This patch reduces the kernel size by removing error messages that duplicate
the normal OOM message.
A simplified version of the semantic patch that finds this problem is as
follows: (http://coccinelle.lip6.fr)
@@
identifier f,print,l;
expression e;
constant char[] c;
@@
e = \(kzalloc\|kmalloc\|devm_kzalloc\|devm_kmalloc\)(...);
if (e == NULL) {
<+...
- print(...,c,...);
... when any
(
goto l;
|
return ...;
)
...+> }
Signed-off-by: Quentin Lambert <[email protected]>
---
drivers/staging/rtl8723au/hal/rtl8723a_cmd.c | 8 ++------
drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c | 8 ++------
drivers/staging/rtl8723au/hal/usb_ops_linux.c | 6 +-----
drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c | 5 -----
4 files changed, 5 insertions(+), 22 deletions(-)
diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
index 7b56411..38f5b7f 100644
--- a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
+++ b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
@@ -462,10 +462,8 @@ static void SetFwRsvdPagePkt(struct rtw_adapter *padapter, bool bDLFinished)
DBG_8723A("%s\n", __func__);
ReservedPagePacket = kzalloc(1000, GFP_KERNEL);
- if (ReservedPagePacket == NULL) {
- DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
+ if (ReservedPagePacket == NULL)
return;
- }
pHalData = GET_HAL_DATA(padapter);
pxmitpriv = &padapter->xmitpriv;
@@ -669,10 +667,8 @@ static void SetFwRsvdPagePkt_BTCoex(struct rtw_adapter *padapter)
DBG_8723A("+%s\n", __func__);
ReservedPagePacket = kzalloc(1024, GFP_KERNEL);
- if (ReservedPagePacket == NULL) {
- DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
+ if (ReservedPagePacket == NULL)
return;
- }
pHalData = GET_HAL_DATA(padapter);
pxmitpriv = &padapter->xmitpriv;
diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
index a5eadd4..6d50b09 100644
--- a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
+++ b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
@@ -401,10 +401,8 @@ hal_ReadEFuse_WiFi(struct rtw_adapter *padapter,
}
efuseTbl = kmalloc(EFUSE_MAP_LEN_8723A, GFP_KERNEL);
- if (efuseTbl == NULL) {
- DBG_8723A("%s: alloc efuseTbl fail!\n", __func__);
+ if (efuseTbl == NULL)
return;
- }
/* 0xff will be efuse default value instead of 0x00. */
memset(efuseTbl, 0xFF, EFUSE_MAP_LEN_8723A);
@@ -494,10 +492,8 @@ hal_ReadEFuse_BT(struct rtw_adapter *padapter,
}
efuseTbl = kmalloc(EFUSE_BT_MAP_LEN, GFP_KERNEL);
- if (efuseTbl == NULL) {
- DBG_8723A("%s: efuseTbl malloc fail!\n", __func__);
+ if (efuseTbl == NULL)
return;
- }
/* 0xff will be efuse default value instead of 0x00. */
memset(efuseTbl, 0xFF, EFUSE_BT_MAP_LEN);
diff --git a/drivers/staging/rtl8723au/hal/usb_ops_linux.c b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
index a6d16ad..f1e9202 100644
--- a/drivers/staging/rtl8723au/hal/usb_ops_linux.c
+++ b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
@@ -256,12 +256,8 @@ static void usb_read_interrupt_complete(struct urb *purb)
c2w = kmalloc(sizeof(struct evt_work),
GFP_ATOMIC);
- if (!c2w) {
- printk(KERN_WARNING "%s: unable to "
- "allocate work buffer\n",
- __func__);
+ if (!c2w)
goto urb_submit;
- }
c2w->adapter = padapter;
INIT_WORK(&c2w->work, rtw_evt_work);
diff --git a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
index 537bd82..40bdf4b 100644
--- a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
+++ b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
@@ -1327,8 +1327,6 @@ static int rtw_cfg80211_set_probe_req_wpsp2pie(struct rtw_adapter *padapter,
pmlmepriv->wps_probe_req_ie = kmemdup(wps_ie, wps_ie[1],
GFP_KERNEL);
if (pmlmepriv->wps_probe_req_ie == NULL) {
- DBG_8723A("%s()-%d: kmalloc() ERROR!\n",
- __func__, __LINE__);
return -EINVAL;
}
pmlmepriv->wps_probe_req_ie_len = wps_ie[1];
@@ -2619,8 +2617,6 @@ static int rtw_cfg80211_add_monitor_if(struct rtw_adapter *padapter, char *name,
/* wdev */
mon_wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
if (!mon_wdev) {
- DBG_8723A("%s(%s): allocate mon_wdev fail\n", __func__,
- padapter->pnetdev->name);
ret = -ENOMEM;
goto out;
}
@@ -3275,7 +3271,6 @@ int rtw_wdev_alloc(struct rtw_adapter *padapter, struct device *dev)
/* wdev */
wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
if (!wdev) {
- DBG_8723A("Couldn't allocate wireless device\n");
ret = -ENOMEM;
goto free_wiphy;
}
--
1.9.1
Joe Perches <[email protected]> writes:
> On Fri, 2015-03-06 at 11:08 -0500, Jes Sorensen wrote:
>> Julia Lawall <[email protected]> writes:
>> > On Fri, 6 Mar 2015, Jes Sorensen wrote:
>> >> Quentin Lambert <[email protected]> writes:
>> >> > This patch reduces the kernel size by removing error messages
>> >> > that duplicate
>> >> > the normal OOM message.
