iio: accel: kxcjk-1013: Fix possible memory leak in probe and remove
PATCH 1/1 iio: accel: kxcjk-1013: Fix possible memory leak in probe and remove
When ACPI type is ACPI_SMO8500, the data->dready_trig will not be set, the memory allocated by iio_triggered_buffer_setup() will not be freed, and cause memory leak as follows:
unreferenced object 0xffff888009551400 (size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):
02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ .......
backtrace:
[<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360
[<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf]
[<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer]
[<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013]
[<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0
[<00000000d9f581a6>] really_probe+0x299/0xc30
[<00000000c6c16cde>] __driver_probe_device+0x357/0x500
[<00000000909852a1>] driver_probe_device+0x4e/0x140
[<000000008419ba53>] __device_attach_driver+0x257/0x340
[<00000000533bb466>] bus_for_each_drv+0x166/0x1e0
[<000000005bf45d75>] __device_attach+0x272/0x420
[<0000000075220311>] bus_probe_device+0x1eb/0x2a0
[<0000000015587e85>] device_add+0xbf0/0x1f90
[<0000000086901b9e>] i2c_new_client_device+0x622/0xb20
[<000000000865ca18>] new_device_store+0x1fa/0x420
[<0000000059a3d183>] dev_attr_store+0x58/0x80Fix it by remove data->dready_trig condition in probe and remove.
--- drivers/iio/accel/kxcjk-1013.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/iio/accel/kxcjk-1013.c +2 -3
@@ -1595,8 +1595,7 @@ static int kxcjk1013_probe(struct i2c_client *client,return 0 ; err_buffer_cleanup: -if (data->dready_trig) -iio_triggered_buffer_cleanup (indio_dev); +iio_triggered_buffer_cleanup (indio_dev); err_trigger_unregister:if (data->dready_trig)iio_trigger_unregister (data->dready_trig);
@@ -1618,8 +1617,8 @@ static int kxcjk1013_remove(struct i2c_client *client)pm_runtime_disable (&client->dev);pm_runtime_set_suspended (&client->dev); +iio_triggered_buffer_cleanup (indio_dev);if (data->dready_trig) { -iio_triggered_buffer_cleanup (indio_dev);iio_trigger_unregister (data->dready_trig);iio_trigger_unregister (data->motion_trig); }
Signature · 1 line
2.25.1
Hi,
Quoted 27 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:unreferenced object
0xffff888009551400(size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace: [<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013] [<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0 [<00000000d9f581a6>] really_probe+0x299/0xc30 [<00000000c6c16cde>] __driver_probe_device+0x357/0x500 [<00000000909852a1>] driver_probe_device+0x4e/0x140 [<000000008419ba53>] __device_attach_driver+0x257/0x340 [<00000000533bb466>] bus_for_each_drv+0x166/0x1e0 [<000000005bf45d75>] __device_attach+0x272/0x420 [<0000000075220311>] bus_probe_device+0x1eb/0x2a0 [<0000000015587e85>] device_add+0xbf0/0x1f90 [<0000000086901b9e>] i2c_new_client_device+0x622/0xb20 [<000000000865ca18>] new_device_store+0x1fa/0x420 [<0000000059a3d183>] dev_attr_store+0x58/0x80
Quoted 5 lines
Fix it by remove
data->dready_trigcondition in probe and remove.Reported-by Hulk Robot <[email protected]>Fixes a25691c1f967 ("iio: accel: kxcjk1013: allow using an external trigger")Signed-off-by Yang Yingliang <[email protected]>
Hmm, wouldn't the right fix be to also move theiio_triggered_buffer_setup() call to inside the:
if (client->irq > 0 && data->acpi_type != ACPI_SMO8500) {
}block ?
Jonathan (jic23) can you take a look at this, to me it seems that having a triggered buffer allocated without any triggers is not useful ?
Regards,
Hans
Quoted 29 lines
--- drivers/iio/accel/kxcjk-1013.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)drivers/iio/accel/kxcjk-1013.c +2
-3@@ -1595,8 +1595,7 @@ static int kxcjk1013_probe(struct i2c_client *client,return 0 ; err_buffer_cleanup: -if (data->dready_trig) -iio_triggered_buffer_cleanup (indio_dev); +iio_triggered_buffer_cleanup (indio_dev); err_trigger_unregister:if (data->dready_trig)iio_trigger_unregister (data->dready_trig);@@ -1618,8 +1617,8 @@ static int kxcjk1013_remove(struct i2c_client *client)pm_runtime_disable (&client->dev);pm_runtime_set_suspended (&client->dev); +iio_triggered_buffer_cleanup (indio_dev);if (data->dready_trig) { -iio_triggered_buffer_cleanup (indio_dev);iio_trigger_unregister (data->dready_trig);iio_trigger_unregister (data->motion_trig); }
Jonathan Cameron
maintainerQuoted 39 lines · On Thu, 14 Oct 2021 10:20:34 +0200 Hans de Goede <[email protected]> wrote:
Hi,
Quoted 27 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:unreferenced object
0xffff888009551400(size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace: [<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013] [<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0 [<00000000d9f581a6>] really_probe+0x299/0xc30 [<00000000c6c16cde>] __driver_probe_device+0x357/0x500 [<00000000909852a1>] driver_probe_device+0x4e/0x140 [<000000008419ba53>] __device_attach_driver+0x257/0x340 [<00000000533bb466>] bus_for_each_drv+0x166/0x1e0 [<000000005bf45d75>] __device_attach+0x272/0x420 [<0000000075220311>] bus_probe_device+0x1eb/0x2a0 [<0000000015587e85>] device_add+0xbf0/0x1f90 [<0000000086901b9e>] i2c_new_client_device+0x622/0xb20 [<000000000865ca18>] new_device_store+0x1fa/0x420 [<0000000059a3d183>] dev_attr_store+0x58/0x80Quoted 5 lines
Fix it by remove
data->dready_trigcondition in probe and remove.Reported-by Hulk Robot <[email protected]>Fixes a25691c1f967 ("iio: accel: kxcjk1013: allow using an external trigger")Signed-off-by Yang Yingliang <[email protected]>Hmm, wouldn't the right fix be to also move the
iio_triggered_buffer_setup()call to inside the:
Quoted 7 lines
if (client->irq > 0 && data->acpi_type != ACPI_SMO8500) { }block ?
