Skip to contentSkip to conversation

Reconsidering kfree() calls for null pointers (with SmPL)

cocci In review Markus Elfring 9 participants 55 messages Mar 28, 2023 – Jan 12, 2024

Subsystems (11)
lore
Share & tools
lore
#0 lore raw

Hello,

I tried the following SmPL script out also on the source files from the software “Linux next-20230324”.

@display@
expression call, storage;
identifier target;
@@
*storage = call(...);
*if (!storage)

    goto target;

...
*target:
*kfree(storage)

119 source files were found where a check was performed for a failed function call and a kfree() call is immediately performed for the same variable. Thus a null pointer is probably passed at these source code places.
I imagine that such code should be reconsidered once more and improved accordingly.

How do you think about to achieve any adjustments in this design direction?

Regards,
Markus

#1 lore raw
Quoted 12 lines · Markus Elfring

Hello,

I tried the following SmPL script out also on the source files from the software “Linux next-20230324”.

@display@
expression call, storage;
identifier target;
@@
*storage = call(...);
*if (!storage)

    goto target;
Quoted 8 lines

...
*target:
*kfree(storage)

119 source files were found where a check was performed for a failed function call and a kfree() call is immediately performed for the same variable. Thus a null pointer is probably passed at these source code places.

It is valid to pass a NULL pointer to kfree(), it checks for that.

Quoted 1 line

I imagine that such code should be reconsidered once more and improved accordingly.

The only potential downside in the scenario is checking storage == NULL twice. But this is an error handling path, not a fast path. On the other hand, the code may look like this:

storage = call(...);
if (!storage)

    goto target;

if (some_other_condition)

    goto target;

target:
kfree(storage)

For some_other_condition, we want the kfree(). If you changed the code, to remove the extra NULL pointer check, you would have:

storage = call(...);
if (!storage)

    goto target1;

if (some_other_condition)

    goto target2;

target2:
kfree(storage)
target1:

The extra goto label and larger code is not worth saving the NULL pointer check, because this is error handling path.

Quoted 1 line

How do you think about to achieve any adjustments in this design direction?

Only in cases where it would not make the code more complex, i.e. if there are no "some_other_condition" that jumps to the same target after allocation to storage is successful.

Vlastimil

Quoted 2 lines

Regards,
Markus

#2 lore raw
Quoted 5 lines

It is valid to pass a NULL pointer to kfree(), it checks for that.

Quoted 1 line

I imagine that such code should be reconsidered once more and improved accordingly.

The only potential downside in the scenario is checking storage == NULL twice.

Will programming concerns evolve any further?

Quoted 1 line

But this is an error handling path, not a fast path.

I hope that also exception handling code can become efficient in more use cases.

Quoted 6 lines

On the other hand, the code may look like this:

storage = call(...);
if (!storage)

    goto target;
Quoted 8 lines

if (some_other_condition)

    goto target;

target:
kfree(storage)

For some_other_condition, we want the kfree().

Allocated resources should be properly released.

Quoted 11 lines

If you changed the code, to remove the extra NULL pointer check, you would have:

storage = call(...);
if (!storage)

    goto target1;

if (some_other_condition)

    goto target2;

target2:
kfree(storage)

A label like “free_this_resource” can be nicer, can't it?

Quoted 4 lines

target1:

The extra goto label and larger code is not worth saving the NULL pointer check, because this is error handling path.

Some code reviewers and programmers represent other development views.

Quoted 5 lines
Quoted 1 line

How do you think about to achieve any adjustments in this design direction?

Only in cases where it would not make the code more complex, i.e. if there are no "some_other_condition" that jumps to the same target after allocation to storage is successful.

Do you find any implementation details worth for another look also according to a special information source?
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources#MEM12C.Considerusingagotochainwhenleavingafunctiononerrorwhenusingandreleasingresources-CompliantSolution%28POSIX,GotoChain%29

Will circumstances evolve in ways so that you would adjust a desire to stick to only a single jump label?

Regards,
Markus

#3 lore raw
Quoted 7 lines · Markus Elfring
Quoted 5 lines

It is valid to pass a NULL pointer to kfree(), it checks for that.

Quoted 1 line

I imagine that such code should be reconsidered once more and improved accordingly.

The only potential downside in the scenario is checking storage == NULL twice.

Will programming concerns evolve any further?

It's very unlikely that the behavior of kfree(NULL) will change.

Quoted 3 lines
Quoted 1 line

But this is an error handling path, not a fast path.

I hope that also exception handling code can become efficient in more use cases.

Nobody will notice the efficiency, and it's not the only goal, code readability is also important.

Quoted 9 lines
Quoted 9 lines

On the other hand, the code may look like this:

storage = call(...);
if (!storage)

    goto target;

if (some_other_condition)

    goto target;
Quoted 7 lines
Quoted 5 lines

target:
kfree(storage)

For some_other_condition, we want the kfree().

Allocated resources should be properly released.

Sure. They are properly released in the above example.

Quoted 6 lines
Quoted 6 lines

If you changed the code, to remove the extra NULL pointer check, you would have:

storage = call(...);
if (!storage)

    goto target1;
Quoted 7 lines
Quoted 5 lines

if (some_other_condition)

    goto target2;

target2:
kfree(storage)

A label like “free_this_resource” can be nicer, can't it?

Sure, name was not the point. Just that there are now 2 labels.

Quoted 6 lines
Quoted 4 lines

target1:

The extra goto label and larger code is not worth saving the NULL pointer check, because this is error handling path.

Some code reviewers and programmers represent other development views.

That's the universal truth :)

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

How do you think about to achieve any adjustments in this design direction?

Only in cases where it would not make the code more complex, i.e. if there are no "some_other_condition" that jumps to the same target after allocation to storage is successful.

Quoted 7 lines

Do you find any implementation details worth for another look also according to a special information source?
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources#MEM12C.Considerusingagotochainwhenleavingafunctiononerrorwhenusingandreleasingresources-CompliantSolution%28POSIX,GotoChain%29

Will circumstances evolve in ways so that you would adjust a desire to stick to only a single jump label?

Depends on specific code cleanup. But generally, not if the change is only to add an extra label to skip over kfree() - I will unlikely consider such patches as useful.

Quoted 2 lines

Regards,
Markus

#4 lore raw

cover ARM: Adjustments for init_atags_procfs()

In reply to Markus Elfring

Date: Tue, 28 Mar 2023 19:06:32 +0200

A few update suggestions were taken into account
from static source code analysis.

Markus Elfring (2):
Delete an error message for a failed memory allocation Return directly after a failed kmalloc()

 arch/arm/kernel/atags_proc.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

--
2.40.0

#5 lore raw

PATCH 1/2 ARM: Delete an error message for a failed memory allocation in init_atags_procfs()

Date: Tue, 28 Mar 2023 18:40:26 +0200

Omit an extra message for a memory allocation failure in this function.

---
 arch/arm/kernel/atags_proc.c | 2 --
 1 file changed, 2 deletions(-)
arch/arm/kernel/atags_proc.c +0 -2
index 3ec2afe78423..2f08ab08e68f 100644
@@ -69,8 +69,6 @@ static int __init init_atags_procfs(void)
 
 nomem:
 	kfree(b);
-	pr_err("Exporting ATAGs: not enough memory\n");
-
 	return -ENOMEM;
 }
 arch_initcall(init_atags_procfs);

--
2.40.0

#6 lore raw

PATCH 2/2 ARM: Return directly after a failed kmalloc() in init_atags_procfs()

In reply to Markus Elfring

Date: Tue, 28 Mar 2023 18:56:31 +0200

Return directly after a call of the function “kmalloc” failed at the beginning.

