Re: [PATCH v8 17/21] x86,fs/resctrl: Move the resctrl filesystem code to live in /fs/resctrl

[James Morse]

Hi Fenghua,

On 15/04/2025 01:27, Fenghua Yu wrote:
> On 4/11/25 09:42, James Morse wrote:
>> Resctrl is a filesystem interface to hardware that provides cache
>> allocation policy and bandwidth control for groups of tasks or CPUs.
>>
>> To support more than one architecture, resctrl needs to live in /fs/.
>>
>> Move the code that is concerned with the filesystem interface to
>> /fs/resctrl.

>> diff --git a/fs/resctrl/pseudo_lock_trace.h b/fs/resctrl/pseudo_lock_trace.h
>> index e69de29bb2d1..7a6a1983953a 100644
>> --- a/fs/resctrl/pseudo_lock_trace.h
>> +++ b/fs/resctrl/pseudo_lock_trace.h
>> @@ -0,0 +1,17 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +#undef TRACE_SYSTEM
>> +#define TRACE_SYSTEM resctrl
>> +
>> +#if !defined(_X86_RESCTRL_PSEUDO_LOCK_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
>> +#define _X86_RESCTRL_PSEUDO_LOCK_TRACE_H
>> +
>> +#include <linux/tracepoint.h>
>> +
>> +#endif /* _X86_RESCTRL_PSEUDO_LOCK_TRACE_H */
>> +
>> +#undef TRACE_INCLUDE_PATH
>> +#define TRACE_INCLUDE_PATH .
>> +
>> +#define TRACE_INCLUDE_FILE pseudo_lock_trace
>> +
>> +#include <trace/define_trace.h>

> If applying patch 1-17, building the kernel reports this error, which will cause bisect
> failure:
> 
>   CC      fs/resctrl/pseudo_lock.o
> In file included from fs/resctrl/pseudo_lock_trace.h:17,
>                  from fs/resctrl/pseudo_lock.c:35:
> ./include/trace/define_trace.h:110:42: fatal error: ./pseudo_lock_trace.h: No such file or
> directory
>   110 | #include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
>       |                                          ^
> compilation terminated.
> make[4]: *** [scripts/Makefile.build:203: fs/resctrl/pseudo_lock.o] Error 1
> 
> Again combining patch 18 with this patch together makes the error go away.
> 
> But you said combining them together may cause the conversion python tool complex:
> https://lore.kernel.org/lkml/97aeaf87-b2db-4efb-9d81-43769c6f27b0@xxxxxxx/
> 
> Is there a better way to resolve this error? Maybe update the tool not to generate or
> include fs/resctrl/pseudo_lock_trace.h at all?

Please see my previous description of this, also described in the cover letter:
| Patches X-Y should be squashed together when merged - they are posted like
| this to allow folk to re-generate patch N, then review the differences on
| op. Not squashing them together would expose a ftrace build warning
| during bisect. (but who does that!)
| That would look like this:
| git://git.kernel.org/pub/scm/linux/kernel/git/morse/linux.git mpam/move_to_fs/v8_final

Amusingly, I forgot to fix 'X-Y', that should read as 17-21.

I can post that 'final' version with everything squashed together once patches 18-21 have
been reviewed - but doing that before makes it harder to review the deliberate changes
separately from the mechanical changes. This patch is just the mechanical changes.


Thanks,

James