backports.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH V2] backport: handle change in netdevice destructor usage
@ 2017-06-21  9:40 Arend van Spriel
  0 siblings, 0 replies; 7+ messages in thread
From: Arend van Spriel @ 2017-06-21  9:40 UTC (permalink / raw)
  To: Johannes Berg; +Cc: backports, Arend van Spriel

This patch deals with changes made in struct net_device by commit
cf124db566e6 ("net: Fix inconsistent teardown and release of private
netdev state.").

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
Hi Johannes,

I took a while for me to be happy with this version. More so because I
found that brcmfmac has a double-free bug if register_netdevice() fails
so I needed to solve that puzzle. Below is the changelog.

Regards,
Arend

 V2:
  - changed netdev_set_priv_destructor() to macro.
  - revised semantic patch covering drivers only needing free_netdev().
  - used some tricks suggested by Julia in semantic patch.
---
 backport/backport-include/linux/netdevice.h |  9 ++++
 patches/0079-netdev-destructor.cocci        | 84 +++++++++++++++++++++++++++++
 2 files changed, 93 insertions(+)
 create mode 100644 patches/0079-netdev-destructor.cocci

diff --git a/backport/backport-include/linux/netdevice.h b/backport/backport-include/linux/netdevice.h
index 06230b5..0f5b198 100644
--- a/backport/backport-include/linux/netdevice.h
+++ b/backport/backport-include/linux/netdevice.h
@@ -320,4 +320,13 @@ static inline void netif_trans_update(struct net_device *dev)
 }
 #endif
 
+#if LINUX_VERSION_IS_LESS(4,13,0)
+#define netdev_set_priv_destructor(_dev, _destructor) \
+	(_dev)->destructor = __ ## _destructor
+#else
+	dev->needs_free_netdev = true;
+	if (destructor != free_netdev)
+		dev->priv_destructor = destructor;
+#endif
+
 #endif /* __BACKPORT_NETDEVICE_H */
diff --git a/patches/0079-netdev-destructor.cocci b/patches/0079-netdev-destructor.cocci
new file mode 100644
index 0000000..fab8af1
--- /dev/null
+++ b/patches/0079-netdev-destructor.cocci
@@ -0,0 +1,84 @@
+@r1@
+struct net_device *NDEV;
+identifier D, C;
+identifier TRUE =~ "true";
+@@
+C(...)
+{
+	<...
+-	NDEV->needs_free_netdev = TRUE;
+-	NDEV->priv_destructor = D;
++	netdev_set_priv_destructor(NDEV, D);
+	...>
+}
+
+@r2 depends on r1@
+identifier r1.D, r1.C;
+fresh identifier E = "__" ## D;
+@@
+
++#if LINUX_VERSION_IS_LESS(4,13,0)
++static void E(struct net_device *ndev)
++{
++	D(ndev);
++	free_netdev(ndev);
++}
++#endif
++
+C(...)
+{
+	...
+}
+
+@r3 depends on r1@
+type T;
+identifier NDEV;
+identifier r1.D;
+T RET;
+@@
+
+RET = \(register_netdevice\|register_ndev\)(NDEV);
+if (<+... RET ...+>) {
+	<...
++#if LINUX_VERSION_IS_LESS(4,13,0)
++	D(NDEV);
++#endif
+	free_netdev(NDEV);
+	...>
+}
+
+@r4 depends on r1@
+identifier NDEV;
+identifier r1.D;
+type T;
+T RET;
+@@
+
+if (...)
+	RET = register_netdevice(NDEV);
+else
+	RET = register_netdev(NDEV);
+if (<+... RET ...+>) {
+	<...
++#if LINUX_VERSION_IS_LESS(4,13,0)
++	D(NDEV);
++#endif
+	free_netdev(NDEV);
+	...>
+}
+
+@r5@
+struct net_device *NDEV;
+identifier TRUE =~ "true";
+@@
+
+-NDEV->needs_free_netdev = TRUE;
++netdev_set_priv_destructor(NDEV, free_netdev);
+
+@r6@
+struct net_device *NDEV;
+identifier D;
+@@
+
+-NDEV->priv_destructor = D;
++netdev_set_priv_destructor(NDEV, D);
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH V2] backport: handle change in netdevice destructor usage
  2017-06-21 21:52     ` Johannes Berg
@ 2017-06-21 21:53       ` Arend van Spriel
  0 siblings, 0 replies; 7+ messages in thread
From: Arend van Spriel @ 2017-06-21 21:53 UTC (permalink / raw)
  To: Johannes Berg; +Cc: backports

