[PATCH] net: tracepoint: fix print wrong sysctl_mem value

[PATCH] net: tracepoint: fix print wrong sysctl_mem value

[Dust Li]

sysctl_mem is an point, and tracepoint entry do not support
been visited like an array. Use 3 long type to get sysctl_mem
instead.

tracpoint output with and without this fix:
- without fix:
   28821.074 sock:sock_exceed_buf_limit:proto:UDP
   sysctl_mem=-1741233440,19,322156906942464 allocated=19 sysctl_rmem=4096
   rmem_alloc=75008 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
   kind=SK_MEM_RECV

- with fix:
  2126.136 sock:sock_exceed_buf_limit:proto:UDP
  sysctl_mem=18,122845,184266 allocated=19 sysctl_rmem=4096
  rmem_alloc=73728 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
  kind=SK_MEM_RECV

Fixes: 3847ce32aea9fdf ("core: add tracepoints for queueing skb to rcvbuf")
Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
---
 include/trace/events/sock.h | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/include/trace/events/sock.h b/include/trace/events/sock.h
index a966d4b5ab37..9118dd2353b7 100644
--- a/include/trace/events/sock.h
+++ b/include/trace/events/sock.h
@@ -98,7 +98,9 @@ TRACE_EVENT(sock_exceed_buf_limit,
 
 	TP_STRUCT__entry(
 		__array(char, name, 32)
-		__field(long *, sysctl_mem)
+		__field(long, sysctl_mem0)
+		__field(long, sysctl_mem1)
+		__field(long, sysctl_mem2)
 		__field(long, allocated)
 		__field(int, sysctl_rmem)
 		__field(int, rmem_alloc)
@@ -110,7 +112,9 @@ TRACE_EVENT(sock_exceed_buf_limit,
 
 	TP_fast_assign(
 		strncpy(__entry->name, prot->name, 32);
-		__entry->sysctl_mem = prot->sysctl_mem;
+		__entry->sysctl_mem0 = prot->sysctl_mem[0];
+		__entry->sysctl_mem1 = prot->sysctl_mem[1];
+		__entry->sysctl_mem2 = prot->sysctl_mem[2];
 		__entry->allocated = allocated;
 		__entry->sysctl_rmem = sk_get_rmem0(sk, prot);
 		__entry->rmem_alloc = atomic_read(&sk->sk_rmem_alloc);
@@ -122,9 +126,9 @@ TRACE_EVENT(sock_exceed_buf_limit,
 
 	TP_printk("proto:%s sysctl_mem=%ld,%ld,%ld allocated=%ld sysctl_rmem=%d rmem_alloc=%d sysctl_wmem=%d wmem_alloc=%d wmem_queued=%d kind=%s",
 		__entry->name,
-		__entry->sysctl_mem[0],
-		__entry->sysctl_mem[1],
-		__entry->sysctl_mem[2],
+		__entry->sysctl_mem0,
+		__entry->sysctl_mem1,
+		__entry->sysctl_mem2,
 		__entry->allocated,
 		__entry->sysctl_rmem,
 		__entry->rmem_alloc,
-- 
2.19.1.3.ge56e4f7

Re: [PATCH] net: tracepoint: fix print wrong sysctl_mem value

[Jakub Kicinski]

On Mon,  7 Sep 2020 22:47:57 +0800 Dust Li wrote:
> sysctl_mem is an point, and tracepoint entry do not support
> been visited like an array. Use 3 long type to get sysctl_mem
> instead.
> 
> tracpoint output with and without this fix:
> - without fix:
>    28821.074 sock:sock_exceed_buf_limit:proto:UDP
>    sysctl_mem=-1741233440,19,322156906942464 allocated=19 sysctl_rmem=4096
>    rmem_alloc=75008 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
>    kind=SK_MEM_RECV
> 
> - with fix:
>   2126.136 sock:sock_exceed_buf_limit:proto:UDP
>   sysctl_mem=18,122845,184266 allocated=19 sysctl_rmem=4096
>   rmem_alloc=73728 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
>   kind=SK_MEM_RECV
> 
> Fixes: 3847ce32aea9fdf ("core: add tracepoints for queueing skb to rcvbuf")
> Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
> ---
>  include/trace/events/sock.h | 14 +++++++++-----
>  1 file changed, 9 insertions(+), 5 deletions(-)
> 
> diff --git a/include/trace/events/sock.h b/include/trace/events/sock.h
> index a966d4b5ab37..9118dd2353b7 100644
> --- a/include/trace/events/sock.h
> +++ b/include/trace/events/sock.h
> @@ -98,7 +98,9 @@ TRACE_EVENT(sock_exceed_buf_limit,
>  
>  	TP_STRUCT__entry(
>  		__array(char, name, 32)
> -		__field(long *, sysctl_mem)
> +		__field(long, sysctl_mem0)
> +		__field(long, sysctl_mem1)
> +		__field(long, sysctl_mem2)

Why not make it an __array() ?

>  		__field(long, allocated)
>  		__field(int, sysctl_rmem)
>  		__field(int, rmem_alloc)

Re: [PATCH] net: tracepoint: fix print wrong sysctl_mem value

[dust.li]

On Mon, Sep 07, 2020 at 10:50:30AM -0700, Jakub Kicinski wrote:
>On Mon,  7 Sep 2020 22:47:57 +0800 Dust Li wrote:
>> sysctl_mem is an point, and tracepoint entry do not support
>> been visited like an array. Use 3 long type to get sysctl_mem
>> instead.
>> 
>> tracpoint output with and without this fix:
>> - without fix:
>>    28821.074 sock:sock_exceed_buf_limit:proto:UDP
>>    sysctl_mem=-1741233440,19,322156906942464 allocated=19 sysctl_rmem=4096
>>    rmem_alloc=75008 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
>>    kind=SK_MEM_RECV
>> 
>> - with fix:
>>   2126.136 sock:sock_exceed_buf_limit:proto:UDP
>>   sysctl_mem=18,122845,184266 allocated=19 sysctl_rmem=4096
>>   rmem_alloc=73728 sysctl_wmem=4096 wmem_alloc=1 wmem_queued=0
>>   kind=SK_MEM_RECV
>> 
>> Fixes: 3847ce32aea9fdf ("core: add tracepoints for queueing skb to rcvbuf")
>> Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
>> ---
>>  include/trace/events/sock.h | 14 +++++++++-----
>>  1 file changed, 9 insertions(+), 5 deletions(-)
>> 
>> diff --git a/include/trace/events/sock.h b/include/trace/events/sock.h
>> index a966d4b5ab37..9118dd2353b7 100644
>> --- a/include/trace/events/sock.h
>> +++ b/include/trace/events/sock.h
>> @@ -98,7 +98,9 @@ TRACE_EVENT(sock_exceed_buf_limit,
>>  
>>  	TP_STRUCT__entry(
>>  		__array(char, name, 32)
>> -		__field(long *, sysctl_mem)
>> +		__field(long, sysctl_mem0)
>> +		__field(long, sysctl_mem1)
>> +		__field(long, sysctl_mem2)
>
>Why not make it an __array() ?
Yeah, it's better using an __array(), I will send a v2.
Thanks !

>
>>  		__field(long, allocated)
>>  		__field(int, sysctl_rmem)
>>  		__field(int, rmem_alloc)