diff mbox series

[v7,net-next,2/2] net: drop_monitor: support drop reason

Message ID 20220208072836.3540192-3-imagedong@tencent.com (mailing list archive)
State Superseded
Delegated to: Netdev Maintainers
Headers show
Series net: drop_monitor: support drop reason | expand

Checks

Context Check Description
netdev/tree_selection success Clearly marked for net-next
netdev/fixes_present success Fixes tag not required for -next series
netdev/subject_prefix success Link
netdev/cover_letter success Series has a cover letter
netdev/patch_count success Link
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 2 this patch: 2
netdev/cc_maintainers success CCed 4 of 4 maintainers
netdev/build_clang success Errors and warnings before: 18 this patch: 18
netdev/module_param success Was 0 now: 0
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/verify_fixes success No Fixes tag
netdev/build_allmodconfig_warn success Errors and warnings before: 7 this patch: 7
netdev/checkpatch fail ERROR: space prohibited before open square bracket '['
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/source_inline success Was 0 now: 0

Commit Message

Menglong Dong Feb. 8, 2022, 7:28 a.m. UTC
From: Menglong Dong <imagedong@tencent.com>

In the commit c504e5c2f964 ("net: skb: introduce kfree_skb_reason()")
drop reason is introduced to the tracepoint of kfree_skb. Therefore,
drop_monitor is able to report the drop reason to users by netlink.

The drop reasons are reported as string to users, which is exactly
the same as what we do when reporting it to ftrace.

Signed-off-by: Menglong Dong <imagedong@tencent.com>
---
v7:
- take the size of NET_DM_ATTR_REASON into accounting in
  net_dm_packet_report_size()
- let compiler define the size of drop_reasons

v6:
- check the range of drop reason in net_dm_packet_report_fill()

v5:
- check if drop reason larger than SKB_DROP_REASON_MAX

v4:
- report drop reasons as string

v3:
- referring to cb->reason and cb->pc directly in
  net_dm_packet_report_fill()

v2:
- get a pointer to struct net_dm_skb_cb instead of local var for
  each field
---
 include/uapi/linux/net_dropmon.h |  1 +
 net/core/drop_monitor.c          | 34 ++++++++++++++++++++++++++++----
 2 files changed, 31 insertions(+), 4 deletions(-)

