[lttng-dev] [PATCH 1/3] statedump: adjust to v6.7 cpu topology struct

[lttng-dev] [PATCH 1/3] statedump: adjust to v6.7 cpu topology struct

[Bruce Ashfield via lttng-dev]

From: Bruce Ashfield <bruce.ashfield@gmail.com>

Adjusting the statedump to the following upstream commits:

  commit b9655e702dc5 [x86/cpu: Encapsulate topology information in cpuinfo_x86]
  commit e95256335d45 [x86/cpu: Move cpu_core_id into topology info]
  commit 02fb601d27a7 [x86/cpu: Move phys_proc_id into topology info]

We now have the topo struct, as well as some minor member name
changes.

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
---

Someone else may or may not already have this in flight,
but I didn't see anything on the list or in the git
repo.

I ran into this failure when working on the v6.7-rc
kernel as part of yocto project kernel updates.

Tweak as necessary, as I'm no expert in the right way
to version these sort of changes.

Bruce

 include/instrumentation/events/lttng-statedump.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/include/instrumentation/events/lttng-statedump.h b/include/instrumentation/events/lttng-statedump.h
index 642aa356..5f52a29c 100644
--- a/include/instrumentation/events/lttng-statedump.h
+++ b/include/instrumentation/events/lttng-statedump.h
@@ -263,8 +263,13 @@ LTTNG_TRACEPOINT_EVENT(lttng_statedump_cpu_topology,
 		ctf_integer(uint8_t, family, c->x86)
 		ctf_integer(uint8_t, model, c->x86_model)
 		ctf_string(model_name, c->x86_model_id[0] ? c->x86_model_id : "unknown")
+#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
 		ctf_integer(uint16_t, physical_id, c->phys_proc_id)
 		ctf_integer(uint16_t, core_id, c->cpu_core_id)
+#else
+		ctf_integer(uint16_t, physical_id, c->topo.pkg_id)
+		ctf_integer(uint16_t, core_id, c->topo.core_id)
+#endif
 		ctf_integer(uint16_t, cores, c->booted_cores)
 	)
 )
-- 
2.34.1

_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

[lttng-dev] [PATCH 2/3] wrapper/fdtable: adjust fd lookup to v6.7+

[Bruce Ashfield via lttng-dev]

From: Bruce Ashfield <bruce.ashfield@gmail.com>

commit 0ede61d8589cc2d93 [file: convert to SLAB_TYPESAFE_BY_RCU]
renames lookup_fd_rcu to lookup_fdget_rcu, so we need to
version adjust the fdtable wrapper accordingly.

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
---

Repeating the comment from patch 1 of this series:

Someone else may or may not already have this in flight,
but I didn't see anything on the list or in the git
repo.

I ran into this failure when working on the v6.7-rc
kernel as part of yocto project kernel updates.

Tweak as necessary, as I'm no expert in the right way
to version these sort of changes.

Bruce


 include/wrapper/fdtable.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/include/wrapper/fdtable.h b/include/wrapper/fdtable.h
index fa5f7207..aaf2b9e7 100644
--- a/include/wrapper/fdtable.h
+++ b/include/wrapper/fdtable.h
@@ -16,7 +16,11 @@
 static inline
 struct file *lttng_lookup_fd_rcu(unsigned int fd)
 {
+#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
 	return lookup_fd_rcu(fd);
+#else
+	return lookup_fdget_rcu(fd);
+#endif
 }
 #else
 static inline
-- 
2.34.1

_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

[lttng-dev] [PATCH 3/3] vmscan: drop isolate_mode (v6.7+)

[Bruce Ashfield via lttng-dev]

From: Bruce Ashfield <bruce.ashfield@gmail.com>

commit 3dfbb555c98ac55b9 [mm, vmscan: remove ISOLATE_UNMAPPED]
makes the isolate_mode parameter unused, so it is removed.

This updates the event to match.

Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
---

Repeating the comment from patch 1 & 2 of this series:

Someone else may or may not already have this in flight,
but I didn't see anything on the list or in the git
repo.

I ran into this failure when working on the v6.7-rc
kernel as part of yocto project kernel updates.

Tweak as necessary, as I'm no expert in the right way
to version these sort of changes.