---
 arch/arm/kernel/atags_proc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/kernel/atags_proc.c +1 -1
index 2f08ab08e68f..1c7441f3af85 100644
@@ -56,7 +56,7 @@ static int __init init_atags_procfs(void)
 
 	b = kmalloc(sizeof(*b) + size, GFP_KERNEL);
 	if (!b)
-		goto nomem;
+		return -ENOMEM;
 
 	b->size = size;
 	memcpy(b->data, atags_copy, size);

--
2.40.0

#7 lore raw

Re: [PATCH 0/2] ARM: Adjustments for init_atags_procfs()

In reply to Markus Elfring
Quoted 11 lines

Date: Tue, 28 Mar 2023 19:06:32 +0200

A few update suggestions were taken into account
from static source code analysis.

Markus Elfring (2):
Delete an error message for a failed memory allocation Return directly after a failed kmalloc()

 arch/arm/kernel/atags_proc.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-04/msg00034.html

Regards,
Markus

#9 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 5 lines · Markus Elfring
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry. I don't catch everything that is on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied. See signature below.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#10 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry.

Special mailing list settings probably influenced this situation.

Quoted 3 lines

I don't catch everything that is
on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied.

Can you collaborate also with mentioned mailing list archive interfaces?

Regards,
Markus

#11 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 8 lines · Markus Elfring
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry.

Quoted 7 lines

Special mailing list settings probably influenced this situation.

Quoted 3 lines

I don't catch everything that is
on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied.

Can you collaborate also with mentioned mailing list archive interfaces?

Err what? Sorry, I don't understand your comment.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#12 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry.

Special mailing list settings probably influenced this situation.

Did any communication filters hinder the clarification of further development ideas?

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

I don't catch everything that is
on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied.

Can you collaborate also with mentioned mailing list archive interfaces?

Err what? Sorry, I don't understand your comment.

Are you going to pick any patches up from linked information sources?

Regards,
Markus

#13 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 11 lines · Markus Elfring
Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry.

Special mailing list settings probably influenced this situation.

Did any communication filters hinder the clarification of further development ideas?

How about doing the research yourself? Using lore, it's possible to work it out. Find the message id. Then visit

https://lore.kernel.org/r/<message-id>

and study the result. It will give you the answer you want.

Quoted 9 lines
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

I don't catch everything that is
on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied.

Can you collaborate also with mentioned mailing list archive interfaces?

Err what? Sorry, I don't understand your comment.

Are you going to pick any patches up from linked information sources?

No. Did you read my first reply which told you how patches get applied to arch/arm? If you didn't, your patch won't get applied.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#14 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 4 lines
Quoted 1 line

Are you going to pick any patches up from linked information sources?

No. Did you read my first reply which told you how patches get applied to arch/arm? If you didn't, your patch won't get applied.

Do you request a resend of affected patches here?

Regards,
Markus

#15 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 6 lines · Markus Elfring
Quoted 4 lines
Quoted 1 line

Are you going to pick any patches up from linked information sources?

No. Did you read my first reply which told you how patches get applied to arch/arm? If you didn't, your patch won't get applied.

Do you request a resend of affected patches here?

Welcome to my /dev/null mailbox. It has plenty of space. Feel free to continue sending your claptrap there. Rest assured I won't read it anymore.

Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#16 lore raw

Re: Re: [0/2] ARM: Adjustments for init_atags_procfs()

In reply to Russell King (Oracle)
Quoted 11 lines · Russell King (Oracle)
Quoted 10 lines · Markus Elfring
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00098.html

I suspect no one looked at it, sorry.

Special mailing list settings probably influenced this situation.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

I don't catch everything that is
on the mailing list. Looks fine to me but it needs to end up in the patch system to be applied.

Can you collaborate also with mentioned mailing list archive interfaces?

Err what? Sorry, I don't understand your comment.

I am just generally following along here, but to give some context it seems like Markus is banned from posting to various kernel mailing lists[0][1][2].

Cheers,
Chris

[0]: https://lkml.org/lkml/2023/6/19/38
[1]: https://lore.kernel.org/lkml/CAHC9VhREfdgiCji=uEeCrc4w1kPGfnWGKnJuUYKXwTApdneSjQ@mail.gmail.com/T/#m1a55ecc3205045fe63ab0f12451705df911d31a0
[2]: https://lore.kernel.org/all/[email protected]/

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#17 lore raw

Re: [0/2] ARM: Adjustments for init_atags_procfs()

Quoted 3 lines

I am just generally following along here, but to give some context it seems like Markus is banned from posting to various kernel mailing lists[0][1][2].

This was the case for a while.
Were such communication filters reconsidered anyhow recently?

Regards,
Markus

#18 lore raw

cover perf/x86/intel/pt: Adjustments for pt_pmu_hw_init()

In reply to Markus Elfring

Date: Tue, 28 Mar 2023 20:34:32 +0200

A few update suggestions were taken into account
from static source code analysis.

Markus Elfring (2):
Use -ENOMEM directly for a return statement
Return directly after a failed kzalloc()

 arch/x86/events/intel/pt.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

--
2.40.0

#19 lore raw

PATCH 1/2 perf/x86/intel/pt: Use -ENOMEM directly for a return statement in pt_pmu_hw_init()

Date: Tue, 28 Mar 2023 20:22:05 +0200

* Return the value “-ENOMEM” directly without an extra assignment statement.

* Omit the variable “ret”.

---
 arch/x86/events/intel/pt.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/x86/events/intel/pt.c +1 -4
index 42a55794004a..93af27edded3 100644
@@ -190,7 +190,6 @@ static int __init pt_pmu_hw_init(void)
 	struct attribute **attrs;
 	size_t size;
 	u64 reg;
-	int ret;
 	long i;
 
 	rdmsrl(MSR_PLATFORM_INFO, reg);
@@ -242,7 +241,6 @@ static int __init pt_pmu_hw_init(void)
 			    &pt_pmu.caps[CPUID_EDX + i*PT_CPUID_REGS_NUM]);
 	}
 
-	ret = -ENOMEM;
 	size = sizeof(struct attribute *) * (ARRAY_SIZE(pt_caps)+1);
 	attrs = kzalloc(size, GFP_KERNEL);
 	if (!attrs)
@@ -273,8 +271,7 @@ static int __init pt_pmu_hw_init(void)
 
 fail:
 	kfree(attrs);
-
-	return ret;
+	return -ENOMEM;
 }
 
 #define RTIT_CTL_CYC_PSB (RTIT_CTL_CYCLEACC	| \

--
2.40.0

#20 lore raw

PATCH 2/2 perf/x86/intel/pt: Return directly after a failed kzalloc() in pt_pmu_hw_init()

In reply to Markus Elfring

Date: Tue, 28 Mar 2023 20:24:16 +0200

Return directly after a call of the function “kzalloc” failed at the beginning.

