Skip to contentSkip to conversation

staging: r8188eu: Use kzalloc() with GFP_ATOMIC in atomic context

PATCH In review Fabio M. De Francesco 3 participants 12 messages Nov 1, 2021 – Nov 7, 2021 Staging

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 staging: r8188eu: Use kzalloc() with GFP_ATOMIC in atomic context

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding spinlocks, therefore it is not allowed to sleep. With the GFP_ATOMIC type flag, the allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848 report_del_sta_event()
warn: sleeping in atomic context".

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and kzalloc()")

---

v2->v3: Add the "Fixes:" tag, as requested by Greg Kroah-Hartman.

v1->v2: Fix an error that I introduced with an incorrect copy-paste
of the sizeof() operator.

 drivers/staging/r8188eu/core/rtw_mlme_ext.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/staging/r8188eu/core/rtw_mlme_ext.c +2 -2
index 55c3d4a6faeb..315902682292 100644
@@ -6845,12 +6845,12 @@ void report_del_sta_event(struct adapter *padapter, unsigned char *MacAddr, unsi
 	struct mlme_ext_priv		*pmlmeext = &padapter->mlmeextpriv;
 	struct cmd_priv *pcmdpriv = &padapter->cmdpriv;
 
-	pcmd_obj = kzalloc(sizeof(struct cmd_obj), GFP_KERNEL);
+	pcmd_obj = kzalloc(sizeof(*pcmd_obj), GFP_ATOMIC);
 	if (!pcmd_obj)
 		return;
 
 	cmdsz = (sizeof(struct stadel_event) + sizeof(struct C2HEvent_Header));
-	pevtcmd = kzalloc(cmdsz, GFP_KERNEL);
+	pevtcmd = kzalloc(cmdsz, GFP_ATOMIC);
 	if (!pevtcmd) {
 		kfree(pcmd_obj);
 		return;
Signature · 1 line
2.33.1
#1 lore raw
Quoted 9 lines · Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding spinlocks, therefore it is not allowed to sleep. With the GFP_ATOMIC type flag, the allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848 report_del_sta_event()
warn: sleeping in atomic context".

Quoted 8 lines

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and kzalloc()")

This is not the correct Fixes tag. The original allocation wrappers checked in_interrupt() they did not check in_atomic() so they had same bug. The correct tag is:

Fixes 15865124feed ("staging: r8188eu: introduce new core dir for RTL8188eu driver")

regards,
dan carpenter

#2 lore raw
Quoted 10 lines · CET Dan Carpenter
Quoted 9 lines · Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding spinlocks, therefore it is not allowed to sleep. With the GFP_ATOMIC type flag, the allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848 report_del_sta_event()
warn: sleeping in atomic context".

Quoted 8 lines
Quoted 8 lines

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and

kzalloc()")

Quoted 6 lines

This is not the correct Fixes tag. The original allocation wrappers checked in_interrupt() they did not check in_atomic() so they had same bug. The correct tag is:

Fixes 15865124feed ("staging: r8188eu: introduce new core dir for

RTL8188eu driver")

Quoted 3 lines

regards,
dan carpenter

Hello Dan,

I'm sorry but I surely missing something, therefore, before making changes I need to understand this subject a little better. Let me explain what I am missing...

The two kzalloc() in report_del_sta_event() are called while spinlocks are held and bottom halves are disabled by spin_lock_bh(). If I remember it correctly spin_lock_bh() finally calls __local_bh_disable_ip() to disable bottom halves on local CPU before actually acquiring the lock.

This is the code and inline documentation of in_interrupt():

/* in_interrupt() - We're in NMI,IRQ,SoftIRQ context or have BH disabled" */
#define irq_count() (nmi_count() | hardirq_count() | softirq_count())
#define in_interrupt() (irq_count())

And this is the code and inline documentation of in_atomic():

"/*
* Are we running in atomic context? WARNING: this macro cannot
* always detect atomic context; in particular, it cannot know about
* held spinlocks in non-preemptible kernels. Thus it should not be
* used in the general case to determine whether sleeping is possible.
* Do not use in_atomic() in driver code.
*/
#define in_atomic() (preempt_count() != 0)

To summarize, I think that using in_interrupt() in the old wrappers was the wiser choice. Therefore this patch fixes 79f712ea994d ("staging: r8188eu:
Remove wrappers for kalloc() and kzalloc()").