>> >> > A simplified version of the semantic patch that finds this problem is as
>> >> > follows: (http://coccinelle.lip6.fr)
>> >> This patch removes useful warnings about what allocation failed. The
>> >> messages removed are NOT duplicate!
>> > Is it really the case that the information can't be reconstructed from the
>> > information generated by kmalloc on failure? To my understanding there is
>> > a stack trace, and from scanning through the changes I see only one change
>> > per function, so perhaps the stack trace already makes it clear where the
>> > problem occurred?
>> It may be possible to backtrack, but this change just makes it harder.
>> There are tons of real issues to fix in this driver, this patch just
>> increases the risk of patch conflicts for no real gain.
>
> Making the allocation less likely to fail for
> low memory systems is a gain.
>
> The allocation failures themselves are low
> likelihood events. Determining which specific
> memory allocation failure occurred has near
> nil value.
Joe,
That is bologna, knowing which allocation failed has a lot of value, it
allows the developer to go back and look at the allocation sizes,
parameters applied etc.
This is a classic case of blindly applied script 'fixes' causing more
harm than good.
Jes
On Fri, 2015-03-06 at 14:43 -0500, Jes Sorensen wrote:
> Joe Perches <[email protected]> writes:
> > On Fri, 2015-03-06 at 11:08 -0500, Jes Sorensen wrote:
> >> Julia Lawall <[email protected]> writes:
> >> > On Fri, 6 Mar 2015, Jes Sorensen wrote:
> >> >> Quentin Lambert <[email protected]> writes:
> >> >> > This patch reduces the kernel size by removing error messages
> >> >> > that duplicate
> >> >> > the normal OOM message.
> >> >> > A simplified version of the semantic patch that finds this problem is as
> >> >> > follows: (http://coccinelle.lip6.fr)
> >> >> This patch removes useful warnings about what allocation failed. The
> >> >> messages removed are NOT duplicate!
> >> > Is it really the case that the information can't be reconstructed from the
> >> > information generated by kmalloc on failure? To my understanding there is
> >> > a stack trace, and from scanning through the changes I see only one change
> >> > per function, so perhaps the stack trace already makes it clear where the
> >> > problem occurred?
> >> It may be possible to backtrack, but this change just makes it harder.
> >> There are tons of real issues to fix in this driver, this patch just
> >> increases the risk of patch conflicts for no real gain.
> >
> > Making the allocation less likely to fail for
> > low memory systems is a gain.
> >
> > The allocation failures themselves are low
> > likelihood events. Determining which specific
> > memory allocation failure occurred has near
> > nil value.
>
> Joe,
Jes,
> That is bologna,
We disagree, and I rather like minced meat sausages.
> knowing which allocation failed has a lot of value, it
> allows the developer to go back and look at the allocation sizes,
> parameters applied etc.
Likely enough of this is emitted by the generic OOM
message and stack dump.
> This is a classic case of blindly applied script 'fixes' causing more
> harm than good.
cheesy tomato/tomahto. Goes well with baloney.
cheers, Joe
Julia Lawall <[email protected]> writes:
> On Fri, 6 Mar 2015, Jes Sorensen wrote:
>
>> Quentin Lambert <[email protected]> writes:
>> > This patch reduces the kernel size by removing error messages that duplicate
>> > the normal OOM message.
>> >
>> > A simplified version of the semantic patch that finds this problem is as
>> > follows: (http://coccinelle.lip6.fr)
>>
>> This patch removes useful warnings about what allocation failed. The
>> messages removed are NOT duplicate!
