driver core: fw_devlink: Clean up strings and mutex usages
cover driver core: fw_devlink: Clean up strings and mutex usages
Replace strlen() + kmalloc() + snprintf() by kasprintf() ond other string handling improvements.
Use guard() / scoped_guard() to handle fw_devlink mutex.
Andy Shevchenko (5):
driver core: Sort headers
driver core: Use kasprintf() instead of fixed buffer formatting driver core: Use guards for simple mutex locks
driver core: Make use of returned value of dev_err_probe() driver core: Use 2-argument strscpy()
drivers/base/core.c | 142 ++++++++++++++++++++------------------------ 1 file changed, 64 insertions(+), 78 deletions(-)
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
PATCH 1/5 driver core: Sort headers
Sort the headers in alphabetic order in order to ease the maintenance for this part.
--- drivers/base/core.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/base/core.c +7 -7
@@ -9,29 +9,29 @@*/ #include <linux/acpi.h> +#include <linux/blkdev.h> #include <linux/cpufreq.h> #include <linux/device.h> +#include <linux/dma-map-ops.h> /* for dma_default_coherent */ #include <linux/err.h> #include <linux/fwnode.h> #include <linux/init.h> +#include <linux/kdev_t.h> #include <linux/kstrtox.h> #include <linux/module.h> -#include <linux/slab.h> -#include <linux/kdev_t.h> +#include <linux/mutex.h> +#include <linux/netdevice.h> #include <linux/notifier.h> #include <linux/of.h> #include <linux/of_device.h> -#include <linux/blkdev.h> -#include <linux/mutex.h> #include <linux/pm_runtime.h> -#include <linux/netdevice.h> #include <linux/rcupdate.h> -#include <linux/sched/signal.h> #include <linux/sched/mm.h> +#include <linux/sched/signal.h> +#include <linux/slab.h> #include <linux/string_helpers.h> #include <linux/swiotlb.h> #include <linux/sysfs.h> -#include <linux/dma-map-ops.h> /* for dma_default_coherent */ #include "base.h" #include "physical_location.h"
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
Re: [PATCH v1 1/5] driver core: Sort headers
Quoted 39 lines · Andy Shevchenko
Sort the headers in alphabetic order in order to ease the maintenance for this part.
Signed-off-by Andy Shevchenko <[email protected]>--- drivers/base/core.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)drivers/base/core.c +5
-6@@ -9,29 +9,29 @@*/ #include <linux/acpi.h> +#include <linux/blkdev.h> #include <linux/cpufreq.h> #include <linux/device.h> +#include <linux/dma-map-ops.h> /* for dma_default_coherent */ #include <linux/err.h> #include <linux/fwnode.h> #include <linux/init.h> +#include <linux/kdev_t.h> #include <linux/kstrtox.h> #include <linux/module.h> -#include <linux/slab.h> -#include <linux/kdev_t.h> +#include <linux/mutex.h> +#include <linux/netdevice.h> #include <linux/notifier.h> #include <linux/of.h> #include <linux/of_device.h> -#include <linux/blkdev.h> -#include <linux/mutex.h> #include <linux/pm_runtime.h> -#include <linux/netdevice.h> #include <linux/rcupdate.h> -#include <linux/sched/signal.h>
drivers/base/core.c line 37
Quoted 10 lines
drivers/base/core.c +2
-1#include <linux/sched/mm.h> +#include <linux/sched/signal.h> +#include <linux/slab.h> #include <linux/string_helpers.h> #include <linux/swiotlb.h> #include <linux/sysfs.h> -#include <linux/dma-map-ops.h> /* for dma_default_coherent */ #include "base.h" #include "physical_location.h"
Hi Andy,
i don't think it is good idea to sort headers by alphabetic order.
why ?
1) header's dependency is not related to its file (name|path), their dependency are related to # includes order.
2) it maybe be easy to cause build error.
3) header's path or name maybe be related to subsystem, it is not good to sort one subsystem's headers before the other.
For header's order, my points is that:
1) sort by their dependency.
#include <b_header.h>
#include <a_header.h>
if
a_header.h:
#include <b_header.h>
2) all #include <> block before all #include "" block.
3) sort headers related to source file at the last.
prefix_xyz.c:
#include <>
.....
#include <prefix_xyz.h> // it is the last if it is exposed.
#include "internal_header.h" ....
4)
sort relevant header together as far as possible, for example, they belong to the same subsystem.
Re: [PATCH v1 1/5] driver core: Sort headers
Quoted 3 lines · quic_zijuhu
Quoted 2 lines · Andy Shevchenko
Sort the headers in alphabetic order in order to ease the maintenance for this part.