I know that I have so little experience that I shouldn't even discuss this topics. However, I would appreciate if you may explain with some more details why in_atomic() should have been preferred over in_interrupt() in the old wrappers that were removed with commit 79f712ea994d.

Thank you very much in advance,

Fabio M. De Francesco

#3 lore raw

Oh yeah, you're right. It never *just* does spinlocks (as stated in the commit message btw), it does spin_lock_bh() which bumps the soft IRQ count.

Quoted 2 lines

To summarize, I think that using in_interrupt() in the old wrappers was the wiser choice.

"Wiser" is not the right word. The wrappers were always stupid, but I guess they did work in this case so the fixes tag is correct.

regards,
dan carpenter

#4 lore raw
Quoted 3 lines · CET Dan Carpenter

Oh yeah, you're right. It never *just* does spinlocks (as stated in the commit message btw), it does spin_lock_bh() which bumps the soft IRQ count.

Thank you very much for checking and confirming.

Quoted 1 line
Quoted 1 line

To summarize, I think that using in_interrupt() in the old wrappers was

the

Quoted 4 lines
Quoted 1 line

wiser choice.

"Wiser" is not the right word. The wrappers were always stupid, but I guess they did work in this case so the fixes tag is correct.

Ah, sorry. I was not able to express my thought properly :(

I agree with you that the wrappers were a not a good idea and Larry did well in removing them. Furthermore, I think that delegating the choice to use GFP_KERNEL vs. GFP_ATOMIC depending on the return from in_interrupt() is very bad design and it adds sensible overhead.

I used "wiser" is a stricter sense. I meant that, if wrappers were needed (but they were not), in_interrupt() is "wiser" than "in_atomic()".

Regards,

Fabio M. De Francesco

#5 lore raw
In reply to Fabio M. De Francesco
Quoted 9 lines · CET Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding spinlocks, therefore it is not allowed to sleep. With the GFP_ATOMIC type flag, the allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848 report_del_sta_event()
warn: sleeping in atomic context".

Quoted 8 lines

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and

kzalloc()")

Quoted 10 lines

---

v2->v3: Add the "Fixes:" tag, as requested by Greg Kroah-Hartman.

v1->v2: Fix an error that I introduced with an incorrect copy-paste
of the sizeof() operator.

 drivers/staging/r8188eu/core/rtw_mlme_ext.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Hello Greg,

I've noticed that you have already applied recent changes to drivers/staging up to the patches of November 6th, but my patch is not among them.

This patch has already been acked by Larry and I'm not sure if I should send a v4 with his "Acked-by" tag or if you can add it by yourself when applying to your tree.

Please let me know if there is something that prevents this patch to be applied. I have no problem in changing / adding whatever it is needed.

Thanks,

Fabio M. De Francesco

#6 lore raw
Quoted 27 lines · Fabio M. De Francesco
Quoted 9 lines · CET Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding spinlocks, therefore it is not allowed to sleep. With the GFP_ATOMIC type flag, the allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848 report_del_sta_event()
warn: sleeping in atomic context".

Quoted 8 lines

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and

kzalloc()")

Quoted 8 lines

---

v2->v3: Add the "Fixes:" tag, as requested by Greg Kroah-Hartman.

v1->v2: Fix an error that I introduced with an incorrect copy-paste
of the sizeof() operator.

Quoted 7 lines
Quoted 2 lines
 drivers/staging/r8188eu/core/rtw_mlme_ext.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Hello Greg,

I've noticed that you have already applied recent changes to drivers/staging up to the patches of November 6th, but my patch is not among them.

I have applied patches that are not targeted for 5.16-final, yes.

Quoted 6 lines

This patch has already been acked by Larry and I'm not sure if I should send a v4 with his "Acked-by" tag or if you can add it by yourself when applying to your tree.

Please let me know if there is something that prevents this patch to be applied. I have no problem in changing / adding whatever it is needed.

Nothing needs to be done, I am waiting for 5.16-rc1 to be released before I pick up this patch, and others that will be targeted for 5.16-final. Only then will I queue them up, as the automated email you should have gotten when you submitted the patch said would happen.

Just relax, there is no rush here :)

thanks,

greg k-h

#7 lore raw
Quoted 4 lines · CET Greg Kroah-Hartman
Quoted 3 lines · Fabio M. De Francesco
Quoted 2 lines · CET Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding

spinlocks,

Quoted 1 line
Quoted 1 line
Quoted 1 line

therefore it is not allowed to sleep. With the GFP_ATOMIC type flag,

the

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848

report_del_sta_event()

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

warn: sleeping in atomic context".

