[net-next] perf trace: fix MSG_SPLICE_PAGES build error

  Our MPTCP CI and Stephen got this error:

    In file included from builtin-trace.c:907:
    trace/beauty/msg_flags.c: In function 'syscall_arg__scnprintf_msg_flags':
    trace/beauty/msg_flags.c:28:21: error: 'MSG_SPLICE_PAGES' undeclared (first use in this function)
       28 |         if (flags & MSG_##n) {           |                     ^~~~
    trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
       50 |         P_MSG_FLAG(SPLICE_PAGES);
          |         ^~~~~~~~~~
    trace/beauty/msg_flags.c:28:21: note: each undeclared identifier is reported only once for each function it appears in
       28 |         if (flags & MSG_##n) {           |                     ^~~~
    trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
       50 |         P_MSG_FLAG(SPLICE_PAGES);
          |         ^~~~~~~~~~

The fix is similar to what was done with MSG_FASTOPEN: the new macro is
defined if it is not defined in the system headers.

Fixes: b848b26c6672 ("net: Kill MSG_SENDPAGE_NOTLAST")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: https://lore.kernel.org/r/20230626112847.2ef3d422@canb.auug.org.au/
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
---

Notes:
    @David: I solved it like that in MPTCP tree. Does it work for you too?

    I guess tools/perf/trace/beauty/include/linux/socket.h file still needs
    to be updated, not just to add MSG_SPLICE_PAGES but also other
    modifications done in this file. Maybe best to sync with Arnaldo because
    he might do it soon during the coming merge window I guess.

Cc: David Howells <dhowells@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>

 tools/perf/trace/beauty/msg_flags.c | 3 +++
 1 file changed, 3 insertions(+)


base-commit: 9ae440b8fdd6772b6c007fa3d3766530a09c9045
  

On Mon, 26 Jun 2023 11:02:39 +0200 Matthieu Baerts wrote:
> Our MPTCP CI and Stephen got this error:
> 
>     In file included from builtin-trace.c:907:
>     trace/beauty/msg_flags.c: In function 'syscall_arg__scnprintf_msg_flags':
>     trace/beauty/msg_flags.c:28:21: error: 'MSG_SPLICE_PAGES' undeclared (first use in this function)
>        28 |         if (flags & MSG_##n) {           |                     ^~~~
>     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
>        50 |         P_MSG_FLAG(SPLICE_PAGES);
>           |         ^~~~~~~~~~
>     trace/beauty/msg_flags.c:28:21: note: each undeclared identifier is reported only once for each function it appears in
>        28 |         if (flags & MSG_##n) {           |                     ^~~~
>     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
>        50 |         P_MSG_FLAG(SPLICE_PAGES);
>           |         ^~~~~~~~~~
> 
> The fix is similar to what was done with MSG_FASTOPEN: the new macro is
> defined if it is not defined in the system headers.
> 
> Fixes: b848b26c6672 ("net: Kill MSG_SENDPAGE_NOTLAST")
> Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
> Closes: https://lore.kernel.org/r/20230626112847.2ef3d422@canb.auug.org.au/
> Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
> ---
> 
> Notes:
>     @David: I solved it like that in MPTCP tree. Does it work for you too?
> 
>     I guess tools/perf/trace/beauty/include/linux/socket.h file still needs
>     to be updated, not just to add MSG_SPLICE_PAGES but also other
>     modifications done in this file. Maybe best to sync with Arnaldo because
>     he might do it soon during the coming merge window I guess.
> 
> Cc: David Howells <dhowells@redhat.com>
> Cc: Arnaldo Carvalho de Melo <acme@redhat.com>

Hi Arnaldo, are you okay with us taking this into the networking tree?
Or do you prefer to sync the header after everything lands in Linus's
tree?
  

Hello,

On Mon, Jun 26, 2023 at 2:27 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 26 Jun 2023 11:02:39 +0200 Matthieu Baerts wrote:
> > Our MPTCP CI and Stephen got this error:
> >
> >     In file included from builtin-trace.c:907:
> >     trace/beauty/msg_flags.c: In function 'syscall_arg__scnprintf_msg_flags':
> >     trace/beauty/msg_flags.c:28:21: error: 'MSG_SPLICE_PAGES' undeclared (first use in this function)
> >        28 |         if (flags & MSG_##n) {           |                     ^~~~
> >     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
> >        50 |         P_MSG_FLAG(SPLICE_PAGES);
> >           |         ^~~~~~~~~~
> >     trace/beauty/msg_flags.c:28:21: note: each undeclared identifier is reported only once for each function it appears in
> >        28 |         if (flags & MSG_##n) {           |                     ^~~~
> >     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
> >        50 |         P_MSG_FLAG(SPLICE_PAGES);
> >           |         ^~~~~~~~~~
> >
> > The fix is similar to what was done with MSG_FASTOPEN: the new macro is
> > defined if it is not defined in the system headers.
> >
> > Fixes: b848b26c6672 ("net: Kill MSG_SENDPAGE_NOTLAST")
> > Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
> > Closes: https://lore.kernel.org/r/20230626112847.2ef3d422@canb.auug.org.au/
> > Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
> > ---
> >
> > Notes:
> >     @David: I solved it like that in MPTCP tree. Does it work for you too?
> >
> >     I guess tools/perf/trace/beauty/include/linux/socket.h file still needs
> >     to be updated, not just to add MSG_SPLICE_PAGES but also other
> >     modifications done in this file. Maybe best to sync with Arnaldo because
> >     he might do it soon during the coming merge window I guess.
> >
> > Cc: David Howells <dhowells@redhat.com>
> > Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
>
> Hi Arnaldo, are you okay with us taking this into the networking tree?
> Or do you prefer to sync the header after everything lands in Linus's
> tree?

