soc: fsl: dpio: Use common error handling code in two functions
PATCH 1/1 soc: fsl: dpio: Use common error handling code in two functions
From: Markus Elfring <[email protected]>
Date: Thu, 11 Jun 2026 11:56:18 +0200
Use additional labels so that a bit of exception handling can be better reused at the end of two function implementations.
This issue was detected by using the Coccinelle software.
--- drivers/soc/fsl/dpio/dpio-service.c | 30 ++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-)
drivers/soc/fsl/dpio/dpio-service.c +15 -15
@@ -140,10 +140,8 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc,return NULL ;/* check if CPU is out of range (-1 means any cpu) */ -if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ()) { -kfree (obj); -return NULL ; - } +if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ()) + goto free_obj; obj->dpio_desc = *desc; obj->swp_desc.cena_bar = obj->dpio_desc.regs_cena;
@@ -158,11 +156,8 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc, qman_256_cycles_per_ns =256000 / (obj->swp_desc.qman_clk /1000000 ); obj->swp_desc.qman_256_cycles_per_ns = qman_256_cycles_per_ns; obj->swp =qbman_swp_init (&obj->swp_desc); - -if (!obj->swp) { -kfree (obj); -return NULL ; - } +if (!obj->swp) + goto free_obj;INIT_LIST_HEAD (&obj->node);spin_lock_init (&obj->lock_mgmt_cmd);
@@ -192,6 +187,10 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc, obj->frames =0 ;return obj; + +free_obj: +kfree (obj); +return NULL ; }/**
@@ -665,10 +664,8 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, ret->max = max_frames; size = max_frames *sizeof (struct dpaa2_dq ) +64 ; ret->alloced_addr =kzalloc (size,GFP_KERNEL ); -if (!ret->alloced_addr) { -kfree (ret); -return NULL ; - } +if (!ret->alloced_addr) + goto free_ret; ret->vaddr =PTR_ALIGN (ret->alloced_addr,64 ); ret->paddr = dma_map_single(dev, ret->vaddr,
@@ -676,14 +673,17 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, DMA_FROM_DEVICE);if (dma_mapping_error (dev, ret->paddr)) {kfree (ret->alloced_addr); -kfree (ret); -return NULL ; + goto free_ret; } ret->idx =0 ; ret->dev = dev;return ret; + +free_ret: +kfree (ret); +return NULL ; }EXPORT_SYMBOL_GPL (dpaa2_io_store_create);
Signature · 1 line
2.54.0
Christophe Leroy (CS GROUP)
maintainerLe 11/06/2026 à 12:02, Markus Elfring a écrit :
Quoted 7 lines
From: Markus Elfring <[email protected]>
Date: Thu, 11 Jun 2026 11:56:18 +0200Use additional labels so that a bit of exception handling can be better reused at the end of two function implementations.
This issue was detected by using the Coccinelle software.
I think it would be better to declare obj and ret as automatic free instead:
struct dpaa2_io __free(kfree) *obj = kmalloc_obj(*obj);
Quoted 81 lines
Signed-off-by Markus Elfring <[email protected]>--- drivers/soc/fsl/dpio/dpio-service.c | 30 ++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-)drivers/soc/fsl/dpio/dpio-service.c +15
-15@@ -140,10 +140,8 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc,return NULL ;/* check if CPU is out of range (-1 means any cpu) */ -if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ()) { -kfree (obj); -return NULL ; - } +if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ()) + goto free_obj; obj->dpio_desc = *desc; obj->swp_desc.cena_bar = obj->dpio_desc.regs_cena;@@ -158,11 +156,8 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc, qman_256_cycles_per_ns =256000 / (obj->swp_desc.qman_clk /1000000 ); obj->swp_desc.qman_256_cycles_per_ns = qman_256_cycles_per_ns; obj->swp =qbman_swp_init (&obj->swp_desc); - -if (!obj->swp) { -kfree (obj); -return NULL ; - } +if (!obj->swp) + goto free_obj;INIT_LIST_HEAD (&obj->node);spin_lock_init (&obj->lock_mgmt_cmd);@@ -192,6 +187,10 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc, obj->frames =0 ;return obj; + +free_obj: +kfree (obj); +return NULL ; }/** @@ -665,10 +664,8 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, ret->max = max_frames; size = max_frames *sizeof (struct dpaa2_dq ) +64 ; ret->alloced_addr =kzalloc (size,GFP_KERNEL ); -if (!ret->alloced_addr) { -kfree (ret); -return NULL ; - } +if (!ret->alloced_addr) + goto free_ret; ret->vaddr =PTR_ALIGN (ret->alloced_addr,64 ); ret->paddr = dma_map_single(dev, ret->vaddr,@@ -676,14 +673,17 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, DMA_FROM_DEVICE);if (dma_mapping_error (dev, ret->paddr)) {kfree (ret->alloced_addr); -kfree (ret); -return NULL ; + goto free_ret; } ret->idx =0 ; ret->dev = dev;return ret; + +free_ret: +kfree (ret); +return NULL ; }EXPORT_SYMBOL_GPL (dpaa2_io_store_create);
Quoted 6 lines
Quoted 4 lines
Use additional labels so that a bit of exception handling can be better reused at the end of two function implementations.