---
 arch/x86/events/intel/pt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/events/intel/pt.c +1 -1
index 93af27edded3..18c009bdb827 100644
@@ -244,7 +244,7 @@ static int __init pt_pmu_hw_init(void)
 	size = sizeof(struct attribute *) * (ARRAY_SIZE(pt_caps)+1);
 	attrs = kzalloc(size, GFP_KERNEL);
 	if (!attrs)
-		goto fail;
+		return -ENOMEM;
 
 	size = sizeof(struct dev_ext_attribute) * (ARRAY_SIZE(pt_caps)+1);
 	de_attrs = kzalloc(size, GFP_KERNEL);

--
2.40.0

#21 lore raw

Re: [PATCH 0/2] perf/x86/intel/pt: Adjustments for pt_pmu_hw_init()

In reply to Markus Elfring
Quoted 11 lines

Date: Tue, 28 Mar 2023 20:34:32 +0200

A few update suggestions were taken into account
from static source code analysis.

Markus Elfring (2):
Use -ENOMEM directly for a return statement
Return directly after a failed kzalloc()

 arch/x86/events/intel/pt.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00101.html

Regards,
Markus

#22 lore raw

PATCH 1/1 apparmor: Return directly after a failed kzalloc() in two functions

In reply to Markus Elfring

Date: Wed, 29 Mar 2023 11:50:44 +0200

1. Return directly after a call of the function “kzalloc” failed at the beginning in these function implementations.

2. Omit extra initialisations (for a few local variables) which became unnecessary with this refactoring.

This issue was detected by using the Coccinelle software.

---
 security/apparmor/crypto.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
security/apparmor/crypto.c +5 -5
index b498ed302461..6724e2ff6da8 100644
@@ -28,15 +28,15 @@ unsigned int aa_hash_size(void)
 char *aa_calc_hash(void *data, size_t len)
 {
 	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
-	char *hash = NULL;
-	int error = -ENOMEM;
+	char *hash;
+	int error;
 
 	if (!apparmor_tfm)
 		return NULL;
 
 	hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
 	if (!hash)
-		goto fail;
+		return ERR_PTR(-ENOMEM);
 
 	desc->tfm = apparmor_tfm;
 
@@ -62,7 +62,7 @@ int aa_calc_profile_hash(struct aa_profile *profile, u32 version, void *start,
 			 size_t len)
 {
 	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
-	int error = -ENOMEM;
+	int error;
 	__le32 le32_version = cpu_to_le32(version);
 
 	if (!aa_g_hash_policy)
@@ -73,7 +73,7 @@ int aa_calc_profile_hash(struct aa_profile *profile, u32 version, void *start,
 
 	profile->hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
 	if (!profile->hash)
-		goto fail;
+		return -ENOMEM;
 
 	desc->tfm = apparmor_tfm;
 

--
2.40.0

maintainer Acked-by on v1 1/1
#23 lore raw

Re: [PATCH] apparmor: Return directly after a failed kzalloc() in two functions

Quoted 5 lines · Markus Elfring

Date: Wed, 29 Mar 2023 11:50:44 +0200

1. Return directly after a call of the function “kzalloc” failed at the beginning in these function implementations.

Quoted 7 lines

2. Omit extra initialisations (for a few local variables) which became unnecessary with this refactoring.

This issue was detected by using the Coccinelle software.

thanks, I have pulled this into the apparmor tree

Quoted 48 lines
---
  security/apparmor/crypto.c | 10 +++++-----
  1 file changed, 5 insertions(+), 5 deletions(-)
security/apparmor/crypto.c +5 -6
index b498ed302461..6724e2ff6da8 100644
@@ -28,15 +28,15 @@ unsigned int aa_hash_size(void)
  char *aa_calc_hash(void *data, size_t len)
  {
  	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
-	char *hash = NULL;
-	int error = -ENOMEM;
+	char *hash;
+	int error;
 
  	if (!apparmor_tfm)
  		return NULL;
 
  	hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
  	if (!hash)
-		goto fail;
+		return ERR_PTR(-ENOMEM);
 
  	desc->tfm = apparmor_tfm;
 
@@ -62,7 +62,7 @@ int aa_calc_profile_hash(struct aa_profile *profile, u32 version, void *start,
  			 size_t len)
  {
  	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
-	int error = -ENOMEM;
+	int error;
  	__le32 le32_version = cpu_to_le32(version);
 
  	if (!aa_g_hash_policy)
@@ -73,7 +73,7 @@ int aa_calc_profile_hash(struct aa_profile *profile, u32 version, void *start,
 
  	profile->hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
  	if (!profile->hash)
-		goto fail;
+		return -ENOMEM;
 
  	desc->tfm = apparmor_tfm;
 
--

2.40.0

#24 lore raw

cover lru_cache: Adjustments for lc_create()

In reply to Markus Elfring

Date: Wed, 29 Mar 2023 15:30:23 +0200

Some update suggestions were taken into account
from static source code analysis.

Markus Elfring (3):
Return directly after a failed kzalloc()
Improve two size determinations
Improve exception handling

 lib/lru_cache.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

--
2.40.0

#25 lore raw

PATCH 1/3 lru_cache: Return directly after a failed kzalloc() in lc_create()

Date: Wed, 29 Mar 2023 14:45:34 +0200

1. Return directly after a call of the function “kzalloc” failed at the beginning in these function implementations.

2. Omit extra initialisations (for the variables “slot” and “element”) which became unnecessary with this refactoring.

---
 lib/lru_cache.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
lib/lru_cache.c +4 -3
index b3d9187611de..e0db27b3a2d7 100644
@@ -78,8 +78,8 @@ struct lru_cache *lc_create(const char *name, struct kmem_cache *cache,
 		unsigned max_pending_changes,
 		unsigned e_count, size_t e_size, size_t e_off)
 {
-	struct hlist_head *slot = NULL;
-	struct lc_element **element = NULL;
+	struct hlist_head *slot;
+	struct lc_element **element;
 	struct lru_cache *lc;
 	struct lc_element *e;
 	unsigned cache_obj_size = kmem_cache_size(cache);
@@ -96,7 +96,8 @@ struct lru_cache *lc_create(const char *name, struct kmem_cache *cache,
 
 	slot = kcalloc(e_count, sizeof(struct hlist_head), GFP_KERNEL);
 	if (!slot)
-		goto out_fail;
+		return NULL;
+
 	element = kcalloc(e_count, sizeof(struct lc_element *), GFP_KERNEL);
 	if (!element)
 		goto out_fail;

--
2.40.0

#26 lore raw

PATCH 2/3 lru_cache: Improve two size determinations in lc_create()

In reply to Markus Elfring

Date: Wed, 29 Mar 2023 15:00:13 +0200

Replace the specification of data structures by pointer dereferences as the parameter for the operator “sizeof” to make the corresponding size determination a bit safer according to the Linux coding style convention.

---
 lib/lru_cache.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
lib/lru_cache.c +2 -2
index e0db27b3a2d7..31820f03b146 100644
@@ -94,11 +94,11 @@ struct lru_cache *lc_create(const char *name, struct kmem_cache *cache,
 	if (e_count > LC_MAX_ACTIVE)
 		return NULL;
 
-	slot = kcalloc(e_count, sizeof(struct hlist_head), GFP_KERNEL);
+	slot = kcalloc(e_count, sizeof(*slot), GFP_KERNEL);
 	if (!slot)
 		return NULL;
 
-	element = kcalloc(e_count, sizeof(struct lc_element *), GFP_KERNEL);
+	element = kcalloc(e_count, sizeof(*element), GFP_KERNEL);
 	if (!element)
 		goto out_fail;
 

--
2.40.0

#27 lore raw

PATCH 3/3 lru_cache: Improve exception handling in lc_create()

In reply to Markus Elfring

Date: Wed, 29 Mar 2023 15:20:39 +0200

The label “out_fail” was used to jump to a kfree() call despite of the detail in the implementation of the function “lc_create” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

Thus use more appropriate labels instead.

