Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[Randy Dunlap]

[-- Attachment #1: Type: text/plain, Size: 646 bytes --]

On 5/18/21 2:27 AM, Stephen Rothwell wrote:
> Hi all,
> 
> Changes since 20210514:
> 

on x86_64:
# CONFIG_OF is not set

../drivers/net/dsa/qca8k.c: In function ‘qca8k_mdio_register’:
../drivers/net/dsa/qca8k.c:797:9: error: implicit declaration of function ‘devm_of_mdiobus_register’; did you mean ‘devm_mdiobus_register’? [-Werror=implicit-function-declaration]
  return devm_of_mdiobus_register(priv->dev, bus, mdio);


Should there be a stub for this function in <linux/of_mdio.h>?
or the driver could add a dependency on OF_MDIO.

Full randconfig file is attached.

-- 
~Randy
Reported-by: Randy Dunlap <rdunlap@infradead.org>


[-- Attachment #2: config-r9381.gz --]
[-- Type: application/gzip, Size: 34074 bytes --]

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[Vladimir Oltean]

Hi Randy,

On Tue, May 18, 2021 at 09:32:49AM -0700, Randy Dunlap wrote:
> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
> > Hi all,
> > 
> > Changes since 20210514:
> > 
> 
> on x86_64:
> # CONFIG_OF is not set
> 
> ../drivers/net/dsa/qca8k.c: In function ‘qca8k_mdio_register’:
> ../drivers/net/dsa/qca8k.c:797:9: error: implicit declaration of function ‘devm_of_mdiobus_register’; did you mean ‘devm_mdiobus_register’? [-Werror=implicit-function-declaration]
>   return devm_of_mdiobus_register(priv->dev, bus, mdio);
> 
> 
> Should there be a stub for this function in <linux/of_mdio.h>?
> or the driver could add a dependency on OF_MDIO.
> 
> Full randconfig file is attached.
> 
> -- 
> ~Randy
> Reported-by: Randy Dunlap <rdunlap@infradead.org>
> 

Would something like this work?

-----------------------------[ cut here ]-----------------------------
From 36c0b3f04ebfa51e52bd1bc2dc447d12d1c6e119 Mon Sep 17 00:00:00 2001
From: Vladimir Oltean <olteanv@gmail.com>
Date: Tue, 18 May 2021 19:39:18 +0300
Subject: [PATCH] net: mdio: provide shim implementation of
 devm_of_mdiobus_register

Similar to the way in which of_mdiobus_register() has a fallback to the
non-DT based mdiobus_register() when CONFIG_OF is not set, we can create
a shim for the device-managed devm_of_mdiobus_register() which calls
devm_mdiobus_register() and discards the struct device_node *.

In particular, this solves a build issue with the qca8k DSA driver which
uses devm_of_mdiobus_register and can be compiled without CONFIG_OF.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
---
 include/linux/of_mdio.h | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/include/linux/of_mdio.h b/include/linux/of_mdio.h
index 2b05e7f7c238..da633d34ab86 100644
--- a/include/linux/of_mdio.h
+++ b/include/linux/of_mdio.h
@@ -72,6 +72,13 @@ static inline int of_mdiobus_register(struct mii_bus *mdio, struct device_node *
 	return mdiobus_register(mdio);
 }
 
+static inline int devm_of_mdiobus_register(struct device *dev,
+					   struct mii_bus *mdio,
+					   struct device_node *np)
+{
+	return devm_mdiobus_register(dev, mdio);
+}
+
 static inline struct mdio_device *of_mdio_find_device(struct device_node *np)
 {
 	return NULL;
-----------------------------[ cut here ]-----------------------------

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[Randy Dunlap]

