dir);
> >>+
> >>+static int __init pmbus_init(void)
> >>+{
> >>+ pmbus_debugfs_dir = debugfs_create_dir(pmbus_driver.driver.name, NULL);
> >>+ if (IS_ERR(pmbus_debugfs_dir))
> >>+ /* Don't fail out if we don't have debugfs support. */
> >>+ pmbus_debugfs_dir = NULL;
> >>+
> >>+ return i2c_add_driver(&pmbus_driver);
> >>+}
> >>+
> >>+static void __exit pmbus_exit(void)
> >>+{
> >>+ debugfs_remove_recursive(pmbus_debugfs_dir);
> >>+
> >>+ i2c_del_driver(&pmbus_driver);
> >>+}
> >>+
> >>+module_init(pmbus_init);
> >>+module_exit(pmbus_exit);
> >> MODULE_AUTHOR("Guenter Roeck");
> >> MODULE_DESCRIPTION("Generic PMBus driver");
> >>diff --git a/drivers/hwmon/pmbus/pmbus.h b/drivers/hwmon/pmbus/pmbus.h
> >>index bfcb13b..c772b83 100644
> >>--- a/drivers/hwmon/pmbus/pmbus.h
> >>+++ b/drivers/hwmon/pmbus/pmbus.h
> >>@@ -25,6 +25,8 @@
> >> #include <linux/bitops.h>
> >> #include <linux/regulator/driver.h>
> >>+struct dentry;
> >>+
> >> /*
> >> * Registers
> >> */
> >>@@ -383,6 +385,12 @@ struct pmbus_driver_info {
> >> /* Regulator functionality, if supported by this chip driver. */
> >> int num_regulators;
> >> const struct regulator_desc *reg_desc;
> >>+
> >>+ /*
> >>+ * Controls whether or not to create debugfs entries for this driver's
> >>+ * devices.
> >>+ */
> >>+ bool debugfs;
> >> };
> >> /* Regulator ops */
> >>@@ -401,6 +409,9 @@ struct pmbus_driver_info {
> >> .owner = THIS_MODULE, \
> >> }
> >>+/* Handle for debugfs directory */
> >>+extern struct dentry *pmbus_debugfs_dir;
> >>+
> >> /* Function declarations */
> >> void pmbus_clear_cache(struct i2c_client *client);
> >>diff --git a/drivers/hwmon/pmbus/pmbus_core.c b/drivers/hwmon/pmbus/pmbus_core.c
> >>index 1a51f8f..afbd364 100644
> >>--- a/drivers/hwmon/pmbus/pmbus_core.c
> >>+++ b/drivers/hwmon/pmbus/pmbus_core.c
> >>@@ -19,6 +19,7 @@
> >> * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
> >> */
> >>+#include <linux/debugfs.h>
> >> #include <linux/kernel.h>
> >> #include <linux/module.h>
> >> #include <linux/init.h>
> >>@@ -49,6 +50,9 @@
> >> #define PB_STATUS_FAN34_BASE (PB_STATUS_FAN_BASE + PMBUS_PAGES)
> >> #define PB_STATUS_TEMP_BASE (PB_STATUS_FAN34_BASE + PMBUS_PAGES)
> >> #define PB_STATUS_INPUT_BASE (PB_STATUS_TEMP_BASE + PMBUS_PAGES)
> >>+#define PB_STATUS_CML_BASE (PB_STATUS_INPUT_BASE + PMBUS_PAGES)
> >>+#define PB_STATUS_OTHER_BASE (PB_STATUS_CML_BASE + PMBUS_PAGES)
> >>+#define PB_STATUS_MFR_BASE (PB_STATUS_OTHER_BASE + PMBUS_PAGES)
> >We are not actively using those status registers, are we ?
> >I am a bit concerned that we'll continuously read those status registers
> >but don't do anything with it most of the time. Ultimately I'd prefer
> >to get rid of all caching, not more, since it gets quite expensive
> >on chips with many pages.
> >
> >Maybe just display uncached status register values in debugfs ?
>
> Sounds good.
>
> >
> >> #define PB_STATUS_VMON_BASE (PB_STATUS_INPUT_BASE + 1)
> >> #define PB_NUM_STATUS_REG (PB_STATUS_VMON_BASE + 1)
> >>@@ -101,6 +105,7 @@ struct pmbus_data {
> >> int num_attributes;
> >> struct attribute_group group;
> >> const struct attribute_group *groups[2];
> >>+ struct dentry *debugfs; /* debugfs device directory */
> >> struct pmbus_sensor *sensors;
> >>@@ -119,6 +124,11 @@ struct pmbus_data {
> >> u8 currpage;
> >> };
> >>+struct pmbus_debugfs_entry {
> >>+ struct device *dev;
> >>+ int index;
> >>+};
> >>+
> >> void pmbus_clear_cache(struct i2c_client *client)
> >> {
> >> struct pmbus_data *data = i2c_get_clientdata(client);
> >>@@ -422,6 +432,19 @@ static struct pmbus_data *pmbus_update_device(struct device *dev)
> >> = _pmbus_read_byte_data(client, i,
> >> s->reg);
> >> }
> >>+
> >>+ if (!data->debugfs)
> >>+ continue;
> >>+
> >>+ data->status[PB_STATUS_CML_BASE + i]
> >>+ = _pmbus_read_byte_data(client, i,
> >>+ PMBUS_STATUS_CML);
> >>+ data->status[PB_STATUS_OTHER_BASE + i]
> >>+ = _pmbus_read_byte_data(client, i,
> >>+ PMBUS_STATUS_OTHER);
> >>+ data->status[PB_STATUS_MFR_BASE + i]
> >>+ = _pmbus_read_byte_data(client, i,
> >>+ PMBUS_STATUS_MFR_SPECIFIC);
> >> }
> >> if (info->func[0] & PMBUS_HAVE_STATUS_INPUT)
> >>@@ -1891,6 +1914,151 @@ static int pmbus_regulator_register(struct pmbus_data *data)
> >> }
> >> #endif
> >>+static int pmbus_debugfs_get(void *data, u64 *val)
> >>+{
> >>+ struct pmbus_debugfs_entry *entry = data;
> >>+ struct pmbus_data *pdata = pmbus_update_device(entry->dev);
> >>+
> >>+ *val = pdata->status[entry->index];
> >>+
> >>+ return 0;
> >>+}
> >>+
> >>+DEFINE_DEBUGFS_ATTRIBUTE(pmbus_debugfs_ops8, pmbus_debugfs_get, NULL,
> >>+ "0x%02llx\n");
> >>+DEFINE_DEBUGFS_ATTRIBUTE(pmbus_debugfs_ops16, pmbus_debugfs_get, NULL,
> >>+ "0x%04llx\n");
> >>+
> >>+static int pmbus_init_debugfs(struct i2c_client *client,
> >>+ struct pmbus_data *data)
> >>+{
> >>+ int i, idx = 0;
> >>+ struct dentry *dbg;
> >>+ char name[PMBUS_NAME_SIZE];
> >>+ struct pmbus_debugfs_entry *entries;
> >>+
> >>+ /*
> >>+ * Either user didn't request debugfs or debugfs is not enabled in
> >>+ * kernel. Exit but don't throw an error in these cases.