This issue was detected by using the Coccinelle software.

---
 lib/lru_cache.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
lib/lru_cache.c +4 -3
index 31820f03b146..fdc8bd6fc888 100644
@@ -100,11 +100,11 @@ struct lru_cache *lc_create(const char *name, struct kmem_cache *cache,
 
 	element = kcalloc(e_count, sizeof(*element), GFP_KERNEL);
 	if (!element)
-		goto out_fail;
+		goto free_slot;
 
 	lc = kzalloc(sizeof(*lc), GFP_KERNEL);
 	if (!lc)
-		goto out_fail;
+		goto free_element;
 
 	INIT_LIST_HEAD(&lc->in_use);
 	INIT_LIST_HEAD(&lc->lru);
@@ -142,8 +142,9 @@ struct lru_cache *lc_create(const char *name, struct kmem_cache *cache,
 		kmem_cache_free(cache, p - e_off);
 	}
 	kfree(lc);
-out_fail:
+free_element:
 	kfree(element);
+free_slot:
 	kfree(slot);
 	return NULL;
 }

--
2.40.0

#28 lore raw

Re: [PATCH 0/3] lru_cache: Adjustments for lc_create()

In reply to Markus Elfring
Quoted 5 lines

Date: Wed, 29 Mar 2023 15:30:23 +0200

Some update suggestions were taken into account
from static source code analysis.

Quoted 7 lines

Markus Elfring (3):
Return directly after a failed kzalloc()
Improve two size determinations
Improve exception handling

 lib/lru_cache.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00110.html

Regards,
Markus

#29 lore raw

PATCH 1/1 io_uring: Fix exception handling in io_ring_ctx_alloc()

In reply to Markus Elfring

Date: Wed, 29 Mar 2023 17:35:16 +0200

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

3. Omit the statement “kfree(ctx->io_bl);”.

This issue was detected by using the Coccinelle software.

Fixes 206aefde4f886fdeb3b6339aacab3a85fb74cb7e ("io_uring: reduce/pack size of io_ring_ctx")
Fixes 78076bb64aa8ba5b7207c38b2660a9e10ffa8cc7 ("io_uring: use hash table for poll command lookups")
Fixes 9cfc7e94e42be9c73072606b84d4574a0a2ec270 ("io_uring: get rid of hashed provided buffer groups")
Fixes 9ca9fb24d5febccea354089c41f96a8ad0d853f8 ("io_uring: mutex locked poll hashing")
---
 io_uring/io_uring.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)
io_uring/io_uring.c +10 -7
index 536940675c67..fceb74d7d851 100644
@@ -291,19 +291,19 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	hash_bits = ilog2(p->cq_entries) - 5;
 	hash_bits = clamp(hash_bits, 1, 8);
 	if (io_alloc_hash_table(&ctx->cancel_table, hash_bits))
-		goto err;
+		goto destroy_io_bl_xa;
 	if (io_alloc_hash_table(&ctx->cancel_table_locked, hash_bits))
-		goto err;
+		goto free_cancel_table_hbs;
 
 	ctx->dummy_ubuf = kzalloc(sizeof(*ctx->dummy_ubuf), GFP_KERNEL);
 	if (!ctx->dummy_ubuf)
-		goto err;
+		goto free_cancel_table_locked_hbs;
 	/* set invalid range, so io_import_fixed() fails meeting it */
 	ctx->dummy_ubuf->ubuf = -1UL;
 
 	if (percpu_ref_init(&ctx->refs, io_ring_ctx_ref_free,
 			    0, GFP_KERNEL))
-		goto err;
+		goto free_ubuf;
 
 	ctx->flags = p->flags;
 	init_waitqueue_head(&ctx->sqo_sq_wait);
@@ -337,11 +337,14 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	INIT_DELAYED_WORK(&ctx->fallback_work, io_fallback_req_func);
 	INIT_WQ_LIST(&ctx->submit_state.compl_reqs);
 	return ctx;
-err:
+
+free_ubuf:
 	kfree(ctx->dummy_ubuf);
-	kfree(ctx->cancel_table.hbs);
+free_cancel_table_locked_hbs:
 	kfree(ctx->cancel_table_locked.hbs);
-	kfree(ctx->io_bl);
+free_cancel_table_hbs:
+	kfree(ctx->cancel_table.hbs);
+destroy_io_bl_xa:
 	xa_destroy(&ctx->io_bl_xa);
 	kfree(ctx);
 	return NULL;

--
2.40.0

#30 lore raw

Re: io_uring: Fix exception handling in io_ring_ctx_alloc()

Quoted 10 lines

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

3. Omit the statement “kfree(ctx->io_bl);”.

Is this patch still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00114.html

Regards,
Markus

#31 lore raw

Re: [cocci] [PATCH] io_uring: Fix exception handling in io_ring_ctx_alloc()

In reply to Markus Elfring
Quoted 10 lines · Markus Elfring

Date: Wed, 29 Mar 2023 17:35:16 +0200

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

FWIW, I don't think it makes sense to have the extra labels or re-sort without a good reason. kfree works fine with the NULL pointers, so there is no bug to be fixed and moving code around for no reason just makes life painful for backporters.

Also, the patch no longer applies.

Quoted 1 line

3. Omit the statement “kfree(ctx->io_bl);”.

From a quick look, this might still make sense. can you confirm and make that change into a separate patch?

Signature · 1 line
Gabriel Krisman Bertazi
#32 lore raw

v2 cover io_uring: Adjustments for io_ring_ctx_alloc()

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 21:38:12 +0100

A few update suggestions were taken into account
from static source code analysis.

Markus Elfring (2):
Delete a redundant kfree() call
Improve exception handling

 io_uring/io_uring.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

--
2.43.0

#33 lore raw

PATCH v2 1/2 io_uring: Delete a redundant kfree() call in io_ring_ctx_alloc()

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 20:54:43 +0100

Another useful pointer was not reassigned to the data structure member “io_bl” by this function implementation.
Thus omit a redundant call of the function “kfree” at the end.

---

v2:
A change request by Gabriel Krisman Bertazi was applied here.

 io_uring/io_uring.c | 1 -
 1 file changed, 1 deletion(-)
io_uring/io_uring.c +0 -1
index 86761ec623f9..c9a63c39cdd0 100644
@@ -344,7 +344,6 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 err:
 	kfree(ctx->cancel_table.hbs);
 	kfree(ctx->cancel_table_locked.hbs);
-	kfree(ctx->io_bl);
 	xa_destroy(&ctx->io_bl_xa);
 	kfree(ctx);
 	return NULL;

--
2.43.0

Reviewed-by on v2 1/2
#34 lore raw

Re: [PATCH v2 1/2] io_uring: Delete a redundant kfree() call in io_ring_ctx_alloc()

Quoted 6 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 20:54:43 +0100

Another useful pointer was not reassigned to the data structure member “io_bl” by this function implementation.
Thus omit a redundant call of the function “kfree” at the end.

Perhaps rewrite this to:

ctx->io_bl is initialized later through IORING_OP_PROVIDE_BUFFERS or IORING_REGISTER_PBUF_RING later on, so there is nothing to free in the ctx allocation error path.