Bruce

 include/instrumentation/events/mm_vmscan.h | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/include/instrumentation/events/mm_vmscan.h b/include/instrumentation/events/mm_vmscan.h
index e7e9b613..a3e9ddf2 100644
--- a/include/instrumentation/events/mm_vmscan.h
+++ b/include/instrumentation/events/mm_vmscan.h
@@ -343,12 +343,18 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
 		unsigned long nr_scanned,
 		unsigned long nr_skipped,
 		unsigned long nr_taken,
+#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
 		isolate_mode_t isolate_mode,
+#endif
 		int lru
 	),
 
 	TP_ARGS(classzone_idx, order, nr_requested, nr_scanned, nr_skipped,
-		nr_taken, isolate_mode, lru
+		nr_taken,
+#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
+		isolate_mode,
+#endif
+		lru
 	),
 
 	TP_FIELDS(
@@ -358,7 +364,9 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
 		ctf_integer(unsigned long, nr_scanned, nr_scanned)
 		ctf_integer(unsigned long, nr_skipped, nr_skipped)
 		ctf_integer(unsigned long, nr_taken, nr_taken)
+#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
 		ctf_integer(isolate_mode_t, isolate_mode, isolate_mode)
+#endif
 		ctf_integer(int, lru, lru)
 	)
 )
-- 
2.34.1

_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Re: [lttng-dev] [PATCH 3/3] vmscan: drop isolate_mode (v6.7+)

[Kienan Stewart via lttng-dev]

Hi Bruce,

thanks for the patches! There's a similar series waiting for review: 
https://review.lttng.org/q/topic:%22fix-linux-6.7.0-rc1%22

thanks,
kienan

On 2023-11-23 14:32, Bruce Ashfield via lttng-dev wrote:
> From: Bruce Ashfield <bruce.ashfield@gmail.com>
> 
> commit 3dfbb555c98ac55b9 [mm, vmscan: remove ISOLATE_UNMAPPED]
> makes the isolate_mode parameter unused, so it is removed.
> 
> This updates the event to match.
> 
> Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
> ---
> 
> Repeating the comment from patch 1 & 2 of this series:
> 
> Someone else may or may not already have this in flight,
> but I didn't see anything on the list or in the git
> repo.
> 
> I ran into this failure when working on the v6.7-rc
> kernel as part of yocto project kernel updates.
> 
> Tweak as necessary, as I'm no expert in the right way
> to version these sort of changes.
> 
> Bruce
> 
>   include/instrumentation/events/mm_vmscan.h | 10 +++++++++-
>   1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/include/instrumentation/events/mm_vmscan.h b/include/instrumentation/events/mm_vmscan.h
> index e7e9b613..a3e9ddf2 100644
> --- a/include/instrumentation/events/mm_vmscan.h
> +++ b/include/instrumentation/events/mm_vmscan.h
> @@ -343,12 +343,18 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>   		unsigned long nr_scanned,
>   		unsigned long nr_skipped,
>   		unsigned long nr_taken,
> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>   		isolate_mode_t isolate_mode,
> +#endif
>   		int lru
>   	),
>   
>   	TP_ARGS(classzone_idx, order, nr_requested, nr_scanned, nr_skipped,
> -		nr_taken, isolate_mode, lru
> +		nr_taken,
> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
> +		isolate_mode,
> +#endif
> +		lru
>   	),
>   
>   	TP_FIELDS(
> @@ -358,7 +364,9 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>   		ctf_integer(unsigned long, nr_scanned, nr_scanned)
>   		ctf_integer(unsigned long, nr_skipped, nr_skipped)
>   		ctf_integer(unsigned long, nr_taken, nr_taken)
> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>   		ctf_integer(isolate_mode_t, isolate_mode, isolate_mode)
> +#endif
>   		ctf_integer(int, lru, lru)
>   	)
>   )
_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Re: [lttng-dev] [PATCH 3/3] vmscan: drop isolate_mode (v6.7+)

[Bruce Ashfield via lttng-dev]

Aha,

So the mailing list isn't used for patch submission anymore ? I didn't
notice any updates in the READMEs, etc, so I just fell back to my
existing workflow.

I just want to know, so I won't waste time in the future.