...
Quoted 1 line
i don't think it is good idea to sort headers by alphabetic order.
I strongly disagree on this on several points:
- the header dependencies has to be resolved on each header by applying IWYU (Include What You Use) principle:
in this case we don't care what is needed for each header in question
- the end developer shouldn't care about header dependencies changes as the project is evolving:
it's way out of human being capacity to follow _all_ the changes in the Linux kernel headers
- it's much easier to maintain the inclusion block when it's sorted (to avoid dups, or to see in a fast manner what's already included):
we are writing code for humans, and not for the machines (leave the optimisation task to the compiler in many cases)
- overall it makes the development process much easier as a whole:
I do not believe there is a single person in the world who may tell you the correct order of inclusion to any, even simple, Linux kernel driver
Quoted 4 lines
why ?
1) header's dependency is not related to its file (name|path), their dependency are related to # includes order.
That's not true. More precisely we are working hard to make it not true (and it's not a Plan 9 OS where as far as I know the idea was that developer knows the exact order).
Quoted 1 line
2) it maybe be easy to cause build error.
Yes, and again we are trying to avoid this by enforcing IWYU principle.
Quoted 2 lines
3) header's path or name maybe be related to subsystem, it is not good to sort one subsystem's headers before the other.
There is a grouping approach which makes this easier to get. See IIO subsystem as a prime example for IWYU implementation in the Linux kernel.
Quoted 3 lines
For header's order, my points is that:
1) sort by their dependency.
See above. No way, it's completely impractical.
Quoted 22 lines
#include <b_header.h> #include <a_header.h> if a_header.h: #include <b_header.h>2) all #include <> block before all #include "" block.
3) sort headers related to source file at the last.
prefix_xyz.c: #include <> ..... #include <prefix_xyz.h> // it is the last if it is exposed. #include "internal_header.h" ....4)
sort relevant header together as far as possible, for example, they belong to the same subsystem.
Signature · 2 lines
With Best Regards, Andy Shevchenko
PATCH 2/5 driver core: Use kasprintf() instead of fixed buffer formatting
In reply to Andy ShevchenkoImprove readability and maintainability by replacing a hardcoded string allocation and formatting by the use of the kasprintf() helper.
--- drivers/base/core.c | 70 +++++++++++++++++++++------------------------ 1 file changed, 32 insertions(+), 38 deletions(-)
drivers/base/core.c +32 -38
@@ -10,6 +10,7 @@#include <linux/acpi.h> #include <linux/blkdev.h> +#include <linux/cleanup.h> #include <linux/cpufreq.h> #include <linux/device.h> #include <linux/dma-map-ops.h> /* for dma_default_coherent */
@@ -563,24 +564,11 @@ static struct class devlink_class = {static int devlink_add_symlinks (struct device *dev) { + char *buf_con __free(kfree) = NULL, *buf_sup __free(kfree) = NULL;int ret; -size_t len;struct device_link *link =to_devlink (dev);struct device *sup = link->supplier;struct device *con = link->consumer; -char *buf; - - len =max (strlen (dev_bus_name (sup)) +strlen (dev_name (sup)), -strlen (dev_bus_name (con)) +strlen (dev_name (con))); - len +=strlen (":" ); -/* -* we kzalloc() memory for symlink name of both supplier and -* consumer, so explicitly take into account both prefix. -*/ - len +=max (strlen ("supplier:" ),strlen ("consumer:" )) +1 ; - buf =kzalloc (len,GFP_KERNEL ); -if (!buf) -return -ENOMEM ; ret =sysfs_create_link (&link->link_dev.kobj, &sup->kobj,"supplier" );if (ret)
