| Message ID | 20210728035605.24510-1-yajun.deng@linux.dev |
|---|---|
| State | New |
| Headers | show |
| Series | Revert "net: Get rid of consume_skb when tracing is off" | expand |
On Wed, Jul 28, 2021 at 11:56:05AM +0800, Yajun Deng wrote: > This reverts commit be769db2f95861cc8c7c8fedcc71a8c39b803b10. > There is trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is > also a trace function. > > Fixes: be769db2f958 (net: Get rid of consume_skb when tracing is off) > Signed-off-by: Yajun Deng <yajun.deng@linux.dev> Please explain in more detail why your patch is needed. As it stands I do not understand the reasoning. Thanks,
July 28, 2021 8:52 PM, "Herbert Xu" <herbert@gondor.apana.org.au> wrote: > On Wed, Jul 28, 2021 at 11:56:05AM +0800, Yajun Deng wrote: > >> This reverts commit be769db2f95861cc8c7c8fedcc71a8c39b803b10. >> There is trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is >> also a trace function. >> >> Fixes: be769db2f958 (net: Get rid of consume_skb when tracing is off) >> Signed-off-by: Yajun Deng <yajun.deng@linux.dev> > > Please explain in more detail why your patch is needed. As it > stands I do not understand the reasoning. > if we don't define CONFIG_TRACEPOINTS, consume_skb() wolud called kfree_skb(), there have trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is also a trace function. So we can trace consume_skb() even if we don't define CONFIG_TRACEPOINTS. This patch "net: Get rid of consume_skb when tracing is off" does not seem to be effective. > Thanks, > -- > Email: Herbert Xu <herbert@gondor.apana.org.au> > Home Page: http://gondor.apana.org.au/~herbert > PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
On Thu, Jul 29, 2021 at 04:01:28AM +0000, yajun.deng@linux.dev wrote: > > if we don't define CONFIG_TRACEPOINTS, consume_skb() wolud called kfree_skb(), there have > trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is also a trace function. So we > can trace consume_skb() even if we don't define CONFIG_TRACEPOINTS. > This patch "net: Get rid of consume_skb when tracing is off" does not seem to be effective. The point of my patch was to get rid of consume_skb because its only purpose is to provide extra information for tracing. If you're not tracing then you don't need that extra information (and overhead). Cheers, -- Email: Herbert Xu <herbert@gondor.apana.org.au> Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
July 29, 2021 12:03 PM, "Herbert Xu" <herbert@gondor.apana.org.au> wrote: > On Thu, Jul 29, 2021 at 04:01:28AM +0000, yajun.deng@linux.dev wrote: > >> if we don't define CONFIG_TRACEPOINTS, consume_skb() wolud called kfree_skb(), there have >> trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is also a trace function. So we >> can trace consume_skb() even if we don't define CONFIG_TRACEPOINTS. >> This patch "net: Get rid of consume_skb when tracing is off" does not seem to be effective. > > The point of my patch was to get rid of consume_skb because its > only purpose is to provide extra information for tracing. If you're > not tracing then you don't need that extra information (and overhead). > Ok, I didn't understand it well. > Cheers, > -- > Email: Herbert Xu <herbert@gondor.apana.org.au> > Home Page: http://gondor.apana.org.au/~herbert > PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index f19190820e63..da897d455d58 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -1072,16 +1072,7 @@ void kfree_skb(struct sk_buff *skb); void kfree_skb_list(struct sk_buff *segs); void skb_dump(const char *level, const struct sk_buff *skb, bool full_pkt); void skb_tx_error(struct sk_buff *skb); - -#ifdef CONFIG_TRACEPOINTS void consume_skb(struct sk_buff *skb); -#else -static inline void consume_skb(struct sk_buff *skb) -{ - return kfree_skb(skb); -} -#endif - void __consume_stateless_skb(struct sk_buff *skb); void __kfree_skb(struct sk_buff *skb); extern struct kmem_cache *skbuff_head_cache; diff --git a/net/core/skbuff.c b/net/core/skbuff.c index fc7942c0dddc..f4c6529ce6b9 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -893,7 +893,6 @@ void skb_tx_error(struct sk_buff *skb) } EXPORT_SYMBOL(skb_tx_error); -#ifdef CONFIG_TRACEPOINTS /** * consume_skb - free an skbuff * @skb: buffer to free @@ -911,7 +910,6 @@ void consume_skb(struct sk_buff *skb) __kfree_skb(skb); } EXPORT_SYMBOL(consume_skb); -#endif /** * __consume_stateless_skb - free an skbuff, assuming it is stateless
This reverts commit be769db2f95861cc8c7c8fedcc71a8c39b803b10. There is trace_kfree_skb() in kfree_skb(), the trace_kfree_skb() is also a trace function. Fixes: be769db2f958 (net: Get rid of consume_skb when tracing is off) Signed-off-by: Yajun Deng <yajun.deng@linux.dev> --- include/linux/skbuff.h | 9 --------- net/core/skbuff.c | 2 -- 2 files changed, 11 deletions(-)