Re: [PATCH v6 42/42] x86/resctrl: Add python script to move resctrl code to /fs/resctrl

[James Morse]

Hi Reinette,

On 25/02/2025 16:16, Reinette Chatre wrote:
> On 2/19/25 10:10 PM, Reinette Chatre wrote:
>> On 2/7/25 10:18 AM, James Morse wrote:
>>> To support more than one architecture resctrl needs to move from arch/x86
>>> to live under fs. Moving all the code breaks any series on the mailing
>>> list, so needs scheduling carefully.
>>>
>>> Maintaining the patch that moves all this code has proved labour intensive.
>>> It's also near-impossible to review that no inadvertent changes have
>>> crept in.
>>>
>>> To solve these problems, temporarily add a hacky python program that
>>> lists all the functions that should move, and those that should stay.
>>>
>>> No attempt to parse C code is made, this thing tries to name 'blocks'
>>> based on hueristics about the kernel coding style. It's fragile, but
>>
>> (heuristics)
>>
>>> good enough for its single use here.
>>>
>>> This only exists to show I have nothing up my sleeve.
>>> I don't suggested this gets merged.
>>>
>>> The patch this script generaets has the following corner cases:
>> (generates)
>>
>>> * The original files are regenerated, which will add newlines that are
>>>   not present in the original file.
>>> * An trace-point header file the only contains boiler-plate is created
>>>   in the arch and filesystem code. The parser doesn't know how to remove
>>>   the includes for these - but its easy to 'keep' the file contents on
>>>   the correct side. A follow-up patch will remove these files and their
>>>   includes.
>>
>> Related to the tracepoints I also noticed that there are some leftover
>> tracing defines in files that no longer make use of tracing.
>> For example, arch/x86/kernel/cpu/resctrl/monitor.c contains:
>> #define CREATE_TRACE_POINTS
>> #include monitor_trace.h
>>
>> and fs/resctrl/pseudo_lock.c contains:
>> #define CREATE_TRACE_POINTS
>> #include "pseudo_lock_trace.h"
>>
>> I assumed this will also be removed in this follow-up patch?

Yup:
https://web.git.kernel.org/pub/scm/linux/kernel/git/morse/linux.git/commit/?h=mpam/snapshot/v6.14-rc1&id=3d0430324a0c7e7ad765140f9e78a9a312a13573

I'll include this patch in v7, you found a case where its not as harmless as I thought.


>>> * asm/cpu_device_id.h and a relative path for 'X86_CONFIG()' are kept
>>>   in the filesystem code to ensure x86 builds. A follow-up patch will
>>>   remove these.
>>> * This script doesn't know how to move the documentation, and update the
>>>   links in comments. A follow-up patch does this.
>>
>> One unexpected thing I noticed is that fs/resctr/internal.h contains:
>> #ifndef _ASM_X86_RESCTRL_INTERNAL_H
>> #define _ASM_X86_RESCTRL_INTERNAL_H
>> ...
>> #endif /* _ASM_X86_RESCTRL_INTERNAL_H */

That's a new one - I'll add a follow-up patch to change those.


> It looks like another item for this list of "corner cases" is that the
> #include of all files need to be reviewed after the code move. There are
> functions depending on a particular #include that are moved out of the .c
> file but the (no longer needed) #include remains.

Indeed, that is one of the followups that I'll include in v7.
https://web.git.kernel.org/pub/scm/linux/kernel/git/morse/linux.git/commit/?h=mpam/snapshot/v6.14-rc1&id=9e2bd53f5e2b33fef69db1aae2dd7aeeaf1dd24c

I suggest all these get merged into the patch that moves the code - but I'll post them
separately in case anyone is interested in regenerating the patch using this script.


Thanks,

James