@@ -590,58 +578,64 @@ static int devlink_add_symlinks(struct device *dev)if (ret) goto err_con; -snprintf (buf, len,"consumer:%s:%s" ,dev_bus_name (con),dev_name (con)); - ret =sysfs_create_link (&sup->kobj, &link->link_dev.kobj, buf); + buf_con =kasprintf (GFP_KERNEL ,"consumer:%s:%s" ,dev_bus_name (con),dev_name (con)); +if (!buf_con) { + ret = -ENOMEM ; + goto err_con_dev; + } + + ret =sysfs_create_link (&sup->kobj, &link->link_dev.kobj, buf_con);if (ret) goto err_con_dev; -snprintf (buf, len,"supplier:%s:%s" ,dev_bus_name (sup),dev_name (sup)); - ret =sysfs_create_link (&con->kobj, &link->link_dev.kobj, buf); + buf_sup =kasprintf (GFP_KERNEL ,"supplier:%s:%s" ,dev_bus_name (sup),dev_name (sup)); +if (!buf_sup) { + ret = -ENOMEM ; + goto err_sup_dev; + } + + ret =sysfs_create_link (&con->kobj, &link->link_dev.kobj, buf_sup);if (ret) goto err_sup_dev; goto out; err_sup_dev: -snprintf (buf, len,"consumer:%s:%s" ,dev_bus_name (con),dev_name (con)); -sysfs_remove_link (&sup->kobj, buf); +sysfs_remove_link (&sup->kobj, buf_con); err_con_dev:sysfs_remove_link (&link->link_dev.kobj,"consumer" ); err_con:sysfs_remove_link (&link->link_dev.kobj,"supplier" ); out: -kfree (buf);return ret; }static void devlink_remove_symlinks (struct device *dev) { + char *buf_con __free(kfree) = NULL, *buf_sup __free(kfree) = NULL;struct device_link *link =to_devlink (dev); -size_t len;struct device *sup = link->supplier;struct device *con = link->consumer; -char *buf;sysfs_remove_link (&link->link_dev.kobj,"consumer" );sysfs_remove_link (&link->link_dev.kobj,"supplier" ); - len =max (strlen (dev_bus_name (sup)) +strlen (dev_name (sup)), -strlen (dev_bus_name (con)) +strlen (dev_name (con))); - len +=strlen (":" ); - len +=max (strlen ("supplier:" ),strlen ("consumer:" )) +1 ; - buf =kzalloc (len,GFP_KERNEL ); -if (!buf) { -WARN (1 ,"Unable to properly free device link symlinks!\n" ); -return ; +if (device_is_registered (con)) { + buf_sup =kasprintf (GFP_KERNEL ,"supplier:%s:%s" ,dev_bus_name (sup),dev_name (sup)); +if (!buf_sup) + goto out; +sysfs_remove_link (&con->kobj, buf_sup); } -if (device_is_registered (con)) { -snprintf (buf, len,"supplier:%s:%s" ,dev_bus_name (sup),dev_name (sup)); -sysfs_remove_link (&con->kobj, buf); - } -snprintf (buf, len,"consumer:%s:%s" ,dev_bus_name (con),dev_name (con)); -sysfs_remove_link (&sup->kobj, buf); -kfree (buf); + buf_con =kasprintf (GFP_KERNEL ,"consumer:%s:%s" ,dev_bus_name (con),dev_name (con)); +if (!buf_con) + goto out; +sysfs_remove_link (&sup->kobj, buf_con); + +return ; + +out: +WARN (1 ,"Unable to properly free device link symlinks!\n" ); }static struct class_interface devlink_class_intf = {
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
PATCH 3/5 driver core: Use guards for simple mutex locks
In reply to Andy ShevchenkoGuards can help to make the code more readable. So use it wherever they do so.
--- drivers/base/core.c | 50 ++++++++++++++++++++------------------------- 1 file changed, 22 insertions(+), 28 deletions(-)
drivers/base/core.c +22 -28
@@ -98,12 +98,9 @@ static int __fwnode_link_add(struct fwnode_handle *con,int fwnode_link_add (struct fwnode_handle *con,struct fwnode_handle *sup,u8 flags) { -int ret; +guard (mutex)(&fwnode_link_lock); -mutex_lock (&fwnode_link_lock); - ret =__fwnode_link_add (con, sup, flags); -mutex_unlock (&fwnode_link_lock); -return ret; +return __fwnode_link_add (con, sup, flags); }/**
@@ -144,10 +141,10 @@ static void fwnode_links_purge_suppliers(struct fwnode_handle *fwnode) {struct fwnode_link *link, *tmp; -mutex_lock (&fwnode_link_lock); +guard (mutex)(&fwnode_link_lock); +list_for_each_entry_safe (link, tmp, &fwnode->suppliers, c_hook)__fwnode_link_del (link); -mutex_unlock (&fwnode_link_lock); }/**
@@ -160,10 +157,10 @@ static void fwnode_links_purge_consumers(struct fwnode_handle *fwnode) {struct fwnode_link *link, *tmp; -mutex_lock (&fwnode_link_lock); +guard (mutex)(&fwnode_link_lock); +list_for_each_entry_safe (link, tmp, &fwnode->consumers, s_hook)__fwnode_link_del (link); -mutex_unlock (&fwnode_link_lock); }/**
