From: David Smith <dsmith@redhat.com>
To: Arkady <arkady.miasnikov@gmail.com>
Cc: systemtap@sourceware.org
Subject: Re: Is there a roadmap for the STAP development?
Date: Thu, 16 Feb 2017 19:38:00 -0000 [thread overview]
Message-ID: <b63f928f-9cfe-95d2-d69a-fc5aed37aaf5@redhat.com> (raw)
In-Reply-To: <CANA-60oZ7QknOzLKxG0Re0Y8KmCof16_JSssmYhpH2oGNyyPbw@mail.gmail.com>
On 02/14/2017 12:00 PM, Arkady wrote:
> On Tue, Feb 14, 2017 at 6:13 PM, David Smith <dsmith@redhat.com> wrote:
... stuff deleted ...
>>> * Add RPM spec file supporting automatic build of the kernel driver
>>> after kernel updates
>>
>> What do you mean by "the kernel driver"?
>
> I need to build the module automatically every time the kernel
> updates. I am preparing the RPM/DEB which does the trick, but to have
> a generic example in /docs/examples would be nice or even a script
> which generates such RPM based on the stap command line options.
Hmm, interesting. This sounds a bit like the kmod system that rebuilds
external kernel modules when the kernel gets updated.
>>> * In the _stp_reserve_bytes() return indication that the trace buffer
>>> is overwritten
>>> * Add support for zero copy data transfer
>>
>> You'd have to flesh those last two items out a bit for me to understand
>> what you are looking to do.
>
> When I send the events from the stap module to the user space I am doing
> event_type* event = _stp_reserve_bytes(sizeof(event_type));
> event->field1 = value1;
> .....
> I do not handle case when _stp_reserve_bytes() starts to overwrite the
> data in the trace buffer. It would be cool to have some feedback that
> buffer is full or is getting full. In some situations it is better to
> drop the event and avoid overwriting the buffer.
As a side note I'll state that you seem to be depending on internal
systemtap interfaces here by using _stp_reserve_bytes(). In general we
try to keep "external" interfaces, like tapsets, compatible between
systemtap versions. There is no such guarantee for internal interfaces.
(That being said, I know of no plans to change _stp_reserve_bytes().)
> In another approach I am planning to use the shared memory for sending
> events from the kernel to the user space in one specific case - very
> frequent system calls. Similarly to transport.c with zero copy API.
Interesting. Yep, that code would be interesting to see.
--
David Smith
dsmith@redhat.com
Red Hat
http://www.redhat.com
256.217.0141 (direct)
256.837.0057 (fax)
next prev parent reply other threads:[~2017-02-16 19:38 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-14 6:52 Arkady
2017-02-14 16:13 ` David Smith
2017-02-14 18:00 ` Arkady
2017-02-16 19:38 ` David Smith [this message]
2017-02-17 8:48 ` Arkady
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b63f928f-9cfe-95d2-d69a-fc5aed37aaf5@redhat.com \
--to=dsmith@redhat.com \
--cc=arkady.miasnikov@gmail.com \
--cc=systemtap@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).