Comments

Ido Schimmel Feb. 8, 2022, 2:58 p.m. UTC | #1
On Tue, Feb 08, 2022 at 03:28:36PM +0800, menglong8.dong@gmail.com wrote:
> From: Menglong Dong <imagedong@tencent.com>
> 
> In the commit c504e5c2f964 ("net: skb: introduce kfree_skb_reason()")
> drop reason is introduced to the tracepoint of kfree_skb. Therefore,
> drop_monitor is able to report the drop reason to users by netlink.
> 
> The drop reasons are reported as string to users, which is exactly
> the same as what we do when reporting it to ftrace.
> 
> Signed-off-by: Menglong Dong <imagedong@tencent.com>
> ---
> v7:
> - take the size of NET_DM_ATTR_REASON into accounting in
>   net_dm_packet_report_size()
> - let compiler define the size of drop_reasons
> 
> v6:
> - check the range of drop reason in net_dm_packet_report_fill()
> 
> v5:
> - check if drop reason larger than SKB_DROP_REASON_MAX
> 
> v4:
> - report drop reasons as string
> 
> v3:
> - referring to cb->reason and cb->pc directly in
>   net_dm_packet_report_fill()
> 
> v2:
> - get a pointer to struct net_dm_skb_cb instead of local var for
>   each field
> ---
>  include/uapi/linux/net_dropmon.h |  1 +
>  net/core/drop_monitor.c          | 34 ++++++++++++++++++++++++++++----
>  2 files changed, 31 insertions(+), 4 deletions(-)
> 
> diff --git a/include/uapi/linux/net_dropmon.h b/include/uapi/linux/net_dropmon.h
> index 66048cc5d7b3..1bbea8f0681e 100644
> --- a/include/uapi/linux/net_dropmon.h
> +++ b/include/uapi/linux/net_dropmon.h
> @@ -93,6 +93,7 @@ enum net_dm_attr {
>  	NET_DM_ATTR_SW_DROPS,			/* flag */
>  	NET_DM_ATTR_HW_DROPS,			/* flag */
>  	NET_DM_ATTR_FLOW_ACTION_COOKIE,		/* binary */
> +	NET_DM_ATTR_REASON,			/* string */
>  
>  	__NET_DM_ATTR_MAX,
>  	NET_DM_ATTR_MAX = __NET_DM_ATTR_MAX - 1
> diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
> index 7b288a121a41..28c55d605566 100644
> --- a/net/core/drop_monitor.c
> +++ b/net/core/drop_monitor.c
> @@ -48,6 +48,19 @@
>  static int trace_state = TRACE_OFF;
>  static bool monitor_hw;
>  
> +#undef EM
> +#undef EMe
> +
> +#define EM(a, b)	[a] = #b,
> +#define EMe(a, b)	[a] = #b
> +
> +/* drop_reasons is used to translate 'enum skb_drop_reason' to string,
> + * which is reported to user space.
> + */
> +static const char * const drop_reasons[] = {
> +	TRACE_SKB_DROP_REASON
> +};
> +
>  /* net_dm_mutex
>   *
>   * An overall lock guarding every operation coming from userspace.
> @@ -126,6 +139,7 @@ struct net_dm_skb_cb {
>  		struct devlink_trap_metadata *hw_metadata;
>  		void *pc;
>  	};
> +	enum skb_drop_reason reason;
>  };
>  
>  #define NET_DM_SKB_CB(__skb) ((struct net_dm_skb_cb *)&((__skb)->cb[0]))
> @@ -498,6 +512,7 @@ static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
>  {
>  	ktime_t tstamp = ktime_get_real();
>  	struct per_cpu_dm_data *data;
> +	struct net_dm_skb_cb *cb;
>  	struct sk_buff *nskb;
>  	unsigned long flags;
>  
> @@ -508,7 +523,9 @@ static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
>  	if (!nskb)
>  		return;
>  
> -	NET_DM_SKB_CB(nskb)->pc = location;
> +	cb = NET_DM_SKB_CB(nskb);
> +	cb->reason = reason;
> +	cb->pc = location;
>  	/* Override the timestamp because we care about the time when the
>  	 * packet was dropped.
>  	 */
> @@ -574,6 +591,8 @@ static size_t net_dm_packet_report_size(size_t payload_len)
>  	       nla_total_size(sizeof(u32)) +
>  	       /* NET_DM_ATTR_PROTO */
>  	       nla_total_size(sizeof(u16)) +
> +	       /* NET_DM_ATTR_REASON */
> +	       nla_total_size(SKB_DR_MAX_LEN + 1) +

Nothing ensures that the reason is not longer than this length and
nothing ensures that this assumption remains valid as more reasons are
added.

I think "SKB_DR_MAX_LEN" can be removed completely. Pass "reason" to
this function and do "strlen(drop_reasons[reason]) + 1". Any reason it
can't work?

>  	       /* NET_DM_ATTR_PAYLOAD */
>  	       nla_total_size(payload_len);
>  }
> @@ -606,8 +625,9 @@ static int net_dm_packet_report_in_port_put(struct sk_buff *msg, int ifindex,
>  static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
>  				     size_t payload_len)
>  {
> -	u64 pc = (u64)(uintptr_t) NET_DM_SKB_CB(skb)->pc;
> +	struct net_dm_skb_cb *cb = NET_DM_SKB_CB(skb);
>  	char buf[NET_DM_MAX_SYMBOL_LEN];
> +	unsigned int reason;
>  	struct nlattr *attr;
>  	void *hdr;
>  	int rc;
> @@ -620,10 +640,16 @@ static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
>  	if (nla_put_u16(msg, NET_DM_ATTR_ORIGIN, NET_DM_ORIGIN_SW))
>  		goto nla_put_failure;
>  
> -	if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, pc, NET_DM_ATTR_PAD))
> +	if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, (u64)(uintptr_t)cb->pc,
> +			      NET_DM_ATTR_PAD))
> +		goto nla_put_failure;
> +
> +	reason = (unsigned int)cb->reason;
> +	if (reason < SKB_DROP_REASON_MAX &&

In which cases can this happen? Might be better to perform this
validation in net_dm_packet_trace_kfree_skb_hit() and set "cb->reason"
to "SKB_DROP_REASON_NOT_SPECIFIED" in this case. That way we don't need
to perform the validation in later code paths