Bruce

On Thu, Nov 23, 2023 at 2:49 PM Kienan Stewart <kstewart@efficios.com> wrote:
>
> Hi Bruce,
>
> thanks for the patches! There's a similar series waiting for review:
> https://review.lttng.org/q/topic:%22fix-linux-6.7.0-rc1%22
>
> thanks,
> kienan
>
> On 2023-11-23 14:32, Bruce Ashfield via lttng-dev wrote:
> > From: Bruce Ashfield <bruce.ashfield@gmail.com>
> >
> > commit 3dfbb555c98ac55b9 [mm, vmscan: remove ISOLATE_UNMAPPED]
> > makes the isolate_mode parameter unused, so it is removed.
> >
> > This updates the event to match.
> >
> > Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
> > ---
> >
> > Repeating the comment from patch 1 & 2 of this series:
> >
> > Someone else may or may not already have this in flight,
> > but I didn't see anything on the list or in the git
> > repo.
> >
> > I ran into this failure when working on the v6.7-rc
> > kernel as part of yocto project kernel updates.
> >
> > Tweak as necessary, as I'm no expert in the right way
> > to version these sort of changes.
> >
> > Bruce
> >
> >   include/instrumentation/events/mm_vmscan.h | 10 +++++++++-
> >   1 file changed, 9 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/instrumentation/events/mm_vmscan.h b/include/instrumentation/events/mm_vmscan.h
> > index e7e9b613..a3e9ddf2 100644
> > --- a/include/instrumentation/events/mm_vmscan.h
> > +++ b/include/instrumentation/events/mm_vmscan.h
> > @@ -343,12 +343,18 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
> >               unsigned long nr_scanned,
> >               unsigned long nr_skipped,
> >               unsigned long nr_taken,
> > +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
> >               isolate_mode_t isolate_mode,
> > +#endif
> >               int lru
> >       ),
> >
> >       TP_ARGS(classzone_idx, order, nr_requested, nr_scanned, nr_skipped,
> > -             nr_taken, isolate_mode, lru
> > +             nr_taken,
> > +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
> > +             isolate_mode,
> > +#endif
> > +             lru
> >       ),
> >
> >       TP_FIELDS(
> > @@ -358,7 +364,9 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
> >               ctf_integer(unsigned long, nr_scanned, nr_scanned)
> >               ctf_integer(unsigned long, nr_skipped, nr_skipped)
> >               ctf_integer(unsigned long, nr_taken, nr_taken)
> > +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
> >               ctf_integer(isolate_mode_t, isolate_mode, isolate_mode)
> > +#endif
> >               ctf_integer(int, lru, lru)
> >       )
> >   )



-- 
- Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end
- "Use the force Harry" - Gandalf, Star Trek II
_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Re: [lttng-dev] [PATCH 3/3] vmscan: drop isolate_mode (v6.7+)

[Kienan Stewart via lttng-dev]

Hi Bruce,

Others may chime in here as I'm not sure I have the full context off-hand.

On 2023-11-23 14:56, Bruce Ashfield wrote:
> Aha,
> 
> So the mailing list isn't used for patch submission anymore ?

Patches submitted to the mailing list are reviewed. That being said, I 
think it's fair to say that the principal workflow has shifted away from 
the mailing list to https://review.lttng.org - many patches are reviewed 
there and not posted on the mailing list.

I think some patch series also get submitted to the mailing list in 
addition to https://review.lttng.org in order to solicit a wider 
feedback from the community.

> I didn't
> notice any updates in the READMEs, etc, so I just fell back to my
> existing workflow.

You are correct that the README for lttng-modules doesn't make any 
mention of this. Thanks for bringing this up, it should definitely be 
addressed!

I'll go through the READMEs and CONTRIBUTING.md files to make sure they 
indicate the current workflows.

thanks,
kienan

> 
> I just want to know, so I won't waste time in the future.
> 
> Bruce
> 
> On Thu, Nov 23, 2023 at 2:49 PM Kienan Stewart <kstewart@efficios.com> wrote:
>>
>> Hi Bruce,
>>
>> thanks for the patches! There's a similar series waiting for review:
>> https://review.lttng.org/q/topic:%22fix-linux-6.7.0-rc1%22
>>
>> thanks,
>> kienan
>>
>> On 2023-11-23 14:32, Bruce Ashfield via lttng-dev wrote:
>>> From: Bruce Ashfield <bruce.ashfield@gmail.com>
>>>
>>> commit 3dfbb555c98ac55b9 [mm, vmscan: remove ISOLATE_UNMAPPED]
>>> makes the isolate_mode parameter unused, so it is removed.
>>>
>>> This updates the event to match.
>>>
>>> Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
>>> ---
>>>
>>> Repeating the comment from patch 1 & 2 of this series:
>>>
>>> Someone else may or may not already have this in flight,
>>> but I didn't see anything on the list or in the git
>>> repo.
>>>
>>> I ran into this failure when working on the v6.7-rc
>>> kernel as part of yocto project kernel updates.
>>>
>>> Tweak as necessary, as I'm no expert in the right way
>>> to version these sort of changes.
>>>
>>> Bruce
>>>
>>>    include/instrumentation/events/mm_vmscan.h | 10 +++++++++-
>>>    1 file changed, 9 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/instrumentation/events/mm_vmscan.h b/include/instrumentation/events/mm_vmscan.h
>>> index e7e9b613..a3e9ddf2 100644
>>> --- a/include/instrumentation/events/mm_vmscan.h
>>> +++ b/include/instrumentation/events/mm_vmscan.h
>>> @@ -343,12 +343,18 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>>>                unsigned long nr_scanned,
>>>                unsigned long nr_skipped,
>>>                unsigned long nr_taken,
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>>                isolate_mode_t isolate_mode,
>>> +#endif
>>>                int lru
>>>        ),
>>>
>>>        TP_ARGS(classzone_idx, order, nr_requested, nr_scanned, nr_skipped,
>>> -             nr_taken, isolate_mode, lru
>>> +             nr_taken,
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>> +             isolate_mode,
>>> +#endif
>>> +             lru
>>>        ),
>>>
>>>        TP_FIELDS(
>>> @@ -358,7 +364,9 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>>>                ctf_integer(unsigned long, nr_scanned, nr_scanned)
>>>                ctf_integer(unsigned long, nr_skipped, nr_skipped)
>>>                ctf_integer(unsigned long, nr_taken, nr_taken)
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>>                ctf_integer(isolate_mode_t, isolate_mode, isolate_mode)
>>> +#endif
>>>                ctf_integer(int, lru, lru)
>>>        )
>>>    )
> 
> 
> 
_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Re: [lttng-dev] [PATCH 3/3] vmscan: drop isolate_mode (v6.7+)