On 5/18/21 9:43 AM, Vladimir Oltean wrote:
> Hi Randy,
> 
> Would something like this work?
> 
> -----------------------------[ cut here ]-----------------------------
> From 36c0b3f04ebfa51e52bd1bc2dc447d12d1c6e119 Mon Sep 17 00:00:00 2001
> From: Vladimir Oltean <olteanv@gmail.com>
> Date: Tue, 18 May 2021 19:39:18 +0300
> Subject: [PATCH] net: mdio: provide shim implementation of
>  devm_of_mdiobus_register
> 
> Similar to the way in which of_mdiobus_register() has a fallback to the
> non-DT based mdiobus_register() when CONFIG_OF is not set, we can create
> a shim for the device-managed devm_of_mdiobus_register() which calls
> devm_mdiobus_register() and discards the struct device_node *.
> 
> In particular, this solves a build issue with the qca8k DSA driver which
> uses devm_of_mdiobus_register and can be compiled without CONFIG_OF.
> 
> Reported-by: Randy Dunlap <rdunlap@infradead.org>
> Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
> ---
>  include/linux/of_mdio.h | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/include/linux/of_mdio.h b/include/linux/of_mdio.h
> index 2b05e7f7c238..da633d34ab86 100644
> --- a/include/linux/of_mdio.h
> +++ b/include/linux/of_mdio.h
> @@ -72,6 +72,13 @@ static inline int of_mdiobus_register(struct mii_bus *mdio, struct device_node *
>  	return mdiobus_register(mdio);
>  }
>  
> +static inline int devm_of_mdiobus_register(struct device *dev,
> +					   struct mii_bus *mdio,
> +					   struct device_node *np)
> +{
> +	return devm_mdiobus_register(dev, mdio);
> +}
> +
>  static inline struct mdio_device *of_mdio_find_device(struct device_node *np)
>  {
>  	return NULL;
> -----------------------------[ cut here ]-----------------------------
> 

Yes, that's all good. Thanks.

Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested

-- 
~Randy

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[Vladimir Oltean]

On Tue, May 18, 2021 at 09:57:13AM -0700, Randy Dunlap wrote:
> On 5/18/21 9:43 AM, Vladimir Oltean wrote:
> > Hi Randy,
> > 
> > Would something like this work?
> > 
> > -----------------------------[ cut here ]-----------------------------
> > From 36c0b3f04ebfa51e52bd1bc2dc447d12d1c6e119 Mon Sep 17 00:00:00 2001
> > From: Vladimir Oltean <olteanv@gmail.com>
> > Date: Tue, 18 May 2021 19:39:18 +0300
> > Subject: [PATCH] net: mdio: provide shim implementation of
> >  devm_of_mdiobus_register
> > 
> > Similar to the way in which of_mdiobus_register() has a fallback to the
> > non-DT based mdiobus_register() when CONFIG_OF is not set, we can create
> > a shim for the device-managed devm_of_mdiobus_register() which calls
> > devm_mdiobus_register() and discards the struct device_node *.
> > 
> > In particular, this solves a build issue with the qca8k DSA driver which
> > uses devm_of_mdiobus_register and can be compiled without CONFIG_OF.
> > 
> > Reported-by: Randy Dunlap <rdunlap@infradead.org>
> > Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
> > ---
> >  include/linux/of_mdio.h | 7 +++++++
> >  1 file changed, 7 insertions(+)
> > 
> > diff --git a/include/linux/of_mdio.h b/include/linux/of_mdio.h
> > index 2b05e7f7c238..da633d34ab86 100644
> > --- a/include/linux/of_mdio.h
> > +++ b/include/linux/of_mdio.h
> > @@ -72,6 +72,13 @@ static inline int of_mdiobus_register(struct mii_bus *mdio, struct device_node *
> >  	return mdiobus_register(mdio);
> >  }
> >  
> > +static inline int devm_of_mdiobus_register(struct device *dev,
> > +					   struct mii_bus *mdio,
> > +					   struct device_node *np)
> > +{
> > +	return devm_mdiobus_register(dev, mdio);
> > +}
> > +
> >  static inline struct mdio_device *of_mdio_find_device(struct device_node *np)
> >  {
> >  	return NULL;
> > -----------------------------[ cut here ]-----------------------------
> > 
> 
> Yes, that's all good. Thanks.
> 
> Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested

Thanks. Waiting for a little bit longer for somebody a little more
authoritative on the MDIO/PHY topic, will submit formally afterwards.