> +	    nla_put_string(msg, NET_DM_ATTR_REASON, drop_reasons[reason]))
>  		goto nla_put_failure;
>  
> -	snprintf(buf, sizeof(buf), "%pS", NET_DM_SKB_CB(skb)->pc);
> +	snprintf(buf, sizeof(buf), "%pS", cb->pc);
>  	if (nla_put_string(msg, NET_DM_ATTR_SYMBOL, buf))
>  		goto nla_put_failure;
>  
> -- 
> 2.34.1
>
Menglong Dong Feb. 9, 2022, 2:49 a.m. UTC | #2
On Tue, Feb 8, 2022 at 10:58 PM Ido Schimmel <idosch@idosch.org> wrote:
>
> On Tue, Feb 08, 2022 at 03:28:36PM +0800, menglong8.dong@gmail.com wrote:
> > From: Menglong Dong <imagedong@tencent.com>
> >
> > In the commit c504e5c2f964 ("net: skb: introduce kfree_skb_reason()")
> > drop reason is introduced to the tracepoint of kfree_skb. Therefore,
> > drop_monitor is able to report the drop reason to users by netlink.
> >
> > The drop reasons are reported as string to users, which is exactly
> > the same as what we do when reporting it to ftrace.
> >
> > Signed-off-by: Menglong Dong <imagedong@tencent.com>
> > ---
> > v7:
> > - take the size of NET_DM_ATTR_REASON into accounting in
> >   net_dm_packet_report_size()
> > - let compiler define the size of drop_reasons
> >
> > v6:
> > - check the range of drop reason in net_dm_packet_report_fill()
> >
> > v5:
> > - check if drop reason larger than SKB_DROP_REASON_MAX
> >
> > v4:
> > - report drop reasons as string
> >
> > v3:
> > - referring to cb->reason and cb->pc directly in
> >   net_dm_packet_report_fill()
> >
> > v2:
> > - get a pointer to struct net_dm_skb_cb instead of local var for
> >   each field
> > ---
> >  include/uapi/linux/net_dropmon.h |  1 +
> >  net/core/drop_monitor.c          | 34 ++++++++++++++++++++++++++++----
> >  2 files changed, 31 insertions(+), 4 deletions(-)
> >
> > diff --git a/include/uapi/linux/net_dropmon.h b/include/uapi/linux/net_dropmon.h
> > index 66048cc5d7b3..1bbea8f0681e 100644
> > --- a/include/uapi/linux/net_dropmon.h
> > +++ b/include/uapi/linux/net_dropmon.h
> > @@ -93,6 +93,7 @@ enum net_dm_attr {
> >       NET_DM_ATTR_SW_DROPS,                   /* flag */
> >       NET_DM_ATTR_HW_DROPS,                   /* flag */
> >       NET_DM_ATTR_FLOW_ACTION_COOKIE,         /* binary */
> > +     NET_DM_ATTR_REASON,                     /* string */
> >
> >       __NET_DM_ATTR_MAX,
> >       NET_DM_ATTR_MAX = __NET_DM_ATTR_MAX - 1
> > diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
> > index 7b288a121a41..28c55d605566 100644
> > --- a/net/core/drop_monitor.c
> > +++ b/net/core/drop_monitor.c
> > @@ -48,6 +48,19 @@
> >  static int trace_state = TRACE_OFF;
> >  static bool monitor_hw;
> >
> > +#undef EM
> > +#undef EMe
> > +
> > +#define EM(a, b)     [a] = #b,
> > +#define EMe(a, b)    [a] = #b
> > +
> > +/* drop_reasons is used to translate 'enum skb_drop_reason' to string,
> > + * which is reported to user space.
> > + */
> > +static const char * const drop_reasons[] = {
> > +     TRACE_SKB_DROP_REASON
> > +};
> > +
> >  /* net_dm_mutex
> >   *
> >   * An overall lock guarding every operation coming from userspace.
> > @@ -126,6 +139,7 @@ struct net_dm_skb_cb {
> >               struct devlink_trap_metadata *hw_metadata;
> >               void *pc;
> >       };
> > +     enum skb_drop_reason reason;
> >  };
> >
> >  #define NET_DM_SKB_CB(__skb) ((struct net_dm_skb_cb *)&((__skb)->cb[0]))
> > @@ -498,6 +512,7 @@ static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
> >  {
> >       ktime_t tstamp = ktime_get_real();
> >       struct per_cpu_dm_data *data;
> > +     struct net_dm_skb_cb *cb;
> >       struct sk_buff *nskb;
> >       unsigned long flags;
> >
> > @@ -508,7 +523,9 @@ static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
> >       if (!nskb)
> >               return;
> >
> > -     NET_DM_SKB_CB(nskb)->pc = location;
> > +     cb = NET_DM_SKB_CB(nskb);
> > +     cb->reason = reason;
> > +     cb->pc = location;
> >       /* Override the timestamp because we care about the time when the
> >        * packet was dropped.
> >        */
> > @@ -574,6 +591,8 @@ static size_t net_dm_packet_report_size(size_t payload_len)
> >              nla_total_size(sizeof(u32)) +
> >              /* NET_DM_ATTR_PROTO */
> >              nla_total_size(sizeof(u16)) +
> > +            /* NET_DM_ATTR_REASON */
> > +            nla_total_size(SKB_DR_MAX_LEN + 1) +
>
> Nothing ensures that the reason is not longer than this length and
> nothing ensures that this assumption remains valid as more reasons are
> added.
>
> I think "SKB_DR_MAX_LEN" can be removed completely. Pass "reason" to
> this function and do "strlen(drop_reasons[reason]) + 1". Any reason it
> can't work?