Other than that, and for this patch only:

thanks,

Quoted 26 lines

---

v2:
A change request by Gabriel Krisman Bertazi was applied here.

 io_uring/io_uring.c | 1 -
 1 file changed, 1 deletion(-)
io_uring/io_uring.c +0 -2
index 86761ec623f9..c9a63c39cdd0 100644
@@ -344,7 +344,6 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 err:
 	kfree(ctx->cancel_table.hbs);
 	kfree(ctx->cancel_table_locked.hbs);
-	kfree(ctx->io_bl);
 	xa_destroy(&ctx->io_bl_xa);
 	kfree(ctx);
 	return NULL;
--

2.43.0

Signature · 1 line
Gabriel Krisman Bertazi
maintainer
#35 lore raw

Re: [PATCH v2 1/2] io_uring: Delete a redundant kfree() call in io_ring_ctx_alloc()

Quoted 8 lines · Gabriel Krisman Bertazi
Quoted 6 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 20:54:43 +0100

Another useful pointer was not reassigned to the data structure member ?io_bl? by this function implementation.
Thus omit a redundant call of the function ?kfree? at the end.

This is just nonsense...

On top of that, this patch is pointless, and the 2nd patch is even worse in that it just makes a mess of cleanup. And for what reasoning? Absolutely none.

There's a reason why I filter emails from this particular author straight to the trash, there's a long history of this kind of thing and not understanding feedback.

Signature · 1 line
Jens Axboe
#36 lore raw

Re: [PATCH v2 1/2] io_uring: Delete a redundant kfree() call in io_ring_ctx_alloc()

Quoted 10 lines · Jens Axboe
Quoted 8 lines · Gabriel Krisman Bertazi
Quoted 6 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 20:54:43 +0100

Another useful pointer was not reassigned to the data structure member ?io_bl? by this function implementation.
Thus omit a redundant call of the function ?kfree? at the end.

Quoted 5 lines

This is just nonsense...

On top of that, this patch is pointless, and the 2nd patch is even worse in that it just makes a mess of cleanup. And for what reasoning? Absolutely none.

Ah, The description is non-sense, but the change in this patch seemed correct to me, even if pointless, which is why I reviewed it. patch 2 is just garbage.

Quoted 3 lines

There's a reason why I filter emails from this particular author straight to the trash, there's a long history of this kind of thing and not understanding feedback.

Clearly there is background with this author that I wasn't aware, and just based on his responses, I can see your point. So I apologize for giving him space to continue the spamming. My bad.

Signature · 1 line
Gabriel Krisman Bertazi
#37 lore raw

Re: [v2 1/2] io_uring: Delete a redundant kfree() call in io_ring_ctx_alloc()

Quoted 1 line

patch 2 is just garbage.

Such a view can eventually be reconsidered if the support would ever grow also for the application of additional labels.
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

Quoted 3 lines

Clearly there is background with this author that I wasn't aware, and just based on his responses, I can see your point. So I apologize for giving him space to continue the spamming.

The change reluctance can be adapted according to the spectrum of presented source code adjustment possibilities, can't it?

Regards,
Markus

#38 lore raw

PATCH v2 2/2 io_uring: Improve exception handling in io_ring_ctx_alloc()

In reply to Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 21:15:48 +0100

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

---

See also:
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

 io_uring/io_uring.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
io_uring/io_uring.c +10 -5
index c9a63c39cdd0..7727cdd505ae 100644
@@ -295,12 +295,14 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	hash_bits = ilog2(p->cq_entries) - 5;
 	hash_bits = clamp(hash_bits, 1, 8);
 	if (io_alloc_hash_table(&ctx->cancel_table, hash_bits))
-		goto err;
+		goto destroy_io_bl_xa;
+
 	if (io_alloc_hash_table(&ctx->cancel_table_locked, hash_bits))
-		goto err;
+		goto free_cancel_table_hbs;
+
 	if (percpu_ref_init(&ctx->refs, io_ring_ctx_ref_free,
 			    0, GFP_KERNEL))
-		goto err;
+		goto free_cancel_table_locked_hbs;
 
 	ctx->flags = p->flags;
 	init_waitqueue_head(&ctx->sqo_sq_wait);
@@ -341,9 +343,12 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	INIT_WQ_LIST(&ctx->submit_state.compl_reqs);
 	INIT_HLIST_HEAD(&ctx->cancelable_uring_cmd);
 	return ctx;
-err:
-	kfree(ctx->cancel_table.hbs);
+
+free_cancel_table_locked_hbs:
 	kfree(ctx->cancel_table_locked.hbs);
+free_cancel_table_hbs:
+	kfree(ctx->cancel_table.hbs);
+destroy_io_bl_xa:
 	xa_destroy(&ctx->io_bl_xa);
 	kfree(ctx);
 	return NULL;

--
2.43.0

#39 lore raw

Re: [PATCH v2 2/2] io_uring: Improve exception handling in io_ring_ctx_alloc()

Quoted 7 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 21:15:48 +0100

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

It's _much_ simpler the way it currently is, compare it with maintaining a bunch of labels. That is the advantage of being able to distinguish un-allocated state like NULL, just kfree them and don't care about jumping to a wrong one or keeping them in order.

Quoted 54 lines

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

---

See also:
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

  io_uring/io_uring.c | 15 ++++++++++-----
  1 file changed, 10 insertions(+), 5 deletions(-)
io_uring/io_uring.c +10 -6
index c9a63c39cdd0..7727cdd505ae 100644
@@ -295,12 +295,14 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
  	hash_bits = ilog2(p->cq_entries) - 5;
  	hash_bits = clamp(hash_bits, 1, 8);
  	if (io_alloc_hash_table(&ctx->cancel_table, hash_bits))
-		goto err;
+		goto destroy_io_bl_xa;
+
  	if (io_alloc_hash_table(&ctx->cancel_table_locked, hash_bits))
-		goto err;
+		goto free_cancel_table_hbs;
+
  	if (percpu_ref_init(&ctx->refs, io_ring_ctx_ref_free,
  			    0, GFP_KERNEL))
-		goto err;
+		goto free_cancel_table_locked_hbs;
 
  	ctx->flags = p->flags;
  	init_waitqueue_head(&ctx->sqo_sq_wait);
@@ -341,9 +343,12 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
  	INIT_WQ_LIST(&ctx->submit_state.compl_reqs);
  	INIT_HLIST_HEAD(&ctx->cancelable_uring_cmd);
  	return ctx;
-err:
-	kfree(ctx->cancel_table.hbs);
+
+free_cancel_table_locked_hbs:
  	kfree(ctx->cancel_table_locked.hbs);
+free_cancel_table_hbs:
+	kfree(ctx->cancel_table.hbs);
+destroy_io_bl_xa:
  	xa_destroy(&ctx->io_bl_xa);
  	kfree(ctx);
  	return NULL;
--

2.43.0

Signature · 1 line
Pavel Begunkov
#40 lore raw

Re: [PATCH v2 2/2] io_uring: Improve exception handling in io_ring_ctx_alloc()

In reply to Markus Elfring
Quoted 12 lines · Markus Elfring

From: Markus Elfring <[email protected]>
Date: Wed, 10 Jan 2024 21:15:48 +0100

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

As I mentioned on v1, this doesn't do us any good, as kfree can handle NULL pointers just fine, and changes like this becomes churn later when backporting or modifying the code.

Signature · 1 line
Gabriel Krisman Bertazi
#41 lore raw

Re: [v2 2/2] io_uring: Improve exception handling in io_ring_ctx_alloc()

Quoted 11 lines
Quoted 9 lines

The label “err” was used to jump to a kfree() call despite of the detail in the implementation of the function “io_ring_ctx_alloc” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Reorder jump targets at the end.

