Re: [PATCH v15 0/4] trace: Introduce objtrace trigger to trace the kernel object

[Steven Rostedt]

On Tue, 31 Jan 2023 17:55:16 +0800
Jeff Xie <xiehuan09@xxxxxxxxx> wrote:

>  Hi Steve,
> 
> On Sat, Dec 17, 2022 at 4:27 PM Jeff Xie <xiehuan09@xxxxxxxxx> wrote:
> >
> > On Wed, Nov 30, 2022 at 12:30 AM Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:  
> > >
> > > On Wed, 30 Nov 2022 00:09:51 +0800
> > > Jeff Xie <xiehuan09@xxxxxxxxx> wrote:
> > >  
> > > > > Now instead of searching the function arguments for "obj", we should just
> > > > > read the obj value at every function and report when it changed, as well as
> > > > > the last function it was still the same. The report will be like the
> > > > > "func_repeats" entry. And instead we can print when the object changes,
> > > > > something like this:
> > > > >
> > > > >               cat-117     [002] ...2.     1.602245: bio_add_page <-ext4_mpage_readpages object:0xffff88811bee4000 value:0x2000 (last value:0x1000 at [002] __bio_add_page <-bio_add_page ts: 1.602245)  
> > > >
> > > >
> > > > I'm just curious if we'll see this rewritten version in the next merge
> > > > window  ;-)  
> > >
> > > Unfortunately, this ended up getting dropped in priority, as my workload
> > > increased internally. :-/  
> >  
> 
> I would like to ask if there is any hope that this patchset can be
> merged into the linux kernel in the end.
> 
> We spent a lot of time on this patchset, I don't want to give up so easily ;-)
> 

I think we were near the end, but it wasn't quite done, as I want it to be
integrated with the kprobes indirection interface:

  https://lore.kernel.org/all/20220924160136.5029e942@xxxxxxxxxxxxxxxxxxxx/

Which I started to work on but didn't have time to finish, nor even get
very far with it.

I think once we can get that implemented, then this can get in. I don't
want to give up on it either.

-- Steve