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc()

and

Quoted 9 lines
Quoted 8 lines

kzalloc()")

Quoted 4 lines

---

[...]

Please let me know if there is something that prevents this patch to be applied. I have no problem in changing / adding whatever it is needed.

Quoted 7 lines

Nothing needs to be done, I am waiting for 5.16-rc1 to be released before I pick up this patch, and others that will be targeted for 5.16-final. Only then will I queue them up, as the automated email you should have gotten when you submitted the patch said would happen.

Just relax, there is no rush here :)

Oh, sorry Greg. There must be something that I haven't understand about the development process... :(

Obviously I agree that there is no rush here :)

As I said, this morning I read git log and saw patches that seemed more recent; thus I thought that was the case to ask. I just (wrongly) thought that the v3 of the patch got unnoticed or dropped because of some requests that I had missed.

Thanks for the explanation,

Fabio

Quoted 4 lines

thanks,

greg k-h

#8 lore raw
Quoted 44 lines · Fabio M. De Francesco
Quoted 4 lines · CET Greg Kroah-Hartman
Quoted 3 lines · Fabio M. De Francesco
Quoted 2 lines · CET Fabio M. De Francesco

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding

spinlocks,

Quoted 1 line
Quoted 1 line
Quoted 1 line

therefore it is not allowed to sleep. With the GFP_ATOMIC type flag,

the

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848

report_del_sta_event()

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

warn: sleeping in atomic context".

After the change, the post-commit hook output the following message:
"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the first kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc()

and

Quoted 9 lines
Quoted 8 lines

kzalloc()")

Quoted 4 lines

---

[...]

Please let me know if there is something that prevents this patch to be applied. I have no problem in changing / adding whatever it is needed.

Quoted 7 lines

Nothing needs to be done, I am waiting for 5.16-rc1 to be released before I pick up this patch, and others that will be targeted for 5.16-final. Only then will I queue them up, as the automated email you should have gotten when you submitted the patch said would happen.

Just relax, there is no rush here :)

Oh, sorry Greg. There must be something that I haven't understand about the development process... :(

Quoted 6 lines

Obviously I agree that there is no rush here :)

As I said, this morning I read git log and saw patches that seemed more recent; thus I thought that was the case to ask. I just (wrongly) thought that the v3 of the patch got unnoticed or dropped because of some requests that I had missed.

Be sure to notice what branch commits are being applied to. There are different branches for a reason :)

thanks,

greg k-h

#9 lore raw
Quoted 4 lines · CET Greg Kroah-Hartman
Quoted 3 lines · Fabio M. De Francesco
Quoted 2 lines · CET Greg Kroah-Hartman
Quoted 1 line · Fabio M. De Francesco

On Monday, November 1, 2021 8:18:47 PM CET Fabio M. De Francesco

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding

spinlocks,

Quoted 1 line
Quoted 1 line
Quoted 1 line

therefore it is not allowed to sleep. With the GFP_ATOMIC type

flag,

Quoted 9 lines
Quoted 9 lines

the

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848

report_del_sta_event()

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

warn: sleeping in atomic context".

After the change, the post-commit hook output the following

message:

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over
kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the

first

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for

kalloc()

Quoted 8 lines
Quoted 8 lines

and

Quoted 7 lines
Quoted 7 lines

kzalloc()")

Quoted 4 lines

---

[...]

Please let me know if there is something that prevents this patch to

be

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

applied. I have no problem in changing / adding whatever it is

needed.

Quoted 10 lines
Quoted 10 lines
Quoted 8 lines

Nothing needs to be done, I am waiting for 5.16-rc1 to be released before I pick up this patch, and others that will be targeted for 5.16-final. Only then will I queue them up, as the automated email you should have gotten when you submitted the patch said would happen.

Just relax, there is no rush here :)

Oh, sorry Greg. There must be something that I haven't understand about

the

Quoted 7 lines
Quoted 7 lines

development process... :(

Obviously I agree that there is no rush here :)

As I said, this morning I read git log and saw patches that seemed more recent; thus I thought that was the case to ask. I just (wrongly) thought that the v3 of the patch got unnoticed or dropped because of some

requests

Quoted 5 lines
Quoted 1 line

that I had missed.