Jonathan (jic23) can you take a look at this, to me it seems that having a triggered buffer allocated without any triggers is not useful ?
It can use another trigger not supplied by this particular device. e.g. sysfs or hrtimer trigger. This is common for cases where we may or may not have an irq wired and the validate_* callbacks are not provided (which would indicate we had to use the device's own trigger).
Jonathan
Quoted 37 lines
Regards,
Hans
Quoted 29 lines
--- drivers/iio/accel/kxcjk-1013.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)drivers/iio/accel/kxcjk-1013.c +2
-3@@ -1595,8 +1595,7 @@ static int kxcjk1013_probe(struct i2c_client *client,return 0 ; err_buffer_cleanup: -if (data->dready_trig) -iio_triggered_buffer_cleanup (indio_dev); +iio_triggered_buffer_cleanup (indio_dev); err_trigger_unregister:if (data->dready_trig)iio_trigger_unregister (data->dready_trig);@@ -1618,8 +1617,8 @@ static int kxcjk1013_remove(struct i2c_client *client)pm_runtime_disable (&client->dev);pm_runtime_set_suspended (&client->dev); +iio_triggered_buffer_cleanup (indio_dev);if (data->dready_trig) { -iio_triggered_buffer_cleanup (indio_dev);iio_trigger_unregister (data->dready_trig);iio_trigger_unregister (data->motion_trig); }
Hans de Goede
Reviewed-byHi,
Quoted 47 lines · Jonathan Cameron
Quoted 44 lines · On Thu, 14 Oct 2021 10:20:34 +0200 Hans de Goede <[email protected]> wrote:
Hi,
Quoted 27 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:unreferenced object
0xffff888009551400(size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace: [<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013] [<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0 [<00000000d9f581a6>] really_probe+0x299/0xc30 [<00000000c6c16cde>] __driver_probe_device+0x357/0x500 [<00000000909852a1>] driver_probe_device+0x4e/0x140 [<000000008419ba53>] __device_attach_driver+0x257/0x340 [<00000000533bb466>] bus_for_each_drv+0x166/0x1e0 [<000000005bf45d75>] __device_attach+0x272/0x420 [<0000000075220311>] bus_probe_device+0x1eb/0x2a0 [<0000000015587e85>] device_add+0xbf0/0x1f90 [<0000000086901b9e>] i2c_new_client_device+0x622/0xb20 [<000000000865ca18>] new_device_store+0x1fa/0x420 [<0000000059a3d183>] dev_attr_store+0x58/0x80Quoted 5 lines
Fix it by remove
data->dready_trigcondition in probe and remove.Reported-by Hulk Robot <[email protected]>Fixes a25691c1f967 ("iio: accel: kxcjk1013: allow using an external trigger")Signed-off-by Yang Yingliang <[email protected]>Hmm, wouldn't the right fix be to also move the
iio_triggered_buffer_setup()call to inside the:if (client->irq > 0 && data->acpi_type != ACPI_SMO8500) { }block ?
Quoted 7 lines
Quoted 2 lines
Jonathan (jic23) can you take a look at this, to me it seems that having a triggered buffer allocated without any triggers is not useful ?
It can use another trigger not supplied by this particular device. e.g. sysfs or hrtimer trigger. This is common for cases where we may or may not have an irq wired and the validate_* callbacks are not provided (which would indicate we had to use the device's own trigger).
Ok, thank you for clarifying this.