This issue was detected by using the Coccinelle software.
I think it would be better to declare obj and ret as automatic free instead:
Please reconsider such information in more detail.
Quoted 1 line
struct dpaa2_io __free(kfree) *obj = kmalloc_obj(*obj);
The application of scope-base resource management can eventually be increased further.
Regards,
Markus
v2 cover soc: fsl: dpio: Use scope-based resource management in two functions
In reply to Christophe Leroy (CS GROUP)From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 11:42:10 +0200
Some adjustment opportunities were picked up.
Markus Elfring (2):
Use scope-based resource management in dpaa2_io_store_create() Use scope-based resource management in dpaa2_io_create()
v2:
Christophe Leroy requested to apply the attribute “__free(kfree)”.
drivers/soc/fsl/dpio/dpio-service.c | 37 ++++++++++------------------- 1 file changed, 12 insertions(+), 25 deletions(-)
Signature · 1 line
2.54.0
PATCH v2 1/2 soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 10:34:43 +0200
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit 54da6a0924311c7cf5015533991e44fb8eb12773 ("locking:
Introduce __cleanup() based infrastructure").
* Thus use the attribute “__free(kfree)”.
* Reduce the scope for the local variable “ret”.
* Omit two kfree() calls accordingly.
* Omit the local variable “size” (for another memory allocation).
* Use the macro call “return_ptr(ret)” at the end.
--- drivers/soc/fsl/dpio/dpio-service.c | 15 ++++----------- 1 file changed, 4 insertions(+), 11 deletions(-)
drivers/soc/fsl/dpio/dpio-service.c +4 -11
@@ -652,23 +652,17 @@ EXPORT_SYMBOL_GPL(dpaa2_io_service_acquire);struct dpaa2_io_store *dpaa2_io_store_create (unsigned int max_frames,struct device *dev) { -struct dpaa2_io_store *ret; -size_t size; -if (!max_frames || (max_frames >32 ))return NULL ; - ret =kmalloc_obj (*ret); + struct dpaa2_io_store *ret __free(kfree) = kmalloc_obj(*ret);if (!ret)return NULL ; ret->max = max_frames; - size = max_frames *sizeof (struct dpaa2_dq ) +64 ; - ret->alloced_addr =kzalloc (size,GFP_KERNEL ); -if (!ret->alloced_addr) { -kfree (ret); + ret->alloced_addr =kzalloc (max_frames *sizeof (struct dpaa2_dq ) +64 ,GFP_KERNEL ); +if (!ret->alloced_addr)return NULL ; - } ret->vaddr =PTR_ALIGN (ret->alloced_addr,64 ); ret->paddr = dma_map_single(dev, ret->vaddr,
@@ -676,14 +670,13 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, DMA_FROM_DEVICE);if (dma_mapping_error (dev, ret->paddr)) {kfree (ret->alloced_addr); -kfree (ret);return NULL ; } ret->idx =0 ; ret->dev = dev; -return ret; +return_ptr (ret); }EXPORT_SYMBOL_GPL (dpaa2_io_store_create);
Signature · 1 line
2.54.0
Re: [PATCH v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 8 lines · Markus Elfring
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 10:34:43 +0200Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").