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[Andrew Lunn]

> Would something like this work?
> 
> -----------------------------[ cut here ]-----------------------------
> >From 36c0b3f04ebfa51e52bd1bc2dc447d12d1c6e119 Mon Sep 17 00:00:00 2001
> From: Vladimir Oltean <olteanv@gmail.com>
> Date: Tue, 18 May 2021 19:39:18 +0300
> Subject: [PATCH] net: mdio: provide shim implementation of
>  devm_of_mdiobus_register
> 
> Similar to the way in which of_mdiobus_register() has a fallback to the
> non-DT based mdiobus_register() when CONFIG_OF is not set, we can create
> a shim for the device-managed devm_of_mdiobus_register() which calls
> devm_mdiobus_register() and discards the struct device_node *.
> 
> In particular, this solves a build issue with the qca8k DSA driver which
> uses devm_of_mdiobus_register and can be compiled without CONFIG_OF.
> 
> Reported-by: Randy Dunlap <rdunlap@infradead.org>
> Signed-off-by: Vladimir Oltean <olteanv@gmail.com>

This should be O.K.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>

Thanks
    Andrew

Re: linux-next: Tree for May 18 (drivers/net/dsa/qca8k.c)

[John Crispin]

On 18.05.21 19:33, Andrew Lunn wrote:
>> Would something like this work?
>>
>> -----------------------------[ cut here ]-----------------------------
>> >From 36c0b3f04ebfa51e52bd1bc2dc447d12d1c6e119 Mon Sep 17 00:00:00 2001
>> From: Vladimir Oltean <olteanv@gmail.com>
>> Date: Tue, 18 May 2021 19:39:18 +0300
>> Subject: [PATCH] net: mdio: provide shim implementation of
>>   devm_of_mdiobus_register
>>
>> Similar to the way in which of_mdiobus_register() has a fallback to the
>> non-DT based mdiobus_register() when CONFIG_OF is not set, we can create
>> a shim for the device-managed devm_of_mdiobus_register() which calls
>> devm_mdiobus_register() and discards the struct device_node *.
>>
>> In particular, this solves a build issue with the qca8k DSA driver which
>> uses devm_of_mdiobus_register and can be compiled without CONFIG_OF.
>>
>> Reported-by: Randy Dunlap <rdunlap@infradead.org>
>> Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
> This should be O.K.
>
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
>
> Thanks
>      Andrew

Just did a x86 build with the patch applied and it completed ...

Acked-by: John Crispin <john@phrozen.org>

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Randy Dunlap]

[-- Attachment #1: Type: text/plain, Size: 437 bytes --]

On 5/18/21 2:27 AM, Stephen Rothwell wrote:
> Hi all,
> 
> Changes since 20210514:
> 

on i386:
# CONFIG_NET is not set

ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'


Full randconfig file is attached.

-- 
~Randy
Reported-by: Randy Dunlap <rdunlap@infradead.org>


[-- Attachment #2: config-r9369.gz --]
[-- Type: application/gzip, Size: 34789 bytes --]

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Randy Dunlap]

On 5/18/21 10:02 AM, Randy Dunlap wrote:
> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
>> Hi all,
>>
>> Changes since 20210514:
>>
> 
> on i386:
> # CONFIG_NET is not set
> 
> ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
> bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
> ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'
> 
> 
> Full randconfig file is attached.
> 

Hi,
I am still seeing this build error in linux-next-20210525.

-- 
~Randy
Reported-by: Randy Dunlap <rdunlap@infradead.org>

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Daniel Borkmann]

On 5/25/21 7:30 PM, Randy Dunlap wrote:
> On 5/18/21 10:02 AM, Randy Dunlap wrote:
>> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
>>> Hi all,
>>>
>>> Changes since 20210514:
>>>
>>
>> on i386:
>> # CONFIG_NET is not set
>>
>> ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
>> bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
>> ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'
>>
>>
>> Full randconfig file is attached.
>>
> 
> Hi,
> I am still seeing this build error in linux-next-20210525.

Will take a look and get back.

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Daniel Borkmann]

Hi Randy,

On 5/25/21 8:26 PM, Daniel Borkmann wrote:
> On 5/25/21 7:30 PM, Randy Dunlap wrote:
>> On 5/18/21 10:02 AM, Randy Dunlap wrote:
>>> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
>>>> Hi all,
>>>>
>>>> Changes since 20210514:
>>>>
>>>
>>> on i386:
>>> # CONFIG_NET is not set
>>>
>>> ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
>>> bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
>>> ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'
>>>
>>>
>>> Full randconfig file is attached.
>>>
>>
>> Hi,
>> I am still seeing this build error in linux-next-20210525.
> 
> Will take a look and get back.