Arnaldo is on vacation now, and I'm taking care of the patches
on behalf of him.

As it's introduced in the networking tree, it should be fine to
carry the fix together.  I'll sync the header later.

But in general you don't need to change the copy of the tools
headers together.  It also needs to support old & new kernels
so different care should be taken.  Please separate tooling
changes and let us handle them.

Thanks,
Namhyung
  

On Mon, 26 Jun 2023 14:41:56 -0700 Namhyung Kim wrote:
> > Hi Arnaldo, are you okay with us taking this into the networking tree?
> > Or do you prefer to sync the header after everything lands in Linus's
> > tree?  
> 
> Arnaldo is on vacation now, and I'm taking care of the patches
> on behalf of him.
> 
> As it's introduced in the networking tree, it should be fine to
> carry the fix together.  I'll sync the header later.

Will do, thanks!

> But in general you don't need to change the copy of the tools
> headers together.  It also needs to support old & new kernels
> so different care should be taken.  Please separate tooling
> changes and let us handle them.

Ack, I'm not sure what makes this a special case, from Stephen's
original report:

https://lore.kernel.org/all/20230626112847.2ef3d422@canb.auug.org.au/

it sounded like perf won't build without the fix.
  

Hi Jakub, Namhyung,

@Namhyung: thank you for the explanations about the header files in the other thread!


26 Jun 2023 23:49:36 Jakub Kicinski <kuba@kernel.org>:
> On Mon, 26 Jun 2023 14:41:56 -0700 Namhyung Kim wrote:
>
>> But in general you don't need to change the copy of the tools
>> headers together.  It also needs to support old & new kernels
>> so different care should be taken.  Please separate tooling
>> changes and let us handle them.
>
> Ack, I'm not sure what makes this a special case, from Stephen's
> original report:
>
> https://lore.kernel.org/all/20230626112847.2ef3d422@canb.auug.org.au/
> > it sounded like perf won't build without the fix.

The copy of the header file is confusing :)

If I understood properly, here, we need the modification in msg_flags.c from Perf code because this file has been recently modified by David's series in net-next to support a new flag (not present in the system header files yet). Without this patch, Perf doesn't build.

Now regarding the "internal" copy of header files: we don't need to modify it because it is just used by Perf maintainers to monitor the modifications in the original file. If I'm not mistaken, it might even be better to revert the line that has been removed in David's original series in net-next in this header file. But now that Namhyung is aware of that modification, I guess the revert is probably not needed. But in short, it is then better not to modify this header file in the networking tree :-)

Cheers,
Matt
--
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
  

Hello:

This patch was applied to netdev/net-next.git (main)
by Paolo Abeni <pabeni@redhat.com>:

On Mon, 26 Jun 2023 11:02:39 +0200 you wrote:
> Our MPTCP CI and Stephen got this error:
> 
>     In file included from builtin-trace.c:907:
>     trace/beauty/msg_flags.c: In function 'syscall_arg__scnprintf_msg_flags':
>     trace/beauty/msg_flags.c:28:21: error: 'MSG_SPLICE_PAGES' undeclared (first use in this function)
>        28 |         if (flags & MSG_##n) {           |                     ^~~~
>     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
>        50 |         P_MSG_FLAG(SPLICE_PAGES);
>           |         ^~~~~~~~~~
>     trace/beauty/msg_flags.c:28:21: note: each undeclared identifier is reported only once for each function it appears in
>        28 |         if (flags & MSG_##n) {           |                     ^~~~
>     trace/beauty/msg_flags.c:50:9: note: in expansion of macro 'P_MSG_FLAG'
>        50 |         P_MSG_FLAG(SPLICE_PAGES);
>           |         ^~~~~~~~~~
> 
> [...]

Here is the summary with links:
  - [net-next] perf trace: fix MSG_SPLICE_PAGES build error
    https://git.kernel.org/netdev/net-next/c/2553a5270d6c

You are awesome, thank you!