Quoted 7 lines
* Thus use the attribute “__free(kfree)”.
* Reduce the scope for the local variable “ret”.
* Omit two
kfree()calls accordingly.* Omit the local variable “size” (for another memory allocation).
Why? This seems unrelated...
drivers/soc/fsl/dpio/dpio-service.c EXPORT_SYMBOL_GPL(dpaa2_io_service_acquire);
Quoted 13 lines
* Use the macro call “return_ptr(ret)” at the end.
Signed-off-by Markus Elfring <[email protected]>--- drivers/soc/fsl/dpio/dpio-service.c | 15 ++++----------- 1 file changed, 4 insertions(+), 11 deletions(-)drivers/soc/fsl/dpio/dpio-service.c +0
-0@@ -652,23 +652,17 @@ EXPORT_SYMBOL_GPL(dpaa2_io_service_acquire);
Missing #include <linux/cleanup.h>
Quoted 12 lines
drivers/soc/fsl/dpio/dpio-service.c +1
-4struct dpaa2_io_store *dpaa2_io_store_create (unsigned int max_frames,struct device *dev) { -struct dpaa2_io_store *ret; -size_t size; -if (!max_frames || (max_frames >32 ))return NULL ; - ret =kmalloc_obj (*ret); + struct dpaa2_io_store *ret __free(kfree) = kmalloc_obj(*ret);if (!ret)
drivers/soc/fsl/dpio/dpio-service.c EXPORT_SYMBOL_GPL(dpaa2_io_service_acquire); · line 665
Quoted 10 lines
drivers/soc/fsl/dpio/dpio-service.c +2
-4return NULL ; ret->max = max_frames; - size = max_frames *sizeof (struct dpaa2_dq ) +64 ; - ret->alloced_addr =kzalloc (size,GFP_KERNEL ); -if (!ret->alloced_addr) { -kfree (ret); + ret->alloced_addr =kzalloc (max_frames *sizeof (struct dpaa2_dq ) +64 ,GFP_KERNEL ); +if (!ret->alloced_addr)return NULL ;
Why convert the ret allocation but not the ret->alloced_addr allocation?
regards,
dan carpenter
Quoted 23 lines
drivers/soc/fsl/dpio/dpio-service.c +1
-3- } ret->vaddr =PTR_ALIGN (ret->alloced_addr,64 ); ret->paddr = dma_map_single(dev, ret->vaddr,@@ -676,14 +670,13 @@ struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames, DMA_FROM_DEVICE);if (dma_mapping_error (dev, ret->paddr)) {kfree (ret->alloced_addr); -kfree (ret);return NULL ; } ret->idx =0 ; ret->dev = dev; -return ret; +return_ptr (ret); }EXPORT_SYMBOL_GPL (dpaa2_io_store_create);Signature · 1 line
2.54.0
Re: [PATCH v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
…
Quoted 1 line
Quoted 1 line
+++
b/drivers/soc/fsl/dpio/dpio-service.c
…
Quoted 9 lines
drivers/soc/fsl/dpio/dpio-service.c EXPORT_SYMBOL_GPL(dpaa2_io_service_acquire); · line 665
Quoted 7 lines
drivers/soc/fsl/dpio/dpio-service.c +2
-4- size = max_frames *sizeof (struct dpaa2_dq ) +64 ; - ret->alloced_addr =kzalloc (size,GFP_KERNEL ); -if (!ret->alloced_addr) { -kfree (ret); + ret->alloced_addr =kzalloc (max_frames *sizeof (struct dpaa2_dq ) +64 ,GFP_KERNEL ); +if (!ret->alloced_addr)return NULL ;Why convert the ret allocation but not the
ret->alloced_addrallocation?
