[DTrace-devel] [PATCH v2 0/4] DTrace TCP provider

[Eugene Loh]

Incidentally, looking at the subject lines for the patch series, I see:

         dtrace: move get_member() to dt_cg.c
         dt_impl: bump number of TSLOTS to 8
         dtrace: add tcp provider
         dtrace: sync dlibs with tcp.d, ip.d and net.d changes

That is, each line is prefixed with a component name.  We do that a lot 
for "test:", but otherwise the practice seems to depend on... the patch 
author?  Anyhow, I claim "dtrace:" is not very useful.  For patch 1, all 
the prefix is saying is "there is movement in the dtrace code base."  I 
think the subject would be more direct without the prefix.  Same for 
patch 2.  For patch 3, if one wanted to use a component prefix, I'd 
think one would use "tcp:".

Historically, it looks like we've added new providers with subject lines 
like this:

     rawfbt: new provider
     Implement the io provider
     Implement the ip provider
     Implement the lockstat provider
     Implement the sched provider (first part)
     Implement the proc provider
     provider: Implement a rawtp provider
     Add a CPC provider
     PID provider implementation
     Add a profile provider
     Implement BEGIN and END probes for the dtrace provider
     Added IO provider support for 4.14 kernels

Everyone has their own style.  Shrug.  No big deal.

Anyhow, personally, I think the prefixes don't add much, especially for 
"dtrace:".  My suggestion is to drop the prefixes, but... your call.

On 6/10/25 09:58, Alan Maguire wrote:

> This series is a first draft of TCP provider support, where the
> probes are implemented via underlying fbt and sdt probes.
>
> Due to the use of the sock/inet_sock_set_state tracepoint, intended
> for ~5.15 kernels and later.  Tried replacing this with
>
> fbt::tcp_set_state:entry
>
> but this misses a few state transitions, so stuck with using
> the tracepoint.
>
> All tests under test/unittest/tcp pass unmodified on an upstream
> (6.15) kernel and 5.15 UEK7U3 kernel.
>
> It implements all documented TCP provider probes:
>
> accept-established, accept-refused, connnect-request,
> connect-established, connect-refused, receive, send,
> state-change
>
> Changes since RFC:
>
> - fixed issues with test failures on UEK7 due to missing
>    SYN_RCV state change
> - moved get_member() to dt_cg.c (patch 1)
>
> Alan Maguire (4):
>    dtrace: move get_member() to dt_cg.c
>    dt_impl: bump number of TSLOTS to 8
>    dtrace: add tcp provider
>    dtrace: sync dlibs with tcp.d, ip.d and net.d changes
>
>   dlibs/aarch64/5.14/ip.d  |   1 -
>   dlibs/aarch64/5.14/net.d |   6 +-
>   dlibs/aarch64/5.14/tcp.d |  52 ++---
>   dlibs/aarch64/5.16/ip.d  |   1 -
>   dlibs/aarch64/5.16/net.d |   6 +-
>   dlibs/aarch64/5.16/tcp.d |  52 ++---
>   dlibs/aarch64/6.1/ip.d   |   1 -
>   dlibs/aarch64/6.1/net.d  |   6 +-
>   dlibs/aarch64/6.1/tcp.d  |  52 ++---
>   dlibs/aarch64/6.10/ip.d  |   1 -
>   dlibs/aarch64/6.10/net.d |   6 +-
>   dlibs/aarch64/6.10/tcp.d |  52 ++---
>   dlibs/x86_64/5.14/ip.d   |   1 -
>   dlibs/x86_64/5.14/net.d  |   6 +-
>   dlibs/x86_64/5.14/tcp.d  |  52 ++---
>   dlibs/x86_64/5.16/ip.d   |   1 -
>   dlibs/x86_64/5.16/net.d  |   6 +-
>   dlibs/x86_64/5.16/tcp.d  |  52 ++---
>   dlibs/x86_64/6.1/ip.d    |   1 -
>   dlibs/x86_64/6.1/net.d   |   6 +-
>   dlibs/x86_64/6.1/tcp.d   |  52 ++---
>   dlibs/x86_64/6.10/ip.d   |   1 -
>   dlibs/x86_64/6.10/net.d  |   6 +-
>   dlibs/x86_64/6.10/tcp.d  |  52 ++---
>   libdtrace/Build          |   2 +
>   libdtrace/dt_cg.c        |  39 ++++
>   libdtrace/dt_cg.h        |   2 +
>   libdtrace/dt_impl.h      |   2 +-
>   libdtrace/dt_prov_ip.c   |  45 +----
>   libdtrace/dt_prov_tcp.c  | 405 +++++++++++++++++++++++++++++++++++++++
>   libdtrace/dt_provider.c  |   1 +
>   libdtrace/dt_provider.h  |   1 +
>   libdtrace/ip.d           |   1 -
>   libdtrace/net.d          |   6 +-
>   libdtrace/tcp.d          |  52 ++---
>   35 files changed, 761 insertions(+), 267 deletions(-)
>   create mode 100644 libdtrace/dt_prov_tcp.c
>