Yeah, it can work. But it feels a little weird to pass this param to
net_dm_packet_report_size(). I'll give it a try.

>
> >              /* NET_DM_ATTR_PAYLOAD */
> >              nla_total_size(payload_len);
> >  }
> > @@ -606,8 +625,9 @@ static int net_dm_packet_report_in_port_put(struct sk_buff *msg, int ifindex,
> >  static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
> >                                    size_t payload_len)
> >  {
> > -     u64 pc = (u64)(uintptr_t) NET_DM_SKB_CB(skb)->pc;
> > +     struct net_dm_skb_cb *cb = NET_DM_SKB_CB(skb);
> >       char buf[NET_DM_MAX_SYMBOL_LEN];
> > +     unsigned int reason;
> >       struct nlattr *attr;
> >       void *hdr;
> >       int rc;
> > @@ -620,10 +640,16 @@ static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
> >       if (nla_put_u16(msg, NET_DM_ATTR_ORIGIN, NET_DM_ORIGIN_SW))
> >               goto nla_put_failure;
> >
> > -     if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, pc, NET_DM_ATTR_PAD))
> > +     if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, (u64)(uintptr_t)cb->pc,
> > +                           NET_DM_ATTR_PAD))
> > +             goto nla_put_failure;
> > +
> > +     reason = (unsigned int)cb->reason;
> > +     if (reason < SKB_DROP_REASON_MAX &&
>
> In which cases can this happen? Might be better to perform this
> validation in net_dm_packet_trace_kfree_skb_hit() and set "cb->reason"
> to "SKB_DROP_REASON_NOT_SPECIFIED" in this case. That way we don't need
> to perform the validation in later code paths

Logically speaking, this shouldn't happen, as the reason is always be
'enum skb_drop_reason'. I added this part out of misunderstanding
your previous reply, and now I'm not sure if we should keep this.

For security considering, let's keep it for the moment, and I'll move it
to net_dm_packet_trace_kfree_skb_hit()

Thanks!
Menglong Dong

>
> > +         nla_put_string(msg, NET_DM_ATTR_REASON, drop_reasons[reason]))
> >               goto nla_put_failure;
> >
> > -     snprintf(buf, sizeof(buf), "%pS", NET_DM_SKB_CB(skb)->pc);
> > +     snprintf(buf, sizeof(buf), "%pS", cb->pc);
> >       if (nla_put_string(msg, NET_DM_ATTR_SYMBOL, buf))
> >               goto nla_put_failure;
> >
> > --
> > 2.34.1
> >
diff mbox series

Patch

diff --git a/include/uapi/linux/net_dropmon.h b/include/uapi/linux/net_dropmon.h
index 66048cc5d7b3..1bbea8f0681e 100644
--- a/include/uapi/linux/net_dropmon.h
+++ b/include/uapi/linux/net_dropmon.h
@@ -93,6 +93,7 @@  enum net_dm_attr {
 	NET_DM_ATTR_SW_DROPS,			/* flag */
 	NET_DM_ATTR_HW_DROPS,			/* flag */
 	NET_DM_ATTR_FLOW_ACTION_COOKIE,		/* binary */
+	NET_DM_ATTR_REASON,			/* string */
 