As I mentioned on v1, this doesn't do us any good,

I dare to present other development views.

Quoted 1 line

as kfree can handle NULL pointers just fine,

Yes, this is the case.

Would you dare to categorise such a special function calls as redundant?

May it be skipped in more cases?

Quoted 1 line

and changes like this becomes churn later when backporting or modifying the code.

There are usual opportunities to consider for further collateral evolution.

Regards,
Markus

#42 lore raw

cover overlayfs: Adjustments for ovl_fill_super()

In reply to Markus Elfring

Date: Thu, 30 Mar 2023 10:38:23 +0200

Some update suggestions were taken into account
from static source code analysis.

Markus Elfring (4):
Return directly for two checks
Improve two size determinations
Improve exception handling
Move some assignments for the variable “err”

 fs/overlayfs/super.c | 72 ++++++++++++++++++++++++--------------------
 1 file changed, 39 insertions(+), 33 deletions(-)

--
2.40.0

#43 lore raw

PATCH 1/4 overlayfs: Return directly for two checks in ovl_fill_super()

Date: Thu, 30 Mar 2023 08:48:34 +0200

1. Return directly after a call of the function “kzalloc” failed or the “user_ns” was different in a data structure at the beginning.

2. Delete the label “out” which became unnecessary with this refactoring.

---
 fs/overlayfs/super.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
fs/overlayfs/super.c +2 -5
index f97ad8b40dbb..c0d33ab331ec 100644
@@ -1894,16 +1894,14 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	unsigned int numlower;
 	int err;
 
-	err = -EIO;
 	if (WARN_ON(sb->s_user_ns != current_user_ns()))
-		goto out;
+		return -EIO;
 
 	sb->s_d_op = &ovl_dentry_operations;
 
-	err = -ENOMEM;
 	ofs = kzalloc(sizeof(struct ovl_fs), GFP_KERNEL);
 	if (!ofs)
-		goto out;
+		return -ENOMEM;
 
 	err = -ENOMEM;
 	ofs->creator_cred = cred = prepare_creds();
@@ -2073,7 +2071,6 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	kfree(splitlower);
 	path_put(&upperpath);
 	ovl_free_fs(ofs);
-out:
 	return err;
 }
 

--
2.40.0

#44 lore raw

PATCH 2/4 overlayfs: Improve two size determinations in ovl_fill_super()

In reply to Markus Elfring

Date: Thu, 30 Mar 2023 09:02:10 +0200

Replace the specification of data structures by pointer dereferences as the parameter for the operator “sizeof” to make the corresponding size determination a bit safer according to the Linux coding style convention.

---
 fs/overlayfs/super.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
fs/overlayfs/super.c +2 -2
index c0d33ab331ec..c721d6fe9cb4 100644
@@ -1899,7 +1899,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 
 	sb->s_d_op = &ovl_dentry_operations;
 
-	ofs = kzalloc(sizeof(struct ovl_fs), GFP_KERNEL);
+	ofs = kzalloc(sizeof(*ofs), GFP_KERNEL);
 	if (!ofs)
 		return -ENOMEM;
 
@@ -1941,7 +1941,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	}
 
 	err = -ENOMEM;
-	layers = kcalloc(numlower + 1, sizeof(struct ovl_layer), GFP_KERNEL);
+	layers = kcalloc(numlower + 1, sizeof(*layers), GFP_KERNEL);
 	if (!layers)
 		goto out_err;
 

--
2.40.0

#45 lore raw

PATCH 3/4 overlayfs: Improve exception handling in ovl_fill_super()

In reply to Markus Elfring

Date: Thu, 30 Mar 2023 09:51:02 +0200

The label “out_err” was used to jump to a kfree() call despite of the detail in the implementation of the function “ovl_fill_super” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

1. Thus use more appropriate labels instead.

2. Omit an extra initialisation (for the variable “splitlower”) which became unnecessary with this refactoring.

This issue was detected by using the Coccinelle software.

---
 fs/overlayfs/super.c | 28 +++++++++++++++-------------
 1 file changed, 15 insertions(+), 13 deletions(-)
fs/overlayfs/super.c +15 -13
index c721d6fe9cb4..df5628056d93 100644
@@ -1890,7 +1890,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	struct ovl_fs *ofs;
 	struct ovl_layer *layers;
 	struct cred *cred;
-	char *splitlower = NULL;
+	char *splitlower;
 	unsigned int numlower;
 	int err;
 
@@ -1906,7 +1906,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	err = -ENOMEM;
 	ofs->creator_cred = cred = prepare_creds();
 	if (!cred)
-		goto out_err;
+		goto out_free_ofs;
 
 	/* Is there a reason anyone would want not to share whiteouts? */
 	ofs->share_whiteout = true;
@@ -1918,32 +1918,32 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	ofs->config.metacopy = ovl_metacopy_def;
 	err = ovl_parse_opt((char *) data, &ofs->config);
 	if (err)
-		goto out_err;
+		goto out_free_ofs;
 
 	err = -EINVAL;
 	if (!ofs->config.lowerdir) {
 		if (!silent)
 			pr_err("missing 'lowerdir'\n");
-		goto out_err;
+		goto out_free_ofs;
 	}
 
 	err = -ENOMEM;
 	splitlower = kstrdup(ofs->config.lowerdir, GFP_KERNEL);
 	if (!splitlower)
-		goto out_err;
+		goto out_free_ofs;
 
 	err = -EINVAL;
 	numlower = ovl_split_lowerdirs(splitlower);
 	if (numlower > OVL_MAX_STACK) {
 		pr_err("too many lower directories, limit is %d\n",
 		       OVL_MAX_STACK);
-		goto out_err;
+		goto out_free_splitlower;
 	}
 
 	err = -ENOMEM;
 	layers = kcalloc(numlower + 1, sizeof(*layers), GFP_KERNEL);
 	if (!layers)
-		goto out_err;
+		goto out_free_splitlower;
 
 	ofs->layers = layers;
 	/* Layer 0 is reserved for upper even if there's no upper */
@@ -1970,12 +1970,12 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 		err = -EINVAL;
 		if (!ofs->config.workdir) {
 			pr_err("missing 'workdir'\n");
-			goto out_err;
+			goto out_free_splitlower;
 		}
 
 		err = ovl_get_upper(sb, ofs, &layers[0], &upperpath);
 		if (err)
-			goto out_err;
+			goto out_free_splitlower;
 
 		upper_sb = ovl_upper_mnt(ofs)->mnt_sb;
 		if (!ovl_should_sync(ofs)) {
@@ -1983,13 +1983,13 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 			if (errseq_check(&upper_sb->s_wb_err, ofs->errseq)) {
 				err = -EIO;
 				pr_err("Cannot mount volatile when upperdir has an unseen error. Sync upperdir fs to clear state.\n");
-				goto out_err;
+				goto out_free_splitlower;
 			}
 		}
 
 		err = ovl_get_workdir(sb, ofs, &upperpath);
 		if (err)
-			goto out_err;
+			goto out_free_splitlower;
 
 		if (!ofs->workdir)
 			sb->s_flags |= SB_RDONLY;
@@ -2000,7 +2000,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	oe = ovl_get_lowerstack(sb, splitlower, numlower, ofs, layers);
 	err = PTR_ERR(oe);
 	if (IS_ERR(oe))
-		goto out_err;
+		goto out_free_splitlower;
 
 	/* If the upper fs is nonexistent, we mark overlayfs r/o too */
 	if (!ovl_upper_mnt(ofs))
@@ -2067,9 +2067,11 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 out_free_oe:
 	ovl_entry_stack_free(oe);
 	kfree(oe);
-out_err:
+out_free_splitlower:
 	kfree(splitlower);
+out_put_path:
 	path_put(&upperpath);
+out_free_ofs:
 	ovl_free_fs(ofs);
 	return err;
 }

--
2.40.0

#46 lore raw

PATCH 4/4 overlayfs: Move some assignments for the variable “err” in ovl_fill_super()

In reply to Markus Elfring

Date: Thu, 30 Mar 2023 10:25:22 +0200

One local variable was set to an error code in some cases before a concrete error situation was detected. Thus move the corresponding assignments into if branches to indicate a software failure there.