>
> Is it really the case that the information can't be reconstructed from the
> information generated by kmalloc on failure? To my understanding there is
> a stack trace, and from scanning through the changes I see only one change
> per function, so perhaps the stack trace already makes it clear where the
> problem occurred?
It may be possible to backtrack, but this change just makes it harder.
There are tons of real issues to fix in this driver, this patch just
increases the risk of patch conflicts for no real gain.
Jes
On Fri, 6 Mar 2015, Jes Sorensen wrote:
> Quentin Lambert <[email protected]> writes:
> > This patch reduces the kernel size by removing error messages that duplicate
> > the normal OOM message.
> >
> > A simplified version of the semantic patch that finds this problem is as
> > follows: (http://coccinelle.lip6.fr)
>
> This patch removes useful warnings about what allocation failed. The
> messages removed are NOT duplicate!
Is it really the case that the information can't be reconstructed from the
information generated by kmalloc on failure? To my understanding there is
a stack trace, and from scanning through the changes I see only one change
per function, so perhaps the stack trace already makes it clear where the
problem occurred?
julia
>
> NACK
>
> Jes
>
> >
> > @@
> > identifier f,print,l;
> > expression e;
> > constant char[] c;
> > @@
> >
> > e = \(kzalloc\|kmalloc\|devm_kzalloc\|devm_kmalloc\)(...);
> > if (e == NULL) {
> > <+...
> > - print(...,c,...);
> > ... when any
> > (
> > goto l;
> > |
> > return ...;
> > )
> > ...+> }
> >
> > Signed-off-by: Quentin Lambert <[email protected]>
> > ---
> > drivers/staging/rtl8723au/hal/rtl8723a_cmd.c | 8 ++------
> > drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c | 8 ++------
> > drivers/staging/rtl8723au/hal/usb_ops_linux.c | 6 +-----
> > drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c | 5 -----
> > 4 files changed, 5 insertions(+), 22 deletions(-)
> >
> > diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> > index 7b56411..38f5b7f 100644
> > --- a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> > +++ b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> > @@ -462,10 +462,8 @@ static void SetFwRsvdPagePkt(struct rtw_adapter *padapter, bool bDLFinished)
> > DBG_8723A("%s\n", __func__);
> >
> > ReservedPagePacket = kzalloc(1000, GFP_KERNEL);
> > - if (ReservedPagePacket == NULL) {
> > - DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
> > + if (ReservedPagePacket == NULL)
> > return;
> > - }
> >
> > pHalData = GET_HAL_DATA(padapter);
> > pxmitpriv = &padapter->xmitpriv;
> > @@ -669,10 +667,8 @@ static void SetFwRsvdPagePkt_BTCoex(struct rtw_adapter *padapter)
> > DBG_8723A("+%s\n", __func__);
> >
> > ReservedPagePacket = kzalloc(1024, GFP_KERNEL);
> > - if (ReservedPagePacket == NULL) {
> > - DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
> > + if (ReservedPagePacket == NULL)
> > return;
> > - }
> >
> > pHalData = GET_HAL_DATA(padapter);
> > pxmitpriv = &padapter->xmitpriv;
> > diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> > index a5eadd4..6d50b09 100644
> > --- a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> > +++ b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> > @@ -401,10 +401,8 @@ hal_ReadEFuse_WiFi(struct rtw_adapter *padapter,
> > }
> >
> > efuseTbl = kmalloc(EFUSE_MAP_LEN_8723A, GFP_KERNEL);
> > - if (efuseTbl == NULL) {
> > - DBG_8723A("%s: alloc efuseTbl fail!\n", __func__);
> > + if (efuseTbl == NULL)
> > return;
> > - }
> > /* 0xff will be efuse default value instead of 0x00. */
> > memset(efuseTbl, 0xFF, EFUSE_MAP_LEN_8723A);
> >
> > @@ -494,10 +492,8 @@ hal_ReadEFuse_BT(struct rtw_adapter *padapter,
> > }
> >
> > efuseTbl = kmalloc(EFUSE_BT_MAP_LEN, GFP_KERNEL);
> > - if (efuseTbl == NULL) {
> > - DBG_8723A("%s: efuseTbl malloc fail!\n", __func__);
> > + if (efuseTbl == NULL)
> > return;
> > - }
> > /* 0xff will be efuse default value instead of 0x00. */
