[PATCH net] net: ethtool: tsinfo: Fix dump command

[PATCH net] net: ethtool: tsinfo: Fix dump command

[Kory Maincent]

Fix missing initialization of ts_info->phc_index in the dump command,
which could cause a netdev interface to incorrectly display a PTP provider
at index 0 instead of "none".
Fix it by initializing the phc_index to -1.

In the same time, restore missing initialization of ts_info.cmd for the
IOCTL case, as it was before the transition from ethnl_default_dumpit to
custom ethnl_tsinfo_dumpit.

Fixes: b9e3f7dc9ed95 ("net: ethtool: tsinfo: Enhance tsinfo to support several hwtstamp by net topology")
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
---
 net/ethtool/tsinfo.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/ethtool/tsinfo.c b/net/ethtool/tsinfo.c
index 691be6c445b38..9edc5dc30de88 100644
--- a/net/ethtool/tsinfo.c
+++ b/net/ethtool/tsinfo.c
@@ -291,6 +291,8 @@ static void *ethnl_tsinfo_prepare_dump(struct sk_buff *skb,
 	memset(reply_data, 0, sizeof(*reply_data));
 	reply_data->base.dev = dev;
 	memset(&reply_data->ts_info, 0, sizeof(reply_data->ts_info));
+	reply_data->ts_info.cmd = ETHTOOL_GET_TS_INFO;
+	reply_data->ts_info.phc_index = -1;
 
 	return ehdr;
 }
-- 
2.34.1

Re: [PATCH net] net: ethtool: tsinfo: Fix dump command

[Vadim Fedorenko]

On 05/03/2025 14:03, Kory Maincent wrote:
> Fix missing initialization of ts_info->phc_index in the dump command,
> which could cause a netdev interface to incorrectly display a PTP provider
> at index 0 instead of "none".
> Fix it by initializing the phc_index to -1.
> 
> In the same time, restore missing initialization of ts_info.cmd for the
> IOCTL case, as it was before the transition from ethnl_default_dumpit to
> custom ethnl_tsinfo_dumpit.
> 
> Fixes: b9e3f7dc9ed95 ("net: ethtool: tsinfo: Enhance tsinfo to support several hwtstamp by net topology")
> Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
> ---
>   net/ethtool/tsinfo.c | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/net/ethtool/tsinfo.c b/net/ethtool/tsinfo.c
> index 691be6c445b38..9edc5dc30de88 100644
> --- a/net/ethtool/tsinfo.c
> +++ b/net/ethtool/tsinfo.c
> @@ -291,6 +291,8 @@ static void *ethnl_tsinfo_prepare_dump(struct sk_buff *skb,
>   	memset(reply_data, 0, sizeof(*reply_data));
>   	reply_data->base.dev = dev;
>   	memset(&reply_data->ts_info, 0, sizeof(reply_data->ts_info));
> +	reply_data->ts_info.cmd = ETHTOOL_GET_TS_INFO;
> +	reply_data->ts_info.phc_index = -1;

This change makes sense, but I'm curious why do we need
memset(&reply_data->ts_info, 0, sizeof(reply_data->ts_info))
at all? ts_info is embedded into reply_data which fully zeroed 2 lines
before.

>   
>   	return ehdr;
>   }

Re: [PATCH net] net: ethtool: tsinfo: Fix dump command

[Kory Maincent]

On Thu, 6 Mar 2025 10:38:58 +0000
Vadim Fedorenko <vadim.fedorenko@linux.dev> wrote:

> On 05/03/2025 14:03, Kory Maincent wrote:
> > Fix missing initialization of ts_info->phc_index in the dump command,
> > which could cause a netdev interface to incorrectly display a PTP provider
> > at index 0 instead of "none".
> > Fix it by initializing the phc_index to -1.
> > 
> > In the same time, restore missing initialization of ts_info.cmd for the
> > IOCTL case, as it was before the transition from ethnl_default_dumpit to
> > custom ethnl_tsinfo_dumpit.
> > 
> > Fixes: b9e3f7dc9ed95 ("net: ethtool: tsinfo: Enhance tsinfo to support
> > several hwtstamp by net topology") Signed-off-by: Kory Maincent
> > <kory.maincent@bootlin.com> ---
> >   net/ethtool/tsinfo.c | 2 ++
> >   1 file changed, 2 insertions(+)
> > 
> > diff --git a/net/ethtool/tsinfo.c b/net/ethtool/tsinfo.c
> > index 691be6c445b38..9edc5dc30de88 100644
> > --- a/net/ethtool/tsinfo.c
> > +++ b/net/ethtool/tsinfo.c
> > @@ -291,6 +291,8 @@ static void *ethnl_tsinfo_prepare_dump(struct sk_buff
> > *skb, memset(reply_data, 0, sizeof(*reply_data));
> >   	reply_data->base.dev = dev;
> >   	memset(&reply_data->ts_info, 0, sizeof(reply_data->ts_info));
> > +	reply_data->ts_info.cmd = ETHTOOL_GET_TS_INFO;
> > +	reply_data->ts_info.phc_index = -1;  
> 
> This change makes sense, but I'm curious why do we need
> memset(&reply_data->ts_info, 0, sizeof(reply_data->ts_info))
> at all? ts_info is embedded into reply_data which fully zeroed 2 lines
> before.

Indeed you are right, this is totally useless.
I will remove it.

Regards,

-- 
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com