I suggest to reconsider the suggested influence of the attribute “__free(kfree)” once more also according to data structure members.
Regards,
Markus
Ioana Ciornei
maintainerRe: [PATCH v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
In reply to Markus ElfringQuoted 11 lines · Markus Elfring
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 10:34:43 +0200Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").* Thus use the attribute “__free(kfree)”.
* Reduce the scope for the local variable “ret”.
Please don't. Let all the variables be defined at the start of each function.
Quoted 4 lines
* Omit two
kfree()calls accordingly.* Omit the local variable “size” (for another memory allocation).
Please remove this change, not related to the intention of the patch.
Ioana
Re: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 10 lines
Quoted 8 lines
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").* Thus use the attribute “__free(kfree)”.
* Reduce the scope for the local variable “ret”.
Please don't. Let all the variables be defined at the start of each function.
Do we stumble on another coding style preferences conflict here?
https://elixir.bootlin.com/linux/v7.2-rc4/source/include/linux/cleanup.h#L142-L146
Quoted 5 lines
Quoted 3 lines
* Omit two
kfree()calls accordingly.* Omit the local variable “size” (for another memory allocation).
Please remove this change, not related to the intention of the patch.
What does hinder to perform the required size determination as a direct parameter for a kzalloc() call?
Regards,
Markus
Ioana Ciornei
maintainerRe: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 5 lines · Markus Elfring
Quoted 5 lines
Quoted 5 lines
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines
* Thus use the attribute “__free(kfree)”.
* Reduce the scope for the local variable “ret”.
Please don't. Let all the variables be defined at the start of each function.
Do we stumble on another coding style preferences conflict here?
Yes, especially when moving the variable definition is unnecessary.
The context for the recommendation linked by you is (also from cleanup.h):
* "When multiple variables in the same scope have cleanup attributes,
* at exit from the scope their associated cleanup functions are run in
* reverse order of definition (last defined, first cleanup)." *
* When the unwind order matters it requires that variables be defined
* mid-function scope rather than at the top of the file. Take the
* following example and notice the bug highlighted by "!!"::
Which is not the case of these simple functions which only need a kfree.
Quoted 9 lines
Quoted 7 lines
Quoted 5 lines
* Omit two
kfree()calls accordingly.* Omit the local variable “size” (for another memory allocation).
Please remove this change, not related to the intention of the patch.
What does hinder to perform the required size determination as a direct parameter for a
kzalloc()call?
Nothing, but it's a different change. On top of this, it's just churning the code without an added benefit.
Ioana
Re: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Quoted 1 line
* Omit the local variable “size” (for another memory allocation).
Please remove this change, not related to the intention of the patch.
What does hinder to perform the required size determination as a direct parameter for a
kzalloc()call?Nothing, but it's a different change. On top of this, it's just churning the code without an added benefit.
It influences development concerns also according to preferred variable scopes.
Regards,
Markus
Re: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
In reply to Ioana CiorneiQuoted 13 lines
The context for the recommendation linked by you is (also from
cleanup.h):* "When multiple variables in the same scope have cleanup attributes,
* at exit from the scope their associated cleanup functions are run in
* reverse order of definition (last defined, first cleanup)." *
* When the unwind order matters it requires that variables be defined
* mid-function scope rather than at the top of the file. Take the
* following example and notice the bug highlighted by "!!"::
Quoted 1 line
Which is not the case of these simple functions which only need a kfree.
How will development concerns evolve further also according to previous clarification attempts?
Clarifying confusion of our variable placement rules caused by cleanup.h
https://lore.kernel.org/ksummit/CAHk-=whPZoi03ZwphxiW6cuWPtC3nyKYS8_BThgztCdgPWP1WA@mail.gmail.com/
https://lkml.org/lkml/2025/11/18/1766
Regards,
Markus
Ioana Ciornei
maintainerRe: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 13 lines · Markus Elfring
Quoted 13 lines
The context for the recommendation linked by you is (also from
cleanup.h):* "When multiple variables in the same scope have cleanup attributes,
* at exit from the scope their associated cleanup functions are run in
* reverse order of definition (last defined, first cleanup)." *
* When the unwind order matters it requires that variables be defined
* mid-function scope rather than at the top of the file. Take the
* following example and notice the bug highlighted by "!!"::
Quoted 6 lines
Quoted 1 line
Which is not the case of these simple functions which only need a kfree.