> > memset(efuseTbl, 0xFF, EFUSE_BT_MAP_LEN);
> >
> > diff --git a/drivers/staging/rtl8723au/hal/usb_ops_linux.c b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> > index a6d16ad..f1e9202 100644
> > --- a/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> > +++ b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> > @@ -256,12 +256,8 @@ static void usb_read_interrupt_complete(struct urb *purb)
> > c2w = kmalloc(sizeof(struct evt_work),
> > GFP_ATOMIC);
> >
> > - if (!c2w) {
> > - printk(KERN_WARNING "%s: unable to "
> > - "allocate work buffer\n",
> > - __func__);
> > + if (!c2w)
> > goto urb_submit;
> > - }
> >
> > c2w->adapter = padapter;
> > INIT_WORK(&c2w->work, rtw_evt_work);
> > diff --git a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> > index 537bd82..40bdf4b 100644
> > --- a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> > +++ b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> > @@ -1327,8 +1327,6 @@ static int rtw_cfg80211_set_probe_req_wpsp2pie(struct rtw_adapter *padapter,
> > pmlmepriv->wps_probe_req_ie = kmemdup(wps_ie, wps_ie[1],
> > GFP_KERNEL);
> > if (pmlmepriv->wps_probe_req_ie == NULL) {
> > - DBG_8723A("%s()-%d: kmalloc() ERROR!\n",
> > - __func__, __LINE__);
> > return -EINVAL;
> > }
> > pmlmepriv->wps_probe_req_ie_len = wps_ie[1];
> > @@ -2619,8 +2617,6 @@ static int rtw_cfg80211_add_monitor_if(struct rtw_adapter *padapter, char *name,
> > /* wdev */
> > mon_wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
> > if (!mon_wdev) {
> > - DBG_8723A("%s(%s): allocate mon_wdev fail\n", __func__,
> > - padapter->pnetdev->name);
> > ret = -ENOMEM;
> > goto out;
> > }
> > @@ -3275,7 +3271,6 @@ int rtw_wdev_alloc(struct rtw_adapter *padapter, struct device *dev)
> > /* wdev */
> > wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
> > if (!wdev) {
> > - DBG_8723A("Couldn't allocate wireless device\n");
> > ret = -ENOMEM;
> > goto free_wiphy;
> > }
> --
> To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
Quentin Lambert <[email protected]> writes:
> This patch reduces the kernel size by removing error messages that duplicate
> the normal OOM message.
>
> A simplified version of the semantic patch that finds this problem is as
> follows: (http://coccinelle.lip6.fr)
This patch removes useful warnings about what allocation failed. The
messages removed are NOT duplicate!
NACK
Jes
>
> @@
> identifier f,print,l;
> expression e;
> constant char[] c;
> @@
>
> e = \(kzalloc\|kmalloc\|devm_kzalloc\|devm_kmalloc\)(...);
> if (e == NULL) {
> <+...
> - print(...,c,...);
> ... when any
> (
> goto l;
> |
> return ...;
> )
> ...+> }
>
> Signed-off-by: Quentin Lambert <[email protected]>
> ---
> drivers/staging/rtl8723au/hal/rtl8723a_cmd.c | 8 ++------
> drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c | 8 ++------
> drivers/staging/rtl8723au/hal/usb_ops_linux.c | 6 +-----
> drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c | 5 -----
> 4 files changed, 5 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> index 7b56411..38f5b7f 100644
> --- a/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> +++ b/drivers/staging/rtl8723au/hal/rtl8723a_cmd.c
> @@ -462,10 +462,8 @@ static void SetFwRsvdPagePkt(struct rtw_adapter *padapter, bool bDLFinished)
> DBG_8723A("%s\n", __func__);
>
> ReservedPagePacket = kzalloc(1000, GFP_KERNEL);
> - if (ReservedPagePacket == NULL) {
> - DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
> + if (ReservedPagePacket == NULL)
> return;
> - }
>
> pHalData = GET_HAL_DATA(padapter);
> pxmitpriv = &padapter->xmitpriv;
> @@ -669,10 +667,8 @@ static void SetFwRsvdPagePkt_BTCoex(struct rtw_adapter *padapter)
> DBG_8723A("+%s\n", __func__);
>
> ReservedPagePacket = kzalloc(1024, GFP_KERNEL);
> - if (ReservedPagePacket == NULL) {
> - DBG_8723A("%s: alloc ReservedPagePacket fail!\n", __func__);
> + if (ReservedPagePacket == NULL)
> return;
> - }
>
> pHalData = GET_HAL_DATA(padapter);
> pxmitpriv = &padapter->xmitpriv;