With that resolved, this patch looks good to me:
Regards,
Hans
Quoted 31 lines
Quoted 30 lines
Quoted 29 lines
--- drivers/iio/accel/kxcjk-1013.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)drivers/iio/accel/kxcjk-1013.c +2
-3@@ -1595,8 +1595,7 @@ static int kxcjk1013_probe(struct i2c_client *client,return 0 ; err_buffer_cleanup: -if (data->dready_trig) -iio_triggered_buffer_cleanup (indio_dev); +iio_triggered_buffer_cleanup (indio_dev); err_trigger_unregister:if (data->dready_trig)iio_trigger_unregister (data->dready_trig);@@ -1618,8 +1617,8 @@ static int kxcjk1013_remove(struct i2c_client *client)pm_runtime_disable (&client->dev);pm_runtime_set_suspended (&client->dev); +iio_triggered_buffer_cleanup (indio_dev);if (data->dready_trig) { -iio_triggered_buffer_cleanup (indio_dev);iio_trigger_unregister (data->dready_trig);iio_trigger_unregister (data->motion_trig); }
Quoted 3 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:
It seems it's not first time I'm telling you to shrink the noise in the commit message. Can you please LEARN this once and forever?
Quoted 22 lines
unreferenced object
0xffff888009551400(size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace: [<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013] [<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0 [<00000000d9f581a6>] really_probe+0x299/0xc30 [<00000000c6c16cde>] __driver_probe_device+0x357/0x500 [<00000000909852a1>] driver_probe_device+0x4e/0x140 [<000000008419ba53>] __device_attach_driver+0x257/0x340 [<00000000533bb466>] bus_for_each_drv+0x166/0x1e0 [<000000005bf45d75>] __device_attach+0x272/0x420 [<0000000075220311>] bus_probe_device+0x1eb/0x2a0 [<0000000015587e85>] device_add+0xbf0/0x1f90 [<0000000086901b9e>] i2c_new_client_device+0x622/0xb20 [<000000000865ca18>] new_device_store+0x1fa/0x420 [<0000000059a3d183>] dev_attr_store+0x58/0x80
Signature · 2 lines
With Best Regards, Andy Shevchenko
Hi,
Quoted 6 lines · Andy Shevchenko
Quoted 3 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:It seems it's not first time I'm telling you to shrink the noise in the commit message. Can you please LEARN this once and forever?
Thanks for you advice, I searched the whole kernel source tree commit message to learn how to make the memory leak report, I found most of them using the whole backtrace, so I make the report like they did in this patch. I will shrink the noise in v2. How about shrink it like this:
unreferenced object 0xffff888009551400 (size 512):
comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes):
02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace:
[<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013]
Thanks,
Yang
Quoted 23 lines
Quoted 23 lines
unreferenced object
0xffff888009551400(size 512):comm "i2c-SMO8500-125", pid 911, jiffies 4294911787 (age 83.852s) hex dump (first 32 bytes): 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 20 e2 e5 c0 ff ff ff ff ........ ....... backtrace: [<0000000041ce75ee>] kmem_cache_alloc_trace+0x16d/0x360 [<000000000aeb17b0>] iio_kfifo_allocate+0x41/0x130 [kfifo_buf] [<000000004b40c1f5>] iio_triggered_buffer_setup_ext+0x2c/0x210 [industrialio_triggered_buffer] [<000000004375b15f>] kxcjk1013_probe+0x10c3/0x1d81 [kxcjk_1013] [<0000000020115b9a>] i2c_device_probe+0xa31/0xbe0 [<00000000d9f581a6>] really_probe+0x299/0xc30 [<00000000c6c16cde>] __driver_probe_device+0x357/0x500 [<00000000909852a1>] driver_probe_device+0x4e/0x140 [<000000008419ba53>] __device_attach_driver+0x257/0x340 [<00000000533bb466>] bus_for_each_drv+0x166/0x1e0 [<000000005bf45d75>] __device_attach+0x272/0x420 [<0000000075220311>] bus_probe_device+0x1eb/0x2a0 [<0000000015587e85>] device_add+0xbf0/0x1f90 [<0000000086901b9e>] i2c_new_client_device+0x622/0xb20 [<000000000865ca18>] new_device_store+0x1fa/0x420 [<0000000059a3d183>] dev_attr_store+0x58/0x80
Quoted 5 lines · Yang Yingliang
Quoted 4 lines · Andy Shevchenko
Quoted 3 lines · Yang Yingliang
When ACPI type is
ACPI_SMO8500, thedata->dready_trigwill not be set, the memory allocated byiio_triggered_buffer_setup()will not be freed, and cause memory leak as follows:
Quoted 7 lines
Quoted 2 lines
It seems it's not first time I'm telling you to shrink the noise in the commit message. Can you please LEARN this once and forever?
Thanks for you advice, I searched the whole kernel source tree commit message to
learn how to make the memory leak report, I found most of them using the whole
backtrace, so I make the report like they did in this patch.
Some maintainers do not care about bloated sizes of the commit messages, however there are several advantages to have them shorter:
1/ saving resources (followed by disk storages and energy around the world,
means being environment friendly for real);
2/ when reading log, noise make it much harder to understand, besides the
fact that reading itself will be time consuming;
3/ nowadays some people are tending to read or even discuss the changes on
the mobile devices, where screen size is not so big;
4/ the copied'n'pasted backtrace means that the contributor probably
haven't thought through it and the quality of the change is in doubt.
Quoted 3 lines
I will shrink
the noise in v2.
How about shrink it like this:
Much better!
Signature · 2 lines
With Best Regards, Andy Shevchenko