This should resolve it:

diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 5efb2b24012c..da471bf01b97 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -107,10 +107,12 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
  		return &bpf_inode_storage_get_proto;
  	case BPF_FUNC_inode_storage_delete:
  		return &bpf_inode_storage_delete_proto;
+#ifdef CONFIG_NET
  	case BPF_FUNC_sk_storage_get:
  		return &bpf_sk_storage_get_proto;
  	case BPF_FUNC_sk_storage_delete:
  		return &bpf_sk_storage_delete_proto;
+#endif /* CONFIG_NET */
  	case BPF_FUNC_spin_lock:
  		return &bpf_spin_lock_proto;
  	case BPF_FUNC_spin_unlock:

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Randy Dunlap]

On 5/25/21 11:31 AM, Daniel Borkmann wrote:
> Hi Randy,
> 
> On 5/25/21 8:26 PM, Daniel Borkmann wrote:
>> On 5/25/21 7:30 PM, Randy Dunlap wrote:
>>> On 5/18/21 10:02 AM, Randy Dunlap wrote:
>>>> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
>>>>> Hi all,
>>>>>
>>>>> Changes since 20210514:
>>>>>
>>>>
>>>> on i386:
>>>> # CONFIG_NET is not set
>>>>
>>>> ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
>>>> bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
>>>> ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'
>>>>
>>>>
>>>> Full randconfig file is attached.
>>>>
>>>
>>> Hi,
>>> I am still seeing this build error in linux-next-20210525.
>>
>> Will take a look and get back.
> 
> This should resolve it:

Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested


Thanks.

> 
> diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
> index 5efb2b24012c..da471bf01b97 100644
> --- a/kernel/bpf/bpf_lsm.c
> +++ b/kernel/bpf/bpf_lsm.c
> @@ -107,10 +107,12 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
>         return &bpf_inode_storage_get_proto;
>     case BPF_FUNC_inode_storage_delete:
>         return &bpf_inode_storage_delete_proto;
> +#ifdef CONFIG_NET
>     case BPF_FUNC_sk_storage_get:
>         return &bpf_sk_storage_get_proto;
>     case BPF_FUNC_sk_storage_delete:
>         return &bpf_sk_storage_delete_proto;
> +#endif /* CONFIG_NET */
>     case BPF_FUNC_spin_lock:
>         return &bpf_spin_lock_proto;
>     case BPF_FUNC_spin_unlock:


-- 
~Randy

Re: linux-next: Tree for May 18 (kernel/bpf/bpf_lsm.o)

[Daniel Borkmann]

On 5/25/21 9:01 PM, Randy Dunlap wrote:
> On 5/25/21 11:31 AM, Daniel Borkmann wrote:
>> On 5/25/21 8:26 PM, Daniel Borkmann wrote:
>>> On 5/25/21 7:30 PM, Randy Dunlap wrote:
>>>> On 5/18/21 10:02 AM, Randy Dunlap wrote:
>>>>> On 5/18/21 2:27 AM, Stephen Rothwell wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> Changes since 20210514:
>>>>>
>>>>> on i386:
>>>>> # CONFIG_NET is not set
>>>>>
>>>>> ld: kernel/bpf/bpf_lsm.o: in function `bpf_lsm_func_proto':
>>>>> bpf_lsm.c:(.text+0x1a0): undefined reference to `bpf_sk_storage_get_proto'
>>>>> ld: bpf_lsm.c:(.text+0x1b8): undefined reference to `bpf_sk_storage_delete_proto'
>>>>>
>>>>> Full randconfig file is attached.
>>>>
>>>> Hi,
>>>> I am still seeing this build error in linux-next-20210525.
>>>
>>> Will take a look and get back.
>>
>> This should resolve it:
> 
> Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested

Thanks Randy, fix pushed here [0].

   [0] https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf.git/commit/?id=3c9e8fec3c5a7833d891ac8e58fd9f6c2df0be91