 	__NET_DM_ATTR_MAX,
 	NET_DM_ATTR_MAX = __NET_DM_ATTR_MAX - 1
diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
index 7b288a121a41..28c55d605566 100644
--- a/net/core/drop_monitor.c
+++ b/net/core/drop_monitor.c
@@ -48,6 +48,19 @@ 
 static int trace_state = TRACE_OFF;
 static bool monitor_hw;
 
+#undef EM
+#undef EMe
+
+#define EM(a, b)	[a] = #b,
+#define EMe(a, b)	[a] = #b
+
+/* drop_reasons is used to translate 'enum skb_drop_reason' to string,
+ * which is reported to user space.
+ */
+static const char * const drop_reasons[] = {
+	TRACE_SKB_DROP_REASON
+};
+
 /* net_dm_mutex
  *
  * An overall lock guarding every operation coming from userspace.
@@ -126,6 +139,7 @@  struct net_dm_skb_cb {
 		struct devlink_trap_metadata *hw_metadata;
 		void *pc;
 	};
+	enum skb_drop_reason reason;
 };
 
 #define NET_DM_SKB_CB(__skb) ((struct net_dm_skb_cb *)&((__skb)->cb[0]))
@@ -498,6 +512,7 @@  static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
 {
 	ktime_t tstamp = ktime_get_real();
 	struct per_cpu_dm_data *data;
+	struct net_dm_skb_cb *cb;
 	struct sk_buff *nskb;
 	unsigned long flags;
 
@@ -508,7 +523,9 @@  static void net_dm_packet_trace_kfree_skb_hit(void *ignore,
 	if (!nskb)
 		return;
 
-	NET_DM_SKB_CB(nskb)->pc = location;
+	cb = NET_DM_SKB_CB(nskb);
+	cb->reason = reason;
+	cb->pc = location;
 	/* Override the timestamp because we care about the time when the
 	 * packet was dropped.
 	 */
@@ -574,6 +591,8 @@  static size_t net_dm_packet_report_size(size_t payload_len)
 	       nla_total_size(sizeof(u32)) +
 	       /* NET_DM_ATTR_PROTO */
 	       nla_total_size(sizeof(u16)) +
+	       /* NET_DM_ATTR_REASON */
+	       nla_total_size(SKB_DR_MAX_LEN + 1) +
 	       /* NET_DM_ATTR_PAYLOAD */
 	       nla_total_size(payload_len);
 }
@@ -606,8 +625,9 @@  static int net_dm_packet_report_in_port_put(struct sk_buff *msg, int ifindex,
 static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
 				     size_t payload_len)
 {
-	u64 pc = (u64)(uintptr_t) NET_DM_SKB_CB(skb)->pc;
+	struct net_dm_skb_cb *cb = NET_DM_SKB_CB(skb);
 	char buf[NET_DM_MAX_SYMBOL_LEN];
+	unsigned int reason;
 	struct nlattr *attr;
 	void *hdr;
 	int rc;
@@ -620,10 +640,16 @@  static int net_dm_packet_report_fill(struct sk_buff *msg, struct sk_buff *skb,
 	if (nla_put_u16(msg, NET_DM_ATTR_ORIGIN, NET_DM_ORIGIN_SW))
 		goto nla_put_failure;
 
-	if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, pc, NET_DM_ATTR_PAD))
+	if (nla_put_u64_64bit(msg, NET_DM_ATTR_PC, (u64)(uintptr_t)cb->pc,
+			      NET_DM_ATTR_PAD))
+		goto nla_put_failure;
+
+	reason = (unsigned int)cb->reason;
+	if (reason < SKB_DROP_REASON_MAX &&
+	    nla_put_string(msg, NET_DM_ATTR_REASON, drop_reasons[reason]))
 		goto nla_put_failure;
 
-	snprintf(buf, sizeof(buf), "%pS", NET_DM_SKB_CB(skb)->pc);
+	snprintf(buf, sizeof(buf), "%pS", cb->pc);
 	if (nla_put_string(msg, NET_DM_ATTR_SYMBOL, buf))
 		goto nla_put_failure;