> >>+ */
> >Here might be the place to initialize pmbus_debugfs_dir
> >if it is not yet initialized.
> >
> >>+ if (!data->info->debugfs || !pmbus_debugfs_dir)
> >>+ return 0;
> >>+
> >>+ /*
> >>+ * Create the debugfs directory for this device. Use the hwmon device
> >>+ * name to avoid conflicts (hwmon numbers are globally unique).
> >>+ */
> >>+ data->debugfs = debugfs_create_dir(dev_name(data->hwmon_dev),
> >>+ pmbus_debugfs_dir);
> >>+ if (IS_ERR_OR_NULL(data->debugfs)) {
> >>+ data->debugfs = NULL;
> >>+ return -ENODEV;
> >>+ }
> >>+
> >>+ /* Allocate the max possible entries we need. */
> >>+ entries = devm_kzalloc(data->dev,
> >>+ sizeof(*entries) * (data->info->pages * 10),
> >>+ GFP_KERNEL);
> >>+ if (!entries)
> >>+ return -ENOMEM;
> >>+
> >>+ for (i = 0; i < data->info->pages; ++i) {
> >>+ /* check accessibility of status register if it's not page 0 */
> >>+ if (!i || pmbus_check_status_register(client, i)) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops16);
> >What is the point of the dbg variable ?
> >
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_VOUT) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_VOUT_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_vout", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_IOUT) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_IOUT_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_iout", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_INPUT) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_INPUT_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_input", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_TEMP) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_TEMP_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_temp", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (pmbus_check_byte_register(client, i, PMBUS_STATUS_CML)) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_CML_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_cml", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (pmbus_check_byte_register(client, i, PMBUS_STATUS_OTHER)) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_OTHER_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_other", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (pmbus_check_byte_register(client, i,
> >>+ PMBUS_STATUS_MFR_SPECIFIC)) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_MFR_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_mfr", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_FAN12) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_FAN_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_fan12", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+
> >>+ if (data->info->func[i] & PMBUS_HAVE_STATUS_FAN34) {
> >>+ entries[idx].dev = data->hwmon_dev;
> >>+ entries[idx].index = PB_STATUS_FAN34_BASE + i;
> >>+ snprintf(name, PMBUS_NAME_SIZE, "status%d_fan34", i);
> >>+ dbg = debugfs_create_file(name, 0444, data->debugfs,
> >>+ &entries[idx++],
> >>+ &pmbus_debugfs_ops8);
> >>+ }
> >>+ }
> >>+
> >>+ return 0;
> >>+}
> >>+
> >Typically debugfs code is conditional. What happens if DEBUGFS
> >is not enabled ? See below.
> >
> >> int pmbus_do_probe(struct i2c_client *client, const struct i2c_device_id *id,
> >> struct pmbus_driver_info *info)
> >> {
> >>@@ -1950,6 +2118,10 @@ int pmbus_do_probe(struct i2c_client *client, const struct i2c_device_id *id,
> >> if (ret)
> >> goto out_unregister;
> >>+ ret = pmbus_init_debugfs(client, data);
> >>+ if (ret)
> >>+ dev_warn(dev, "Failed to register debugfs\n");
> >I think this warning will be generated if debugfs is disabled.
> >We should not do that. Consider putting the debugfs code into
> >#fidef and have a dummy pmbus_init_debugfs() returning 0 if
> >it is disabled.
>
> Yes, good idea.
>
> >
> >>+
> >> return 0;
> >> out_unregister:
> >>@@ -1963,6 +2135,9 @@ int pmbus_do_probe(struct i2c_client *client, const struct i2c_device_id *id,
> >> int pmbus_do_remove(struct i2c_client *client)
> >> {
> >> struct pmbus_data *data = i2c_get_clientdata(client);
> >>+
> >>+ debugfs_remove_recursive(data->debugfs);
> >>+
> >> hwmon_device_unregister(data->hwmon_dev);
> >> kfree(data->group.attrs);
> >> return 0;
> >>--
> >>1.8.3.1
> >>
> >>--
> >>To unsubscribe from this list: send the line "unsubscribe linux-hwmon" in
> >>the body of a message to majordomo@...r.kernel.org
> >>More majordomo info at http://vger.kernel.org/majordomo-info.html
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-hwmon" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists