* [driver-core:driver-core-testing 12/18] drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
@ 2026-05-23 9:46 ` Dan Carpenter
0 siblings, 0 replies; 4+ messages in thread
From: kernel test robot @ 2026-05-22 18:55 UTC (permalink / raw)
To: oe-kbuild; +Cc: lkp, Dan Carpenter
BCC: lkp@intel.com
CC: oe-kbuild-all@lists.linux.dev
CC: devel@driverdev.osuosl.org
TO: Saravana Kannan <saravanak@google.com>
CC: "Greg Kroah-Hartman" <gregkh@linuxfoundation.org>
CC: Herve Codina <herve.codina@bootlin.com>
tree: https://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core.git driver-core-testing
head: 024480bf8d75bd16894c5b0eb6082b6e6dae4970
commit: 81e7c6befa36cecdcbf7244393bd67e8f8c59bf5 [12/18] of: dynamic: Fix overlayed devices not probing because of fw_devlink
:::::: branch date: 7 hours ago
:::::: commit date: 7 hours ago
config: arm-randconfig-r071-20260523 (https://download.01.org/0day-ci/archive/20260523/202605230211.567ddMEp-lkp@intel.com/config)
compiler: clang version 23.0.0git (https://github.com/llvm/llvm-project 5bac06718f502014fade905512f1d26d578a18f3)
smatch: v0.5.0-9185-gbcc58b9c
If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Reported-by: Dan Carpenter <error27@gmail.com>
| Closes: https://lore.kernel.org/r/202605230211.567ddMEp-lkp@intel.com/
New smatch warnings:
drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
Old smatch warnings:
drivers/base/core.c:2551 device_store_int() warn: impossible condition '(new > ((~0 >> 1))) => (s32min-s32max > s32max)'
drivers/base/core.c:2551 device_store_int() warn: impossible condition '(new < (-((~0 >> 1)) - 1)) => (s32min-s32max < s32min)'
drivers/base/core.c:3297 class_dir_create_and_add() warn: Calling kobject_put|get with state->initialized unset from line: 3286
vim +/dev +286 drivers/base/core.c
81e7c6befa36ce Saravana Kannan 2026-05-11 248
81e7c6befa36ce Saravana Kannan 2026-05-11 249 /**
81e7c6befa36ce Saravana Kannan 2026-05-11 250 * fw_devlink_refresh_fwnode - Recheck the tree under this firmware node
81e7c6befa36ce Saravana Kannan 2026-05-11 251 * @fwnode: The fwnode under which the fwnode tree has changed
81e7c6befa36ce Saravana Kannan 2026-05-11 252 *
81e7c6befa36ce Saravana Kannan 2026-05-11 253 * This function is mainly meant to adjust the supplier/consumer dependencies
81e7c6befa36ce Saravana Kannan 2026-05-11 254 * after a fwnode tree overlay has occurred.
81e7c6befa36ce Saravana Kannan 2026-05-11 255 */
81e7c6befa36ce Saravana Kannan 2026-05-11 256 void fw_devlink_refresh_fwnode(struct fwnode_handle *fwnode)
81e7c6befa36ce Saravana Kannan 2026-05-11 257 {
81e7c6befa36ce Saravana Kannan 2026-05-11 258 struct device *dev;
81e7c6befa36ce Saravana Kannan 2026-05-11 259
81e7c6befa36ce Saravana Kannan 2026-05-11 260 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 261 * Find the closest ancestor fwnode that has been converted to a device
81e7c6befa36ce Saravana Kannan 2026-05-11 262 * that can bind to a driver (bus device).
81e7c6befa36ce Saravana Kannan 2026-05-11 263 */
81e7c6befa36ce Saravana Kannan 2026-05-11 264 fwnode_handle_get(fwnode);
81e7c6befa36ce Saravana Kannan 2026-05-11 265 do {
81e7c6befa36ce Saravana Kannan 2026-05-11 266 if (fwnode_test_flag(fwnode, FWNODE_FLAG_NOT_DEVICE))
81e7c6befa36ce Saravana Kannan 2026-05-11 267 continue;
81e7c6befa36ce Saravana Kannan 2026-05-11 268
81e7c6befa36ce Saravana Kannan 2026-05-11 269 dev = get_dev_from_fwnode(fwnode);
81e7c6befa36ce Saravana Kannan 2026-05-11 @270 if (!dev)
81e7c6befa36ce Saravana Kannan 2026-05-11 271 continue;
81e7c6befa36ce Saravana Kannan 2026-05-11 272
81e7c6befa36ce Saravana Kannan 2026-05-11 273 if (dev->bus)
81e7c6befa36ce Saravana Kannan 2026-05-11 274 break;
81e7c6befa36ce Saravana Kannan 2026-05-11 275
81e7c6befa36ce Saravana Kannan 2026-05-11 276 put_device(dev);
81e7c6befa36ce Saravana Kannan 2026-05-11 277 } while ((fwnode = fwnode_get_next_parent(fwnode)));
81e7c6befa36ce Saravana Kannan 2026-05-11 278
81e7c6befa36ce Saravana Kannan 2026-05-11 279 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 280 * If none of the ancestor fwnodes have (yet) been converted to a device
81e7c6befa36ce Saravana Kannan 2026-05-11 281 * that can bind to a driver, there's nothing to fix up.
81e7c6befa36ce Saravana Kannan 2026-05-11 282 */
81e7c6befa36ce Saravana Kannan 2026-05-11 283 if (!fwnode)
81e7c6befa36ce Saravana Kannan 2026-05-11 284 return;
81e7c6befa36ce Saravana Kannan 2026-05-11 285
81e7c6befa36ce Saravana Kannan 2026-05-11 @286 WARN(device_is_bound(dev) && dev->links.status != DL_DEV_DRIVER_BOUND,
81e7c6befa36ce Saravana Kannan 2026-05-11 287 "Don't multithread overlaying and probing the same device!\n");
81e7c6befa36ce Saravana Kannan 2026-05-11 288
81e7c6befa36ce Saravana Kannan 2026-05-11 289 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 290 * If the device has already bound to a driver, then we need to redo
81e7c6befa36ce Saravana Kannan 2026-05-11 291 * some of the work that was done after the device was bound to a
81e7c6befa36ce Saravana Kannan 2026-05-11 292 * driver. If the device hasn't bound to a driver, running things too
81e7c6befa36ce Saravana Kannan 2026-05-11 293 * soon would incorrectly pick up consumers that it shouldn't.
81e7c6befa36ce Saravana Kannan 2026-05-11 294 */
81e7c6befa36ce Saravana Kannan 2026-05-11 295 if (dev->links.status == DL_DEV_DRIVER_BOUND) {
81e7c6befa36ce Saravana Kannan 2026-05-11 296 fw_devlink_pickup_dangling_consumers(dev);
81e7c6befa36ce Saravana Kannan 2026-05-11 297 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 298 * Some of dangling consumers could have been put previously in
81e7c6befa36ce Saravana Kannan 2026-05-11 299 * the deferred probe list due to the unavailability of their
81e7c6befa36ce Saravana Kannan 2026-05-11 300 * suppliers. Those consumers have been picked up and some of
81e7c6befa36ce Saravana Kannan 2026-05-11 301 * their suppliers links have been updated. Time to re-try their
81e7c6befa36ce Saravana Kannan 2026-05-11 302 * probe sequence.