---
 fs/overlayfs/super.c | 33 ++++++++++++++++++++-------------
 1 file changed, 20 insertions(+), 13 deletions(-)
fs/overlayfs/super.c +20 -13
index df5628056d93..cdef9934faf6 100644
@@ -1903,10 +1903,11 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	if (!ofs)
 		return -ENOMEM;
 
-	err = -ENOMEM;
 	ofs->creator_cred = cred = prepare_creds();
-	if (!cred)
+	if (!cred) {
+		err = -ENOMEM;
 		goto out_free_ofs;
+	}
 
 	/* Is there a reason anyone would want not to share whiteouts? */
 	ofs->share_whiteout = true;
@@ -1920,30 +1921,34 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	if (err)
 		goto out_free_ofs;
 
-	err = -EINVAL;
 	if (!ofs->config.lowerdir) {
 		if (!silent)
 			pr_err("missing 'lowerdir'\n");
+
+		err = -EINVAL;
 		goto out_free_ofs;
 	}
 
-	err = -ENOMEM;
 	splitlower = kstrdup(ofs->config.lowerdir, GFP_KERNEL);
-	if (!splitlower)
+	if (!splitlower) {
+		err = -ENOMEM;
 		goto out_free_ofs;
+	}
 
-	err = -EINVAL;
 	numlower = ovl_split_lowerdirs(splitlower);
 	if (numlower > OVL_MAX_STACK) {
 		pr_err("too many lower directories, limit is %d\n",
 		       OVL_MAX_STACK);
+
+		err = -EINVAL;
 		goto out_free_splitlower;
 	}
 
-	err = -ENOMEM;
 	layers = kcalloc(numlower + 1, sizeof(*layers), GFP_KERNEL);
-	if (!layers)
+	if (!layers) {
+		err = -ENOMEM;
 		goto out_free_splitlower;
+	}
 
 	ofs->layers = layers;
 	/* Layer 0 is reserved for upper even if there's no upper */
@@ -1967,9 +1972,9 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	if (ofs->config.upperdir) {
 		struct super_block *upper_sb;
 
-		err = -EINVAL;
 		if (!ofs->config.workdir) {
 			pr_err("missing 'workdir'\n");
+			err = -EINVAL;
 			goto out_free_splitlower;
 		}
 
@@ -1998,9 +2003,10 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 		sb->s_time_gran = upper_sb->s_time_gran;
 	}
 	oe = ovl_get_lowerstack(sb, splitlower, numlower, ofs, layers);
-	err = PTR_ERR(oe);
-	if (IS_ERR(oe))
+	if (IS_ERR(oe)) {
+		err = PTR_ERR(oe);
 		goto out_free_splitlower;
+	}
 
 	/* If the upper fs is nonexistent, we mark overlayfs r/o too */
 	if (!ovl_upper_mnt(ofs))
@@ -2052,10 +2058,11 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	sb->s_flags |= SB_POSIXACL;
 	sb->s_iflags |= SB_I_SKIP_SYNC;
 
-	err = -ENOMEM;
 	root_dentry = ovl_get_root(sb, upperpath.dentry, oe);
-	if (!root_dentry)
+	if (!root_dentry) {
+		err = -ENOMEM;
 		goto out_free_oe;
+	}
 
 	mntput(upperpath.mnt);
 	kfree(splitlower);

--
2.40.0

#47 lore raw

Re: [PATCH 0/4] overlayfs: Adjustments for ovl_fill_super()

In reply to Markus Elfring
Quoted 5 lines

Date: Thu, 30 Mar 2023 10:38:23 +0200

Some update suggestions were taken into account
from static source code analysis.

Quoted 8 lines

Markus Elfring (4):
Return directly for two checks
Improve two size determinations
Improve exception handling
Move some assignments for the variable “err”

 fs/overlayfs/super.c | 72 ++++++++++++++++++++++++--------------------
 1 file changed, 39 insertions(+), 33 deletions(-)

Is this patch series still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

Regards,
Markus

maintainer
#48 lore raw

Re: [PATCH 0/4] overlayfs: Adjustments for ovl_fill_super()

Quoted 12 lines · Markus Elfring
Quoted 12 lines

Date: Thu, 30 Mar 2023 10:38:23 +0200

Some update suggestions were taken into account
from static source code analysis.

Markus Elfring (4):
Return directly for two checks
Improve two size determinations
Improve exception handling
Move some assignments for the variable “err”

Quoted 5 lines
Quoted 2 lines
 fs/overlayfs/super.c | 72 ++++++++++++++++++++++++--------------------
 1 file changed, 39 insertions(+), 33 deletions(-)

Is this patch series still in review queues?

Sorry, this series was not on my radar.

Quoted 4 lines

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

I will queue cleanup patches 1-2, but I do not like patches 3/4 and 4/4. I do not think that they make the code better to read or maintain.

Thanks,
Amir.

#49 lore raw

Re: [0/4] overlayfs: Adjustments for ovl_fill_super()

Quoted 5 lines
Quoted 3 lines

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

I will queue cleanup patches 1-2,

Thanks for this positive feedback.

Quoted 2 lines

but I do not like patches 3/4 and 4/4.
I do not think that they make the code better to read or maintain.

I would appreciate if the details for such change reluctance can be clarified better.

Regards,
Markus

maintainer
#50 lore raw

Re: [0/4] overlayfs: Adjustments for ovl_fill_super()

Quoted 8 lines · Markus Elfring
Quoted 6 lines
Quoted 4 lines

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

I will queue cleanup patches 1-2,

Thanks for this positive feedback.

Sorry, these patches do not apply to master branch and patch 1 is no longer correct in master branch and the new mount api changes.

Quoted 6 lines
Quoted 4 lines

but I do not like patches 3/4 and 4/4.
I do not think that they make the code better to read or maintain.

I would appreciate if the details for such change reluctance can be clarified better.

patch 3:
I much rather a single error handling label that takes care of all the cleanups - it is harder to make mistakes and jump to the wrong label when adding new error conditions.

patch 4:
Overlayfs uses this coding style all over the place

  err = -ENOMEM;
  ofs->creator_cred = cred = prepare_creds();
  if (!cred)
      goto out_free_ofs;

I don't see the benefit in making err = -ENOMEM conditional. I don't see the style after your patch as clearly better than before.

Thanks,
Amir.

#51 lore raw

Re: [0/4] overlayfs: Adjustments for ovl_fill_super()

Quoted 10 lines
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

I will queue cleanup patches 1-2,

Thanks for this positive feedback.

Sorry, these patches do not apply to master branch and patch 1 is no longer correct in master branch and the new mount api changes.

Do you want that I adapt the linked development ideas to the current situation a bit more?