Be sure to notice what branch commits are being applied to. There are different branches for a reason :)

This is what confuses me:

--- git-log output ---

commit 8a893759d0075ea9556abcf86a4826d9865ba4bf (origin/staging-testing)

Author: Phillip Potter <[email protected]>
Date:   Sat Nov 6 23:16:36 2021 +0000

staging: r8188eu: remove MSG_88E macro

--- end of git-log output ---

Aside from the "Date" field, I know that this patch has been sent to the list during the last night and that it goes to the same branch (staging-testing) to which my patch should go. I know I'm still missing something, but I cannot understand what it is... :(

Anyway, never mind. I don't want to bother you with this silly questions :)

Again, thanks,

Fabio

#10 lore raw
Quoted 73 lines · Fabio M. De Francesco
Quoted 4 lines · CET Greg Kroah-Hartman
Quoted 3 lines · Fabio M. De Francesco
Quoted 2 lines · CET Greg Kroah-Hartman
Quoted 1 line · Fabio M. De Francesco

On Monday, November 1, 2021 8:18:47 PM CET Fabio M. De Francesco

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in report_del_sta_event(). This function is called while holding

spinlocks,

Quoted 1 line
Quoted 1 line
Quoted 1 line

therefore it is not allowed to sleep. With the GFP_ATOMIC type

flag,

Quoted 9 lines
Quoted 9 lines

the

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

allocation is high priority and must not sleep.

This issue is detected by Smatch which emits the following warning:
"drivers/staging/r8188eu/core/rtw_mlme_ext.c:6848

report_del_sta_event()

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

warn: sleeping in atomic context".

After the change, the post-commit hook output the following

message:

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

"CHECK: Prefer kzalloc(sizeof(*pcmd_obj)...) over
kzalloc(sizeof(struct cmd_obj)...)".

According to the above "CHECK", use the preferred style in the

first

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

kzalloc().

Fixes 79f712ea994d ("staging: r8188eu: Remove wrappers for

kalloc()

Quoted 8 lines
Quoted 8 lines

and

Quoted 7 lines
Quoted 7 lines

kzalloc()")

Quoted 4 lines

---

[...]

Please let me know if there is something that prevents this patch to

be

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

applied. I have no problem in changing / adding whatever it is

needed.

Quoted 10 lines
Quoted 10 lines
Quoted 8 lines

Nothing needs to be done, I am waiting for 5.16-rc1 to be released before I pick up this patch, and others that will be targeted for 5.16-final. Only then will I queue them up, as the automated email you should have gotten when you submitted the patch said would happen.

Just relax, there is no rush here :)

Oh, sorry Greg. There must be something that I haven't understand about

the

Quoted 7 lines
Quoted 7 lines

development process... :(

Obviously I agree that there is no rush here :)

As I said, this morning I read git log and saw patches that seemed more recent; thus I thought that was the case to ask. I just (wrongly) thought that the v3 of the patch got unnoticed or dropped because of some

requests

Quoted 5 lines
Quoted 1 line

that I had missed.

Be sure to notice what branch commits are being applied to. There are different branches for a reason :)

This is what confuses me:

--- git-log output ---

commit 8a893759d0075ea9556abcf86a4826d9865ba4bf (origin/staging-testing)

Author: Phillip Potter <[email protected]>
Date:   Sat Nov 6 23:16:36 2021 +0000
Quoted 8 lines

staging: r8188eu: remove MSG_88E macro

--- end of git-log output ---

Aside from the "Date" field, I know that this patch has been sent to the list during the last night and that it goes to the same branch (staging-testing) to which my patch should go. I know I'm still missing something, but I cannot understand what it is... :(

No, your change will go to the staging-linus branch, as it needs to go into 5.16-final and get sent to Linus much sooner than 5.17-rc1, which is where things are being queued up in the staging-testing branch at the moment.

hope this helps,

greg k-h

#11 lore raw
Quoted 4 lines · CET Greg Kroah-Hartman

No, your change will go to the staging-linus branch, as it needs to go into 5.16-final and get sent to Linus much sooner than 5.17-rc1, which is where things are being queued up in the staging-testing branch at the moment.

Oh, I didn't even remotely guess that this kinds of patches usually go to the staging-linus branch so they get sent to Linus much sooner.

Thank you so much for your patience and for taking the time to explain the workflow to me.

Fabio

Quoted 5 lines

hope this helps,

greg k-h