81e7c6befa36ce Saravana Kannan 2026-05-11 303 */
81e7c6befa36ce Saravana Kannan 2026-05-11 304 driver_deferred_probe_trigger();
81e7c6befa36ce Saravana Kannan 2026-05-11 305 }
81e7c6befa36ce Saravana Kannan 2026-05-11 306
81e7c6befa36ce Saravana Kannan 2026-05-11 307 put_device(dev);
81e7c6befa36ce Saravana Kannan 2026-05-11 308 fwnode_handle_put(fwnode);
81e7c6befa36ce Saravana Kannan 2026-05-11 309 }
81e7c6befa36ce Saravana Kannan 2026-05-11 310
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
^ permalink raw reply [flat|nested] 4+ messages in thread
* [driver-core:driver-core-testing 12/18] drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
@ 2026-05-23 9:46 ` Dan Carpenter
0 siblings, 0 replies; 4+ messages in thread
From: Dan Carpenter @ 2026-05-23 9:46 UTC (permalink / raw)
To: oe-kbuild, Saravana Kannan
Cc: lkp, oe-kbuild-all, devel, Greg Kroah-Hartman, Herve Codina
tree: https://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core.git driver-core-testing
head: 024480bf8d75bd16894c5b0eb6082b6e6dae4970
commit: 81e7c6befa36cecdcbf7244393bd67e8f8c59bf5 [12/18] of: dynamic: Fix overlayed devices not probing because of fw_devlink
config: arm-randconfig-r071-20260523 (https://download.01.org/0day-ci/archive/20260523/202605230211.567ddMEp-lkp@intel.com/config)
compiler: clang version 23.0.0git (https://github.com/llvm/llvm-project 5bac06718f502014fade905512f1d26d578a18f3)
smatch: v0.5.0-9185-gbcc58b9c
If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Reported-by: Dan Carpenter <error27@gmail.com>
| Closes: https://lore.kernel.org/r/202605230211.567ddMEp-lkp@intel.com/
New smatch warnings:
drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
vim +/dev +286 drivers/base/core.c
81e7c6befa36ce Saravana Kannan 2026-05-11 256 void fw_devlink_refresh_fwnode(struct fwnode_handle *fwnode)
81e7c6befa36ce Saravana Kannan 2026-05-11 257 {
81e7c6befa36ce Saravana Kannan 2026-05-11 258 struct device *dev;
81e7c6befa36ce Saravana Kannan 2026-05-11 259
81e7c6befa36ce Saravana Kannan 2026-05-11 260 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 261 * Find the closest ancestor fwnode that has been converted to a device
81e7c6befa36ce Saravana Kannan 2026-05-11 262 * that can bind to a driver (bus device).
81e7c6befa36ce Saravana Kannan 2026-05-11 263 */
81e7c6befa36ce Saravana Kannan 2026-05-11 264 fwnode_handle_get(fwnode);
81e7c6befa36ce Saravana Kannan 2026-05-11 265 do {
81e7c6befa36ce Saravana Kannan 2026-05-11 266 if (fwnode_test_flag(fwnode, FWNODE_FLAG_NOT_DEVICE))
81e7c6befa36ce Saravana Kannan 2026-05-11 267 continue;
81e7c6befa36ce Saravana Kannan 2026-05-11 268
81e7c6befa36ce Saravana Kannan 2026-05-11 269 dev = get_dev_from_fwnode(fwnode);
81e7c6befa36ce Saravana Kannan 2026-05-11 @270 if (!dev)
81e7c6befa36ce Saravana Kannan 2026-05-11 271 continue;
Smatch doesn't like how this code tests for NULL and but then later
we dereference "dev" without testing.
get_dev_from_fwnode() can only return NULL if "fwnode" is NULL which
it can't be. We can just remove this test.
81e7c6befa36ce Saravana Kannan 2026-05-11 272
81e7c6befa36ce Saravana Kannan 2026-05-11 273 if (dev->bus)
81e7c6befa36ce Saravana Kannan 2026-05-11 274 break;
81e7c6befa36ce Saravana Kannan 2026-05-11 275
81e7c6befa36ce Saravana Kannan 2026-05-11 276 put_device(dev);
81e7c6befa36ce Saravana Kannan 2026-05-11 277 } while ((fwnode = fwnode_get_next_parent(fwnode)));
81e7c6befa36ce Saravana Kannan 2026-05-11 278
81e7c6befa36ce Saravana Kannan 2026-05-11 279 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 280 * If none of the ancestor fwnodes have (yet) been converted to a device
81e7c6befa36ce Saravana Kannan 2026-05-11 281 * that can bind to a driver, there's nothing to fix up.