How will development concerns evolve further also according to previous clarification attempts?
Clarifying confusion of our variable placement rules caused by
cleanup.h
https://lore.kernel.org/ksummit/CAHk-=whPZoi03ZwphxiW6cuWPtC3nyKYS8_BThgztCdgPWP1WA@mail.gmail.com/
https://lkml.org/lkml/2025/11/18/1766
My preference still stands and I do not think they contradict the views expressed in the above links:
- Variables which do not have anything to do with cleanup.h should still be defined at the start of the function. This is the case of the qman_256_cycles_per_ns which you moved in 2/2 without any motive.
- In the case of a variable used with __free, if it can be done, I prefer to see the definition and initialization done at the start of the function.
For example:
struct dpaa2_io *obj __free(kfree) = kmalloc_obj(*obj); u32 qman_256_cycles_per_ns; if (!obj) return NULL;
I am not going to engage past this message.
Ioana
Re: [v2 1/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_store_create()
Quoted 8 lines
Quoted 5 lines
How will development concerns evolve further also according to previous clarification attempts?
Clarifying confusion of our variable placement rules caused by
cleanup.h
https://lore.kernel.org/ksummit/CAHk-=whPZoi03ZwphxiW6cuWPtC3nyKYS8_BThgztCdgPWP1WA@mail.gmail.com/
https://lkml.org/lkml/2025/11/18/1766My preference still stands and I do not think they contradict the views expressed in the above links:
It seems that more time and guidance adjustments would be needed to adapt development habits into further directions.
Quoted 6 lines
- Variables which do not have anything to do with
cleanup.hshould still be defined at the start of the function. This is the case of theqman_256_cycles_per_nswhich you moved in 2/2 without any motive.
- In the case of a variable used with__free, if it can be done, I prefer to see the definition and initialization done at the start of the function.
The discussed function implementations contain also input parameter validation. Thus I imagine that it would occasionally be nicer to avoid extra data processing accordingly.
Regards,
Markus
PATCH v2 2/2 soc: fsl: dpio: Use scope-based resource management in dpaa2_io_create()
In reply to Markus ElfringFrom: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 11:18:39 +0200
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit 54da6a0924311c7cf5015533991e44fb8eb12773 ("locking:
Introduce __cleanup() based infrastructure").
* Thus use the attribute “__free(kfree)”.
* Omit two kfree() calls accordingly.
* Reduce the scopes for the local variables “obj”
and “qman_256_cycles_per_ns”.
* Use the macro call “return_ptr(obj)” at the end.
--- drivers/soc/fsl/dpio/dpio-service.c | 22 ++++++++-------------- 1 file changed, 8 insertions(+), 14 deletions(-)
drivers/soc/fsl/dpio/dpio-service.c +8 -14
@@ -133,17 +133,14 @@ static void dpaa2_io_dim_work(struct work_struct *w)struct dpaa2_io *dpaa2_io_create (const struct dpaa2_io_desc *desc,struct device *dev) { -struct dpaa2_io *obj =kmalloc_obj (*obj); -u32 qman_256_cycles_per_ns; - -if (!obj) +/* check if CPU is out of range (-1 means any cpu) */ +if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ())return NULL ; -/* check if CPU is out of range (-1 means any cpu) */ -if (desc->cpu !=DPAA2_IO_ANY_CPU && desc->cpu >=num_possible_cpus ()) { -kfree (obj); + struct dpaa2_io *obj __free(kfree) = kmalloc_obj(*obj); + +if (!obj)return NULL ; - } obj->dpio_desc = *desc; obj->swp_desc.cena_bar = obj->dpio_desc.regs_cena;
@@ -155,14 +152,11 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc,* the interrupt timeout period register needs to be specified in QBMAN * clock cycles in increments of 256. */ - qman_256_cycles_per_ns =256000 / (obj->swp_desc.qman_clk /1000000 ); +u32 qman_256_cycles_per_ns =256000 / (obj->swp_desc.qman_clk /1000000 ); obj->swp_desc.qman_256_cycles_per_ns = qman_256_cycles_per_ns; obj->swp =qbman_swp_init (&obj->swp_desc); - -if (!obj->swp) { -kfree (obj); +if (!obj->swp)return NULL ; - }INIT_LIST_HEAD (&obj->node);spin_lock_init (&obj->lock_mgmt_cmd);
@@ -191,7 +185,7 @@ struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc, obj->bytes =0 ; obj->frames =0 ; -return obj; +return_ptr (obj); }/**
Signature · 1 line
2.54.0
Re: [PATCH v2 2/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_create()
Quoted 8 lines · Markus Elfring
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 11:18:39 +0200Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").