On 21-06-17 23:52, Johannes Berg wrote:
> On Wed, 2017-06-21 at 23:50 +0200, Arend van Spriel wrote:
>>
>>>> ++#if LINUX_VERSION_IS_LESS(4,13,0)
>>>> ++static void E(struct net_device *ndev)
> 
>> So I will send a V3 fixing the macro in netdevice.h, but I don't
>> think we need the maybe_unused attribute.
> 
> Ummm, yeah, I totally missed the #if above :)
> 
> Maybe it's time to get some sleep...

Dito. That V3 will have to wait until tomorrow.

Gr. AvS
--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH V2] backport: handle change in netdevice destructor usage
  2017-06-21 21:50   ` Arend van Spriel
@ 2017-06-21 21:52     ` Johannes Berg
  2017-06-21 21:53       ` Arend van Spriel
  0 siblings, 1 reply; 7+ messages in thread
From: Johannes Berg @ 2017-06-21 21:52 UTC (permalink / raw)
  To: Arend van Spriel; +Cc: backports

On Wed, 2017-06-21 at 23:50 +0200, Arend van Spriel wrote:
> 
> > > ++#if LINUX_VERSION_IS_LESS(4,13,0)
> > > ++static void E(struct net_device *ndev)

> So I will send a V3 fixing the macro in netdevice.h, but I don't
> think we need the maybe_unused attribute.

Ummm, yeah, I totally missed the #if above :)

Maybe it's time to get some sleep...

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH V2] backport: handle change in netdevice destructor usage
  2017-06-21 21:18 ` Johannes Berg
@ 2017-06-21 21:50   ` Arend van Spriel
  2017-06-21 21:52     ` Johannes Berg
  0 siblings, 1 reply; 7+ messages in thread
From: Arend van Spriel @ 2017-06-21 21:50 UTC (permalink / raw)
  To: Johannes Berg; +Cc: backports



On 21-06-17 23:18, Johannes Berg wrote:
> Perhaps that can be fixed as such:
> 
>> +@r2 depends on r1@
>> +identifier r1.D, r1.C;
>> +fresh identifier E = "__" ## D;
> 
> attribute __maybe_unused;
> 
>> +@@
>> +
>> ++#if LINUX_VERSION_IS_LESS(4,13,0)
>> ++static void E(struct net_device *ndev)
> 
> __maybe_unused

Ok. I clearly made a mistake in #else of the macro:

+#if LINUX_VERSION_IS_LESS(4,13,0)
+#define netdev_set_priv_destructor(_dev, _destructor) \
+	(_dev)->destructor = __ ## _destructor
+#else
+	(_dev)->needs_free_netdev = true;
+	if (_destructor != free_netdev)
+		(_dev)->priv_destructor = _destructor;
+#endif

So I will fix that, but I do not quite understand your comment.
Expanding the macro in spatch output for net/mac80211/iface.c:

+#if LINUX_VERSION_IS_LESS(4,13,0)
+static void __ieee80211_if_free(struct net_device *ndev)
+{
+	ieee80211_if_free(ndev);
+	free_netdev(ndev);
+}
+#endif
+
 static void ieee80211_if_setup(struct net_device *dev)
 {
 	ether_setup(dev);
 	dev->priv_flags &= ~IFF_TX_SKB_SHARING;
 	dev->netdev_ops = &ieee80211_dataif_ops;
-	dev->needs_free_netdev = true;
-	dev->priv_destructor = ieee80211_if_free;
+#if LINUX_VERSION_IS_LESS(4,13,0)
+       dev->destructor = __ieee80211_if_free
+#else
+       dev->needs_free_netdev = true;
+       if (ieee80211_if_free != free_netdev)
+               dev->priv_destructor = ieee80211_if_free;
+#endif
 }

So __ieee80211_if_free is only defined for lower kernels and it is only
used for lower kernels. Adding the __ieee80211_if_free depends on r1 in
the semantic patch.