81e7c6befa36ce Saravana Kannan 2026-05-11 282 */
81e7c6befa36ce Saravana Kannan 2026-05-11 283 if (!fwnode)
81e7c6befa36ce Saravana Kannan 2026-05-11 284 return;
81e7c6befa36ce Saravana Kannan 2026-05-11 285
81e7c6befa36ce Saravana Kannan 2026-05-11 @286 WARN(device_is_bound(dev) && dev->links.status != DL_DEV_DRIVER_BOUND,
^^^^^^^^^^^^^^^^^
81e7c6befa36ce Saravana Kannan 2026-05-11 287 "Don't multithread overlaying and probing the same device!\n");
81e7c6befa36ce Saravana Kannan 2026-05-11 288
81e7c6befa36ce Saravana Kannan 2026-05-11 289 /*
81e7c6befa36ce Saravana Kannan 2026-05-11 290 * If the device has already bound to a driver, then we need to redo
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [driver-core:driver-core-testing 12/18] drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
2026-05-23 9:46 ` Dan Carpenter
(?)
@ 2026-05-23 11:01 ` Herve Codina
2026-05-25 6:35 ` Dan Carpenter
-1 siblings, 1 reply; 4+ messages in thread
From: Herve Codina @ 2026-05-23 11:01 UTC (permalink / raw)
To: Dan Carpenter
Cc: oe-kbuild, Saravana Kannan, lkp, oe-kbuild-all, devel,
Greg Kroah-Hartman
Hi Dan,
On Sat, 23 May 2026 12:46:14 +0300
Dan Carpenter <error27@gmail.com> wrote:
> tree: https://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core.git driver-core-testing
> head: 024480bf8d75bd16894c5b0eb6082b6e6dae4970
> commit: 81e7c6befa36cecdcbf7244393bd67e8f8c59bf5 [12/18] of: dynamic: Fix overlayed devices not probing because of fw_devlink
> config: arm-randconfig-r071-20260523 (https://download.01.org/0day-ci/archive/20260523/202605230211.567ddMEp-lkp@intel.com/config)
> compiler: clang version 23.0.0git (https://github.com/llvm/llvm-project 5bac06718f502014fade905512f1d26d578a18f3)
> smatch: v0.5.0-9185-gbcc58b9c
>
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@intel.com>
> | Reported-by: Dan Carpenter <error27@gmail.com>
> | Closes: https://lore.kernel.org/r/202605230211.567ddMEp-lkp@intel.com/
>
> New smatch warnings:
> drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
>
> vim +/dev +286 drivers/base/core.c
>
> 81e7c6befa36ce Saravana Kannan 2026-05-11 256 void fw_devlink_refresh_fwnode(struct fwnode_handle *fwnode)
> 81e7c6befa36ce Saravana Kannan 2026-05-11 257 {
> 81e7c6befa36ce Saravana Kannan 2026-05-11 258 struct device *dev;
> 81e7c6befa36ce Saravana Kannan 2026-05-11 259
> 81e7c6befa36ce Saravana Kannan 2026-05-11 260 /*
> 81e7c6befa36ce Saravana Kannan 2026-05-11 261 * Find the closest ancestor fwnode that has been converted to a device
> 81e7c6befa36ce Saravana Kannan 2026-05-11 262 * that can bind to a driver (bus device).
> 81e7c6befa36ce Saravana Kannan 2026-05-11 263 */
> 81e7c6befa36ce Saravana Kannan 2026-05-11 264 fwnode_handle_get(fwnode);
> 81e7c6befa36ce Saravana Kannan 2026-05-11 265 do {
> 81e7c6befa36ce Saravana Kannan 2026-05-11 266 if (fwnode_test_flag(fwnode, FWNODE_FLAG_NOT_DEVICE))
> 81e7c6befa36ce Saravana Kannan 2026-05-11 267 continue;
> 81e7c6befa36ce Saravana Kannan 2026-05-11 268
> 81e7c6befa36ce Saravana Kannan 2026-05-11 269 dev = get_dev_from_fwnode(fwnode);
> 81e7c6befa36ce Saravana Kannan 2026-05-11 @270 if (!dev)
> 81e7c6befa36ce Saravana Kannan 2026-05-11 271 continue;
>
> Smatch doesn't like how this code tests for NULL and but then later
> we dereference "dev" without testing.
>
> get_dev_from_fwnode() can only return NULL if "fwnode" is NULL which
> it can't be. We can just remove this test.
No so sure.
--- 8< ---
struct device *get_dev_from_fwnode(struct fwnode_handle *fwnode)
{
return get_device((fwnode)->dev);
}
EXPORT_SYMBOL_GPL(get_dev_from_fwnode);
--- 8< ---
get_dev_from_fwnode() returns NULL if no device is attached to the
given fwnode. This is what we test here.
At line 286, when dev is derefenced (dev->links.status), it cannot be NULL.
>
> 81e7c6befa36ce Saravana Kannan 2026-05-11 272
> 81e7c6befa36ce Saravana Kannan 2026-05-11 273 if (dev->bus)
> 81e7c6befa36ce Saravana Kannan 2026-05-11 274 break;
> 81e7c6befa36ce Saravana Kannan 2026-05-11 275
> 81e7c6befa36ce Saravana Kannan 2026-05-11 276 put_device(dev);
> 81e7c6befa36ce Saravana Kannan 2026-05-11 277 } while ((fwnode = fwnode_get_next_parent(fwnode)));
We go out of the while() loop here either :
- because dev->bus is not NULL (the break in the loop) and so dev is not
NULL in that case (tested just before)
or
- because fwnode is NULL
> 81e7c6befa36ce Saravana Kannan 2026-05-11 278
> 81e7c6befa36ce Saravana Kannan 2026-05-11 279 /*
> 81e7c6befa36ce Saravana Kannan 2026-05-11 280 * If none of the ancestor fwnodes have (yet) been converted to a device
> 81e7c6befa36ce Saravana Kannan 2026-05-11 281 * that can bind to a driver, there's nothing to fix up.
> 81e7c6befa36ce Saravana Kannan 2026-05-11 282 */
> 81e7c6befa36ce Saravana Kannan 2026-05-11 283 if (!fwnode)
> 81e7c6befa36ce Saravana Kannan 2026-05-11 284 return;
The case fwnode == NULL is handled here.
At this point dev cannot be NULL. Indeed, the loop has been exited because
dev->bus was not NULL (the break).
> 81e7c6befa36ce Saravana Kannan 2026-05-11 285
> 81e7c6befa36ce Saravana Kannan 2026-05-11 @286 WARN(device_is_bound(dev) && dev->links.status != DL_DEV_DRIVER_BOUND,
> ^^^^^^^^^^^^^^^^^
>
> 81e7c6befa36ce Saravana Kannan 2026-05-11 287 "Don't multithread overlaying and probing the same device!\n");
> 81e7c6befa36ce Saravana Kannan 2026-05-11 288
> 81e7c6befa36ce Saravana Kannan 2026-05-11 289 /*
> 81e7c6befa36ce Saravana Kannan 2026-05-11 290 * If the device has already bound to a driver, then we need to redo
>
> --
> 0-DAY CI Kernel Test Service
> https://github.com/intel/lkp-tests/wiki
>
I don't understand why smatch reported a problem.
Do I miss something?
Best regards,
Hervé
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [driver-core:driver-core-testing 12/18] drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270)
2026-05-23 11:01 ` Herve Codina
@ 2026-05-25 6:35 ` Dan Carpenter
0 siblings, 0 replies; 4+ messages in thread
From: Dan Carpenter @ 2026-05-25 6:35 UTC (permalink / raw)
To: Herve Codina
Cc: oe-kbuild, Saravana Kannan, lkp, oe-kbuild-all, devel,
Greg Kroah-Hartman
On Sat, May 23, 2026 at 01:01:42PM +0200, Herve Codina wrote:
>
> > --
> > 0-DAY CI Kernel Test Service
> > https://github.com/intel/lkp-tests/wiki
> >
>
> I don't understand why smatch reported a problem.
> Do I miss something?
Thanks for looking at this. Your analysis is correct. The problem
is that the zero day bot doesn't do cross function analysis so it can't
parse this code correctly. (Cross function analysis can't scale well
enough to enable it).
regards,
dan carpenter
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-05-25 6:35 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-05-22 18:55 [driver-core:driver-core-testing 12/18] drivers/base/core.c:286 fw_devlink_refresh_fwnode() error: we previously assumed 'dev' could be null (see line 270) kernel test robot
2026-05-23 9:46 ` Dan Carpenter
2026-05-23 11:01 ` Herve Codina
2026-05-25 6:35 ` Dan Carpenter
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.