> diff --git a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> index a5eadd4..6d50b09 100644
> --- a/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> +++ b/drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c
> @@ -401,10 +401,8 @@ hal_ReadEFuse_WiFi(struct rtw_adapter *padapter,
> }
>
> efuseTbl = kmalloc(EFUSE_MAP_LEN_8723A, GFP_KERNEL);
> - if (efuseTbl == NULL) {
> - DBG_8723A("%s: alloc efuseTbl fail!\n", __func__);
> + if (efuseTbl == NULL)
> return;
> - }
> /* 0xff will be efuse default value instead of 0x00. */
> memset(efuseTbl, 0xFF, EFUSE_MAP_LEN_8723A);
>
> @@ -494,10 +492,8 @@ hal_ReadEFuse_BT(struct rtw_adapter *padapter,
> }
>
> efuseTbl = kmalloc(EFUSE_BT_MAP_LEN, GFP_KERNEL);
> - if (efuseTbl == NULL) {
> - DBG_8723A("%s: efuseTbl malloc fail!\n", __func__);
> + if (efuseTbl == NULL)
> return;
> - }
> /* 0xff will be efuse default value instead of 0x00. */
> memset(efuseTbl, 0xFF, EFUSE_BT_MAP_LEN);
>
> diff --git a/drivers/staging/rtl8723au/hal/usb_ops_linux.c b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> index a6d16ad..f1e9202 100644
> --- a/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> +++ b/drivers/staging/rtl8723au/hal/usb_ops_linux.c
> @@ -256,12 +256,8 @@ static void usb_read_interrupt_complete(struct urb *purb)
> c2w = kmalloc(sizeof(struct evt_work),
> GFP_ATOMIC);
>
> - if (!c2w) {
> - printk(KERN_WARNING "%s: unable to "
> - "allocate work buffer\n",
> - __func__);
> + if (!c2w)
> goto urb_submit;
> - }
>
> c2w->adapter = padapter;
> INIT_WORK(&c2w->work, rtw_evt_work);
> diff --git a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> index 537bd82..40bdf4b 100644
> --- a/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> +++ b/drivers/staging/rtl8723au/os_dep/ioctl_cfg80211.c
> @@ -1327,8 +1327,6 @@ static int rtw_cfg80211_set_probe_req_wpsp2pie(struct rtw_adapter *padapter,
> pmlmepriv->wps_probe_req_ie = kmemdup(wps_ie, wps_ie[1],
> GFP_KERNEL);
> if (pmlmepriv->wps_probe_req_ie == NULL) {
> - DBG_8723A("%s()-%d: kmalloc() ERROR!\n",
> - __func__, __LINE__);
> return -EINVAL;
> }
> pmlmepriv->wps_probe_req_ie_len = wps_ie[1];
> @@ -2619,8 +2617,6 @@ static int rtw_cfg80211_add_monitor_if(struct rtw_adapter *padapter, char *name,
> /* wdev */
> mon_wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
> if (!mon_wdev) {
> - DBG_8723A("%s(%s): allocate mon_wdev fail\n", __func__,
> - padapter->pnetdev->name);
> ret = -ENOMEM;
> goto out;
> }
> @@ -3275,7 +3271,6 @@ int rtw_wdev_alloc(struct rtw_adapter *padapter, struct device *dev)
> /* wdev */
> wdev = kzalloc(sizeof(struct wireless_dev), GFP_KERNEL);
> if (!wdev) {
> - DBG_8723A("Couldn't allocate wireless device\n");
> ret = -ENOMEM;
> goto free_wiphy;
> }
On Fri, 2015-03-06 at 11:08 -0500, Jes Sorensen wrote:
> Julia Lawall <[email protected]> writes:
> > On Fri, 6 Mar 2015, Jes Sorensen wrote:
> >> Quentin Lambert <[email protected]> writes:
> >> > This patch reduces the kernel size by removing error messages that duplicate
> >> > the normal OOM message.
> >> > A simplified version of the semantic patch that finds this problem is as
> >> > follows: (http://coccinelle.lip6.fr)
> >> This patch removes useful warnings about what allocation failed. The
> >> messages removed are NOT duplicate!
> > Is it really the case that the information can't be reconstructed from the
> > information generated by kmalloc on failure? To my understanding there is
> > a stack trace, and from scanning through the changes I see only one change
> > per function, so perhaps the stack trace already makes it clear where the
> > problem occurred?
> It may be possible to backtrack, but this change just makes it harder.
> There are tons of real issues to fix in this driver, this patch just
> increases the risk of patch conflicts for no real gain.
Making the allocation less likely to fail for
low memory systems is a gain.
The allocation failures themselves are low
likelihood events. Determining which specific
memory allocation failure occurred has near
nil value.