So I will send a V3 fixing the macro in netdevice.h, but I don't think
we need the maybe_unused attribute.

Regards,
Arend
--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH V2] backport: handle change in netdevice destructor usage
  2017-06-21 20:58 Arend van Spriel
  2017-06-21 21:17 ` Johannes Berg
@ 2017-06-21 21:18 ` Johannes Berg
  2017-06-21 21:50   ` Arend van Spriel
  1 sibling, 1 reply; 7+ messages in thread
From: Johannes Berg @ 2017-06-21 21:18 UTC (permalink / raw)
  To: Arend van Spriel; +Cc: backports

Perhaps that can be fixed as such:

> +@r2 depends on r1@
> +identifier r1.D, r1.C;
> +fresh identifier E = "__" ## D;

attribute __maybe_unused;

> +@@
> +
> ++#if LINUX_VERSION_IS_LESS(4,13,0)
> ++static void E(struct net_device *ndev)

__maybe_unused

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH V2] backport: handle change in netdevice destructor usage
  2017-06-21 20:58 Arend van Spriel
@ 2017-06-21 21:17 ` Johannes Berg
  2017-06-21 21:18 ` Johannes Berg
  1 sibling, 0 replies; 7+ messages in thread
From: Johannes Berg @ 2017-06-21 21:17 UTC (permalink / raw)
  To: Arend van Spriel; +Cc: backports

On Wed, 2017-06-21 at 21:58 +0100, Arend van Spriel wrote:
> This patch deals with changes made in struct net_device by commit
> cf124db566e6 ("net: Fix inconsistent teardown and release of private
> netdev state.").
> 
> Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
> ---
> Hi Johannes,
> 
> It took a while for me to be happy with this version. More so because
> I
> found that brcmfmac has a double-free bug if register_netdevice()
> fails
> so I needed to solve that puzzle. Below is the changelog.
> 
> Regards,
> Arend
> 
>  V2:
>   - changed netdev_set_priv_destructor() to macro.
>   - revised semantic patch covering drivers only needing
> free_netdev().
>   - used some tricks suggested by Julia in semantic patch.
>   - added more version checks against 4.13.
> ---
>  backport/backport-include/linux/netdevice.h |  9 ++++
>  patches/0079-netdev-destructor.cocci        | 84
> +++++++++++++++++++++++++++++
>  2 files changed, 93 insertions(+)
>  create mode 100644 patches/0079-netdev-destructor.cocci
> 
> diff --git a/backport/backport-include/linux/netdevice.h
> b/backport/backport-include/linux/netdevice.h
> index 06230b5..0f5b198 100644
> --- a/backport/backport-include/linux/netdevice.h
> +++ b/backport/backport-include/linux/netdevice.h
> @@ -320,4 +320,13 @@ static inline void netif_trans_update(struct
> net_device *dev)
>  }
>  #endif
>  
> +#if LINUX_VERSION_IS_LESS(4,13,0)
> +#define netdev_set_priv_destructor(_dev, _destructor) \
> +	(_dev)->destructor = __ ## _destructor
> +#else

Won't this cause annoying warnings, that __##_destructor is an unused
function, when compiling for higher kernels?

johannes
--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply	[flat|nested] 7+ messages in thread

* [PATCH V2] backport: handle change in netdevice destructor usage
@ 2017-06-21 20:58 Arend van Spriel
  2017-06-21 21:17 ` Johannes Berg
  2017-06-21 21:18 ` Johannes Berg
  0 siblings, 2 replies; 7+ messages in thread
From: Arend van Spriel @ 2017-06-21 20:58 UTC (permalink / raw)
  To: Johannes Berg; +Cc: backports, Arend van Spriel

This patch deals with changes made in struct net_device by commit
cf124db566e6 ("net: Fix inconsistent teardown and release of private
netdev state.").

Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
---
Hi Johannes,

It took a while for me to be happy with this version. More so because I
found that brcmfmac has a double-free bug if register_netdevice() fails
so I needed to solve that puzzle. Below is the changelog.

Regards,
Arend

 V2:
  - changed netdev_set_priv_destructor() to macro.
  - revised semantic patch covering drivers only needing free_netdev().
  - used some tricks suggested by Julia in semantic patch.
  - added more version checks against 4.13.
---
 backport/backport-include/linux/netdevice.h |  9 ++++
 patches/0079-netdev-destructor.cocci        | 84 +++++++++++++++++++++++++++++
 2 files changed, 93 insertions(+)
 create mode 100644 patches/0079-netdev-destructor.cocci

diff --git a/backport/backport-include/linux/netdevice.h b/backport/backport-include/linux/netdevice.h
index 06230b5..0f5b198 100644
--- a/backport/backport-include/linux/netdevice.h
+++ b/backport/backport-include/linux/netdevice.h
@@ -320,4 +320,13 @@ static inline void netif_trans_update(struct net_device *dev)
 }
 #endif
 
+#if LINUX_VERSION_IS_LESS(4,13,0)
+#define netdev_set_priv_destructor(_dev, _destructor) \
+	(_dev)->destructor = __ ## _destructor
+#else
+	(_dev)->needs_free_netdev = true;
+	if (_destructor != free_netdev)
+		(_dev)->priv_destructor = _destructor;
+#endif
+
 #endif /* __BACKPORT_NETDEVICE_H */
diff --git a/patches/0079-netdev-destructor.cocci b/patches/0079-netdev-destructor.cocci
new file mode 100644
index 0000000..fab8af1
--- /dev/null
+++ b/patches/0079-netdev-destructor.cocci
@@ -0,0 +1,84 @@
+@r1@
+struct net_device *NDEV;
+identifier D, C;
+identifier TRUE =~ "true";
+@@
+C(...)
+{
+	<...
+-	NDEV->needs_free_netdev = TRUE;
+-	NDEV->priv_destructor = D;
++	netdev_set_priv_destructor(NDEV, D);
+	...>
+}
+
+@r2 depends on r1@
+identifier r1.D, r1.C;
+fresh identifier E = "__" ## D;
+@@
+
++#if LINUX_VERSION_IS_LESS(4,13,0)
++static void E(struct net_device *ndev)
++{
++	D(ndev);
++	free_netdev(ndev);
++}
++#endif
++
+C(...)
+{
+	...
+}
+
+@r3 depends on r1@
+type T;
+identifier NDEV;
+identifier r1.D;
+T RET;
+@@
+
+RET = \(register_netdevice\|register_ndev\)(NDEV);
+if (<+... RET ...+>) {
+	<...
++#if LINUX_VERSION_IS_LESS(4,13,0)
++	D(NDEV);
++#endif
+	free_netdev(NDEV);
+	...>
+}
+
+@r4 depends on r1@
+identifier NDEV;
+identifier r1.D;
+type T;
+T RET;
+@@
+
+if (...)
+	RET = register_netdevice(NDEV);
+else
+	RET = register_netdev(NDEV);
+if (<+... RET ...+>) {
+	<...
++#if LINUX_VERSION_IS_LESS(4,13,0)
++	D(NDEV);
++#endif
+	free_netdev(NDEV);
+	...>
+}
+
+@r5@
+struct net_device *NDEV;
+identifier TRUE =~ "true";
+@@
+
+-NDEV->needs_free_netdev = TRUE;
++netdev_set_priv_destructor(NDEV, free_netdev);
+
+@r6@
+struct net_device *NDEV;
+identifier D;
+@@
+
+-NDEV->priv_destructor = D;
++netdev_set_priv_destructor(NDEV, D);
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe backports" in

^ permalink raw reply related	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2017-06-21 21:53 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-06-21  9:40 [PATCH V2] backport: handle change in netdevice destructor usage Arend van Spriel
2017-06-21 20:58 Arend van Spriel
2017-06-21 21:17 ` Johannes Berg
2017-06-21 21:18 ` Johannes Berg
2017-06-21 21:50   ` Arend van Spriel
2017-06-21 21:52     ` Johannes Berg
2017-06-21 21:53       ` Arend van Spriel

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).