Quoted 9 lines
Quoted 4 lines
Quoted 2 lines

but I do not like patches 3/4 and 4/4.
I do not think that they make the code better to read or maintain.

I would appreciate if the details for such change reluctance can be clarified better.

patch 3:
I much rather a single error handling label that takes care of all the cleanups - it is harder to make mistakes and jump to the wrong label when adding new error conditions.

There are different coding style preferences involved.

See also:
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

Quoted 10 lines

patch 4:
Overlayfs uses this coding style all over the place

  err = -ENOMEM;
  ofs->creator_cred = cred = prepare_creds();
  if (!cred)
      goto out_free_ofs;

I don't see the benefit in making err = -ENOMEM conditional. I don't see the style after your patch as clearly better than before.

Can it be nicer to set error codes only in exceptional data processing situations?

Regards,
Markus

maintainer
#52 lore raw

Re: [0/4] overlayfs: Adjustments for ovl_fill_super()

Quoted 7 lines · Markus Elfring
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 4 lines

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00115.html

I will queue cleanup patches 1-2,

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

Thanks for this positive feedback.

Sorry, these patches do not apply to master branch and patch 1 is no longer correct in master branch and the new mount api changes.

Do you want that I adapt the linked development ideas to the current situation a bit more?

No thanks.
Patch 1 just doesn't work for the new mount api.

Quoted 11 lines
Quoted 10 lines
Quoted 5 lines
Quoted 3 lines

but I do not like patches 3/4 and 4/4.
I do not think that they make the code better to read or maintain.

I would appreciate if the details for such change reluctance can be clarified better.

patch 3:
I much rather a single error handling label that takes care of all the cleanups - it is harder to make mistakes and jump to the wrong label when adding new error conditions.

Quoted 5 lines

There are different coding style preferences involved.

See also:
https://wiki.sei.cmu.edu/confluence/display/c/MEM12-C.+Consider+using+a+goto+chain+when+leaving+a+function+on+error+when+using+and+releasing+resources

As long as coding styles are not mandatory
I prefer what we have right now.

Quoted 9 lines
Quoted 9 lines

patch 4:
Overlayfs uses this coding style all over the place

  err = -ENOMEM;
  ofs->creator_cred = cred = prepare_creds();
  if (!cred)
      goto out_free_ofs;
Quoted 5 lines
Quoted 2 lines

I don't see the benefit in making err = -ENOMEM conditional. I don't see the style after your patch as clearly better than before.

Can it be nicer to set error codes only in exceptional data processing situations?

It's a matter of taste.
I'll stay with what we have.

Thanks,
Amir.

#53 lore raw

PATCH 1/1 squashfs: Improve exception handling in squashfs_decompressor_create()

In reply to Markus Elfring

Date: Thu, 30 Mar 2023 18:03:47 +0200

The label “out” was used to jump to a kfree() call despite of the detail in the implementation of the function “squashfs_decompressor_create” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

Thus perform the following adjustments:

1. Return directly after a call of the function “kzalloc” failed at the beginning.

2. Use more appropriate labels instead.

3. Omit extra initialisations (for the variables “decomp_strm” and “err”) which became unnecessary with this refactoring.

This issue was detected by using the Coccinelle software.

---
 fs/squashfs/decompressor_multi.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)
fs/squashfs/decompressor_multi.c +10 -7
index 416c53eedbd1..0a054ba4c541 100644
@@ -62,12 +62,12 @@ static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 				void *comp_opts)
 {
 	struct squashfs_stream *stream;
-	struct decomp_stream *decomp_strm = NULL;
-	int err = -ENOMEM;
+	struct decomp_stream *decomp_strm;
+	int err;
 
 	stream = kzalloc(sizeof(*stream), GFP_KERNEL);
 	if (!stream)
-		goto out;
+		return ERR_PTR(-ENOMEM);
 
 	stream->comp_opts = comp_opts;
 	mutex_init(&stream->mutex);
@@ -81,22 +81,25 @@ static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 	 * file system works.
 	 */
 	decomp_strm = kmalloc(sizeof(*decomp_strm), GFP_KERNEL);
-	if (!decomp_strm)
-		goto out;
+	if (!decomp_strm) {
+		err = -ENOMEM;
+		goto free_stream;
+	}
 
 	decomp_strm->stream = msblk->decompressor->init(msblk,
 						stream->comp_opts);
 	if (IS_ERR(decomp_strm->stream)) {
 		err = PTR_ERR(decomp_strm->stream);
-		goto out;
+		goto free_decomp_stream;
 	}
 
 	list_add(&decomp_strm->list, &stream->strm_list);
 	stream->avail_decomp = 1;
 	return stream;
 
-out:
+free_decomp_stream:
 	kfree(decomp_strm);
+free_stream:
 	kfree(stream);
 	return ERR_PTR(err);
 }

--
2.40.0

#54 lore raw

Re: [PATCH] squashfs: Improve exception handling in squashfs_decompressor_create()

Quoted 66 lines

Date: Thu, 30 Mar 2023 18:03:47 +0200

The label “out” was used to jump to a kfree() call despite of the detail in the implementation of the function “squashfs_decompressor_create” that it was determined already that a corresponding variable contained a null pointer because of a failed memory allocation.

Thus perform the following adjustments:

1. Return directly after a call of the function “kzalloc” failed at the beginning.

2. Use more appropriate labels instead.

3. Omit extra initialisations (for the variables “decomp_strm” and “err”) which became unnecessary with this refactoring.

This issue was detected by using the Coccinelle software.

---
 fs/squashfs/decompressor_multi.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)
fs/squashfs/decompressor_multi.c +8 -6
index 416c53eedbd1..0a054ba4c541 100644
@@ -62,12 +62,12 @@ static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 				void *comp_opts)
 {
 	struct squashfs_stream *stream;
-	struct decomp_stream *decomp_strm = NULL;
-	int err = -ENOMEM;
+	struct decomp_stream *decomp_strm;
+	int err;
 
 	stream = kzalloc(sizeof(*stream), GFP_KERNEL);
 	if (!stream)
-		goto out;
+		return ERR_PTR(-ENOMEM);
 
 	stream->comp_opts = comp_opts;
 	mutex_init(&stream->mutex);
@@ -81,22 +81,25 @@ static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 	 * file system works.
 	 */
 	decomp_strm = kmalloc(sizeof(*decomp_strm), GFP_KERNEL);
-	if (!decomp_strm)
-		goto out;
+	if (!decomp_strm) {
+		err = -ENOMEM;
+		goto free_stream;
+	}
 
 	decomp_strm->stream = msblk->decompressor->init(msblk,
 						stream->comp_opts);
 	if (IS_ERR(decomp_strm->stream)) {
 		err = PTR_ERR(decomp_strm->stream);
-		goto out;
+		goto free_decomp_stream;
 	}
 
 	list_add(&decomp_strm->list, &stream->strm_list);

fs/squashfs/decompressor_multi.c static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk, · line 105

Quoted 10 lines
fs/squashfs/decompressor_multi.c +2 -1
 	stream->avail_decomp = 1;
 	return stream;
 
-out:
+free_decomp_stream:
 	kfree(decomp_strm);
+free_stream:
 	kfree(stream);
 	return ERR_PTR(err);
 }

Is this patch still in review queues?

See also:
https://lore.kernel.org/cocci/[email protected]/
https://sympa.inria.fr/sympa/arc/cocci/2023-03/msg00120.html

Regards,
Markus