[Kienan Stewart via lttng-dev]

Hi Bruce,

just a quick follow-up to say that the fixes for building against Linux 
6.7 have been merged into the master and stable-2.13 branches of 
lttng-modules.

thanks,
kienan

On 2023-11-23 14:56, Bruce Ashfield wrote:
> Aha,
> 
> So the mailing list isn't used for patch submission anymore ? I didn't
> notice any updates in the READMEs, etc, so I just fell back to my
> existing workflow.
> 
> I just want to know, so I won't waste time in the future.
> 
> Bruce
> 
> On Thu, Nov 23, 2023 at 2:49 PM Kienan Stewart <kstewart@efficios.com> wrote:
>>
>> Hi Bruce,
>>
>> thanks for the patches! There's a similar series waiting for review:
>> https://review.lttng.org/q/topic:%22fix-linux-6.7.0-rc1%22
>>
>> thanks,
>> kienan
>>
>> On 2023-11-23 14:32, Bruce Ashfield via lttng-dev wrote:
>>> From: Bruce Ashfield <bruce.ashfield@gmail.com>
>>>
>>> commit 3dfbb555c98ac55b9 [mm, vmscan: remove ISOLATE_UNMAPPED]
>>> makes the isolate_mode parameter unused, so it is removed.
>>>
>>> This updates the event to match.
>>>
>>> Signed-off-by: Bruce Ashfield <bruce.ashfield@gmail.com>
>>> ---
>>>
>>> Repeating the comment from patch 1 & 2 of this series:
>>>
>>> Someone else may or may not already have this in flight,
>>> but I didn't see anything on the list or in the git
>>> repo.
>>>
>>> I ran into this failure when working on the v6.7-rc
>>> kernel as part of yocto project kernel updates.
>>>
>>> Tweak as necessary, as I'm no expert in the right way
>>> to version these sort of changes.
>>>
>>> Bruce
>>>
>>>    include/instrumentation/events/mm_vmscan.h | 10 +++++++++-
>>>    1 file changed, 9 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/instrumentation/events/mm_vmscan.h b/include/instrumentation/events/mm_vmscan.h
>>> index e7e9b613..a3e9ddf2 100644
>>> --- a/include/instrumentation/events/mm_vmscan.h
>>> +++ b/include/instrumentation/events/mm_vmscan.h
>>> @@ -343,12 +343,18 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>>>                unsigned long nr_scanned,
>>>                unsigned long nr_skipped,
>>>                unsigned long nr_taken,
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>>                isolate_mode_t isolate_mode,
>>> +#endif
>>>                int lru
>>>        ),
>>>
>>>        TP_ARGS(classzone_idx, order, nr_requested, nr_scanned, nr_skipped,
>>> -             nr_taken, isolate_mode, lru
>>> +             nr_taken,
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>> +             isolate_mode,
>>> +#endif
>>> +             lru
>>>        ),
>>>
>>>        TP_FIELDS(
>>> @@ -358,7 +364,9 @@ LTTNG_TRACEPOINT_EVENT(mm_vmscan_lru_isolate,
>>>                ctf_integer(unsigned long, nr_scanned, nr_scanned)
>>>                ctf_integer(unsigned long, nr_skipped, nr_skipped)
>>>                ctf_integer(unsigned long, nr_taken, nr_taken)
>>> +#if (LTTNG_LINUX_VERSION_CODE < LTTNG_KERNEL_VERSION(6,7,0))
>>>                ctf_integer(isolate_mode_t, isolate_mode, isolate_mode)
>>> +#endif
>>>                ctf_integer(int, lru, lru)
>>>        )
>>>    )
> 
> 
> 
_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev