2022-02-07 16:39:13

by Fabio M. De Francesco

[permalink] [raw]
Subject: [PATCH] staging: r8188eu: Use kzalloc() with GFP_ATOMIC in atomic context

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

This issue is detected by Smatch which emits the following warning:
drivers/staging/r8188eu/core/rtw_pwrctrl.c:79 ips_leave() warn: sleeping in atomic context
drivers/staging/r8188eu/core/rtw_pwrctrl.c:81 ips_leave() warn: sleeping in atomic context

The calls chain (in reverse order) is the following:

rtw_set_key()
-> ips_leave()
-> -> rtw_pwr_wakeup()
-> -> -> rtw_set_802_11_disassociate()

The disabilitation of bottom halves and the acquisition of a spinlock is in
rtw_set_802_11_disassociate().

After the changes, the post-commit hook output the following messages:

CHECK: Prefer kzalloc(sizeof(*pcmd)...) over kzalloc(sizeof(struct cmd_obj)...)
+ pcmd = kzalloc(sizeof(struct cmd_obj), GFP_ATOMIC);

CHECK: Prefer kzalloc(sizeof(*psetkeyparm)...) over kzalloc(sizeof(struct setkey_parm)...)
+ psetkeyparm = kzalloc(sizeof(struct setkey_parm), GFP_ATOMIC).

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

Fixes: 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and kzalloc()")
Fixes: 15865124feed ("staging: r8188eu: introduce new core dir for RTL8188eu driver")
Signed-off-by: Fabio M. De Francesco <[email protected]>
---
drivers/staging/r8188eu/core/rtw_mlme.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/staging/r8188eu/core/rtw_mlme.c b/drivers/staging/r8188eu/core/rtw_mlme.c
index 038bddc361c3..860835e29b79 100644
--- a/drivers/staging/r8188eu/core/rtw_mlme.c
+++ b/drivers/staging/r8188eu/core/rtw_mlme.c
@@ -1600,12 +1600,12 @@ int rtw_set_key(struct adapter *adapter, struct security_priv *psecuritypriv, in
struct mlme_priv *pmlmepriv = &adapter->mlmepriv;
int res = _SUCCESS;

- pcmd = kzalloc(sizeof(struct cmd_obj), GFP_KERNEL);
+ pcmd = kzalloc(sizeof(*pcmd), GFP_ATOMIC);
if (!pcmd) {
res = _FAIL; /* try again */
goto exit;
}
- psetkeyparm = kzalloc(sizeof(struct setkey_parm), GFP_KERNEL);
+ psetkeyparm = kzalloc(sizeof(*psetkeyparm), GFP_ATOMIC);
if (!psetkeyparm) {
kfree(pcmd);
res = _FAIL;
--
2.34.1



2022-02-08 11:28:46

by Greg Kroah-Hartman

[permalink] [raw]
Subject: Re: [PATCH] staging: r8188eu: Use kzalloc() with GFP_ATOMIC in atomic context

On Sun, Feb 06, 2022 at 11:59:43PM +0100, Fabio M. De Francesco wrote:
> Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in
> rtw_set_key(). This function is called while holding spinlocks and with
> disabled bottom halves, therefore it is not allowed to sleep. With the
> GFP_ATOMIC type flag, the allocation is high priority and cannot sleep.
>
> This issue is detected by Smatch which emits the following warning:
> drivers/staging/r8188eu/core/rtw_pwrctrl.c:79 ips_leave() warn: sleeping in atomic context
> drivers/staging/r8188eu/core/rtw_pwrctrl.c:81 ips_leave() warn: sleeping in atomic context
>
> The calls chain (in reverse order) is the following:
>
> rtw_set_key()
> -> ips_leave()
> -> -> rtw_pwr_wakeup()
> -> -> -> rtw_set_802_11_disassociate()
>
> The disabilitation of bottom halves and the acquisition of a spinlock is in
> rtw_set_802_11_disassociate().
>
> After the changes, the post-commit hook output the following messages:
>
> CHECK: Prefer kzalloc(sizeof(*pcmd)...) over kzalloc(sizeof(struct cmd_obj)...)
> + pcmd = kzalloc(sizeof(struct cmd_obj), GFP_ATOMIC);
>
> CHECK: Prefer kzalloc(sizeof(*psetkeyparm)...) over kzalloc(sizeof(struct setkey_parm)...)
> + psetkeyparm = kzalloc(sizeof(struct setkey_parm), GFP_ATOMIC).
>
> According to the above "CHECK[S]", use the preferred style in the first
> kzalloc().
>
> Fixes: 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and kzalloc()")
> Fixes: 15865124feed ("staging: r8188eu: introduce new core dir for RTL8188eu driver")
> Signed-off-by: Fabio M. De Francesco <[email protected]>
> ---
> drivers/staging/r8188eu/core/rtw_mlme.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/staging/r8188eu/core/rtw_mlme.c b/drivers/staging/r8188eu/core/rtw_mlme.c
> index 038bddc361c3..860835e29b79 100644
> --- a/drivers/staging/r8188eu/core/rtw_mlme.c
> +++ b/drivers/staging/r8188eu/core/rtw_mlme.c
> @@ -1600,12 +1600,12 @@ int rtw_set_key(struct adapter *adapter, struct security_priv *psecuritypriv, in
> struct mlme_priv *pmlmepriv = &adapter->mlmepriv;
> int res = _SUCCESS;
>
> - pcmd = kzalloc(sizeof(struct cmd_obj), GFP_KERNEL);
> + pcmd = kzalloc(sizeof(*pcmd), GFP_ATOMIC);
> if (!pcmd) {
> res = _FAIL; /* try again */
> goto exit;
> }
> - psetkeyparm = kzalloc(sizeof(struct setkey_parm), GFP_KERNEL);
> + psetkeyparm = kzalloc(sizeof(*psetkeyparm), GFP_ATOMIC);

You are making two different changes here. Please do the first patch to
change the sizeof() change to fix up checkpatch, and then the second one
for the GFP_ATOMIC change so that if there is a problem with either of
them we can only revert the offending change.

thanks,

greg k-h

2022-02-08 15:32:33

by Fabio M. De Francesco

[permalink] [raw]
Subject: Re: [PATCH] staging: r8188eu: Use kzalloc() with GFP_ATOMIC in atomic context

On marted? 8 febbraio 2022 10:33:04 CET Greg Kroah-Hartman wrote:
> On Sun, Feb 06, 2022 at 11:59:43PM +0100, Fabio M. De Francesco wrote:
> > Use the GFP_ATOMIC flag of kzalloc() with two memory allocation in
> > rtw_set_key(). This function is called while holding spinlocks and with
> > disabled bottom halves, therefore it is not allowed to sleep. With the
> > GFP_ATOMIC type flag, the allocation is high priority and cannot sleep.
> >
> > This issue is detected by Smatch which emits the following warning:
> > drivers/staging/r8188eu/core/rtw_pwrctrl.c:79 ips_leave() warn: sleeping in atomic context
> > drivers/staging/r8188eu/core/rtw_pwrctrl.c:81 ips_leave() warn: sleeping in atomic context
> >
> > The calls chain (in reverse order) is the following:
> >
> > rtw_set_key()
> > -> ips_leave()
> > -> -> rtw_pwr_wakeup()
> > -> -> -> rtw_set_802_11_disassociate()
> >
> > The disable of bottom halves and the acquisition of a spinlock is in
> > rtw_set_802_11_disassociate().
> >
> > After the changes, the post-commit hook output the following messages:
> >
> > CHECK: Prefer kzalloc(sizeof(*pcmd)...) over kzalloc(sizeof(struct cmd_obj)...)
> > + pcmd = kzalloc(sizeof(struct cmd_obj), GFP_ATOMIC);
> >
> > CHECK: Prefer kzalloc(sizeof(*psetkeyparm)...) over kzalloc(sizeof(struct setkey_parm)...)
> > + psetkeyparm = kzalloc(sizeof(struct setkey_parm), GFP_ATOMIC).
> >
> > According to the above "CHECK[S]", use the preferred style in the first
> > kzalloc().
> >
> > Fixes: 79f712ea994d ("staging: r8188eu: Remove wrappers for kalloc() and kzalloc()")
> > Fixes: 15865124feed ("staging: r8188eu: introduce new core dir for RTL8188eu driver")
> > Signed-off-by: Fabio M. De Francesco <[email protected]>
> > ---
> > drivers/staging/r8188eu/core/rtw_mlme.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > [...]
>
> You are making two different changes here. Please do the first patch to
> change the sizeof() change to fix up checkpatch, and then the second one
> for the GFP_ATOMIC change so that if there is a problem with either of
> them we can only revert the offending change.
>
OK, thanks for your reply. I'm about to split this patch in two steps as
you require.

In the while I've noticed that, after git-reset HARD^, Smatch now points directly
to the kzalloc() calls and emits a different output:

"drivers/staging/r8188eu/core/rtw_mlme.c:1603 rtw_set_key() warn: sleeping in
atomic context
CHECK drivers/staging/r8188eu/core/rtw_mlme_ext.c".

So now it complains specifically about the first of the two kzalloc[s]() in
rtw_set_key(). Before sending v2, I'd like to check why when I made v1 it pointed
to the lines that call rtw_set_key(). You'll see that change in the commit message
of v2.

Thanks,

Fabio
>
> thanks,
>
> greg k-h
>
>