@@ -1059,20 +1056,16 @@ int device_links_check_suppliers(struct device *dev)* Device waiting for supplier to become available is not allowed to * probe. */ -mutex_lock (&fwnode_link_lock); - sup_fw =fwnode_links_check_suppliers (dev->fwnode); -if (sup_fw) { -if (!dev_is_best_effort (dev)) { - fwnode_ret = -EPROBE_DEFER ; -dev_err_probe (dev, -EPROBE_DEFER , -"wait for supplier %pfwf\n" , sup_fw); - }else { - fwnode_ret = -EAGAIN ; + scoped_guard(mutex, &fwnode_link_lock) { + sup_fw =fwnode_links_check_suppliers (dev->fwnode); +if (sup_fw) { +if (dev_is_best_effort (dev)) + fwnode_ret = -EAGAIN ; +else +return dev_err_probe (dev, -EPROBE_DEFER , +"wait for supplier %pfwf\n" , sup_fw); } } -mutex_unlock (&fwnode_link_lock); -if (fwnode_ret == -EPROBE_DEFER ) -return fwnode_ret;device_links_write_lock ();
@@ -1247,9 +1240,8 @@ static ssize_t waiting_for_supplier_show(struct device *dev,bool val;device_lock (dev); -mutex_lock (&fwnode_link_lock); - val = !!fwnode_links_check_suppliers (dev->fwnode); -mutex_unlock (&fwnode_link_lock); +scoped_guard (mutex, &fwnode_link_lock) + val = !!fwnode_links_check_suppliers(dev->fwnode);device_unlock (dev);return sysfs_emit (buf,"%u\n" , val); }
@@ -1322,13 +1314,15 @@ void device_links_driver_bound(struct device *dev)*/ if (dev->fwnode && dev->fwnode->dev == dev) {struct fwnode_handle *child; +fwnode_links_purge_suppliers (dev->fwnode); -mutex_lock (&fwnode_link_lock); + +guard (mutex)(&fwnode_link_lock); +fwnode_for_each_available_child_node (dev->fwnode, child)__fw_devlink_pickup_dangling_consumers (child, dev->fwnode);__fw_devlink_link_to_consumers (dev); -mutex_unlock (&fwnode_link_lock); }device_remove_file (dev, &dev_attr_waiting_for_supplier);
@@ -2337,10 +2331,10 @@ static void fw_devlink_link_device(struct device *dev)fw_devlink_parse_fwtree (fwnode); -mutex_lock (&fwnode_link_lock); +guard (mutex)(&fwnode_link_lock); +__fw_devlink_link_to_consumers (dev);__fw_devlink_link_to_suppliers (dev, fwnode); -mutex_unlock (&fwnode_link_lock); }/* Device links support end. */
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
PATCH 4/5 driver core: Make use of returned value of dev_err_probe()
In reply to Andy ShevchenkoInstead of assigning ret explicitly to the same value that is supplied to dev_err_probe(), make use of returned value of the latter.
--- drivers/base/core.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/base/core.c +2 -4
@@ -1084,10 +1084,8 @@ int device_links_check_suppliers(struct device *dev) }device_links_missing_supplier (dev); -dev_err_probe (dev, -EPROBE_DEFER , -"supplier %s not ready\n" , -dev_name (link->supplier)); - ret = -EPROBE_DEFER ; + ret =dev_err_probe (dev, -EPROBE_DEFER , +"supplier %s not ready\n" ,dev_name (link->supplier));break ; }WRITE_ONCE (link->status,DL_STATE_CONSUMER_PROBE );
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
PATCH 5/5 driver core: Use 2-argument strscpy()
In reply to Andy ShevchenkoFrom: Andy Shevchenko <[email protected]>
Use 2-argument strscpy(), which is not only shorter but also provides an additional check that destination buffer is an array.
--- drivers/base/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/base/core.c +1 -1
@@ -4867,7 +4867,7 @@ set_dev_info(const struct device *dev, struct dev_printk_info *dev_info)else return ; -strscpy (dev_info->subsystem, subsys,sizeof (dev_info->subsystem)); +strscpy (dev_info->subsystem, subsys);/* * Add device identifier DEVICE=:
Signature · 1 line
2.43.0.rc1.1336.g36b5255a03ac
Greg Kroah-Hartman
maintainerQuoted 4 lines · Andy Shevchenko
Replace
strlen()+kmalloc()+snprintf()bykasprintf()ond other string handling improvements.Use
guard()/scoped_guard()to handlefw_devlinkmutex.
Nice clean ups, thanks!