Quoted 6 lines
* Thus use the attribute “__free(kfree)”.
* Omit two
kfree()calls accordingly.* Reduce the scopes for the local variables “obj”
and “qman_256_cycles_per_ns”.
Why? No one does this except bcachefs and that was removed.
regards,
dan carpenter
Re: [v2 2/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_create()
Quoted 5 lines
Quoted 5 lines
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").
Quoted 8 lines
Quoted 6 lines
* Thus use the attribute “__free(kfree)”.
* Omit two
kfree()calls accordingly.* Reduce the scopes for the local variables “obj”
and “qman_256_cycles_per_ns”.Why? No one does this except bcachefs and that was removed.
Have you still got development difficulties with scope reductions for selected local variables?
https://elixir.bootlin.com/linux/v7.2-rc4/source/include/linux/cleanup.h#L142-L146
Regards,
Markus
Ioana Ciornei
maintainerRe: [PATCH v2 2/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_create()
In reply to Markus ElfringQuoted 8 lines · Markus Elfring
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 11:18:39 +0200Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").
Quoted 6 lines
* Thus use the attribute “__free(kfree)”.
* Omit two
kfree()calls accordingly.* Reduce the scopes for the local variables “obj”
and “qman_256_cycles_per_ns”.
Remove these changes. All variables should be defined at the start of the function, especially since there is no strong reason not to.
Ioana
Re: [v2 2/2] soc: fsl: dpio: Use scope-based resource management in dpaa2_io_create()
Quoted 7 lines
Quoted 7 lines
Scope-based resource management became supported for some programming interfaces by contributions of Peter Zijlstra on 2023-05-26. See also the commit
54da6a0924311c7cf5015533991e44fb8eb12773("locking:
Introduce__cleanup()based infrastructure").* Thus use the attribute “__free(kfree)”.
Quoted 7 lines
Quoted 4 lines
* Omit two
kfree()calls accordingly.* Reduce the scopes for the local variables “obj”
and “qman_256_cycles_per_ns”.Remove these changes. All variables should be defined at the start of the function, especially since there is no strong reason not to.
Can such a feedback be also interpreted in the way that you have got still difficulties with complete support of scope-based resource management?
Regards,
Markus
Christophe Leroy (CS GROUP)
maintainerRe: [PATCH v2 0/2] soc: fsl: dpio: Use scope-based resource management in two functions
In reply to Markus ElfringHi Ioana,
Le 06/07/2026 à 12:36, Markus Elfring a écrit :
Quoted 9 lines
From: Markus Elfring <[email protected]>
Date: Mon, 6 Jul 2026 11:42:10 +0200Some adjustment opportunities were picked up.
Markus Elfring (2):
Use scope-based resource management indpaa2_io_store_create()Use scope-based resource management indpaa2_io_create()
As the (new) maintainer of drivers/soc/fsl/dpio, could you have a look at this series ?
Thanks
Christophe
Quoted 8 lines
v2:
Christophe Leroy requested to apply the attribute “__free(kfree)”.drivers/soc/fsl/dpio/dpio-service.c | 37 ++++++++++------------------- 1 file changed, 12 insertions(+), 25 deletions(-)