Skip to contentSkip to conversation

checkpatch: add --fix option for INCLUDE_LINUX

PATCH In review Dwaipayan Ray 2 participants 5 messages Nov 21, 2020 Checkpatch

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 checkpatch: add --fix option for INCLUDE_LINUX

Provide fix option to INCLUDE_LINUX check to replace asm includes.

Macros of type:
#include <asm/percpu.h>

are corrected to:
#include <linux/percpu.h>

---
 scripts/checkpatch.pl | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
scripts/checkpatch.pl +5 -2
index 0da6422cd0fd..f852f130e065 100755
@@ -5468,8 +5468,11 @@ sub process {
 						CHK("ARCH_INCLUDE_LINUX",
 						    "Consider using #include <linux/$file> instead of <asm/$file>\n" . $herecurr);
 					} else {
-						WARN("INCLUDE_LINUX",
-						     "Use #include <linux/$file> instead of <asm/$file>\n" . $herecurr);
+						if (WARN("INCLUDE_LINUX",
+							 "Use #include <linux/$file> instead of <asm/$file>\n" . $herecurr) &&
+						    $fix) {
+							$fixed[$fixlinenr] =~ s/\<asm\/$file\>/\<linux\/$file\>/;
+						}
 					}
 				}
 			}
Signature · 6 lines
2.27.0

_______________________________________________
Linux-kernel-mentees mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees
#1 lore raw
Quoted 10 lines · Dwaipayan Ray

Provide fix option to INCLUDE_LINUX check to replace asm includes.

Macros of type:
#include <asm/percpu.h>

are corrected to:
#include <linux/percpu.h>

The patch itself looks okay; but should this fix option not be limited to lines where content is added?

Further we need to investigate:
- is this still a valid rule?
- is this rule and the rationale documented somewhere?
- how many rule violations currently exist in the kernel repository?
- are there known exceptions for some files in some directories?

Lukas

Quoted 24 lines
---
 scripts/checkpatch.pl | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
scripts/checkpatch.pl +5 -3
index 0da6422cd0fd..f852f130e065 100755
@@ -5468,8 +5468,11 @@ sub process {
                                                CHK("ARCH_INCLUDE_LINUX",
                                                    "Consider using #include <linux/$file> instead of <asm/$file>\n" . $herecurr);
                                        } else {
-                                               WARN("INCLUDE_LINUX",
-                                                    "Use #include <linux/$file> instead of <asm/$file>\n" . $herecurr);
+                                               if (WARN("INCLUDE_LINUX",
+                                                        "Use #include <linux/$file> instead of <asm/$file>\n" . $herecurr) &&
+                                                   $fix) {
+                                                       $fixed[$fixlinenr] =~ s/\<asm\/$file\>/\<linux\/$file\>/;
+                                               }
                                        }
                                }
                        }
--
Quoted 2 lines

2.27.0

_______________________________________________
Linux-kernel-mentees mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees

#2 lore raw
Quoted 9 lines · Lukas Bulwahn
Quoted 7 lines · Dwaipayan Ray

Provide fix option to INCLUDE_LINUX check to replace asm includes.

Macros of type:
#include <asm/percpu.h>

Quoted 7 lines
Quoted 3 lines

are corrected to:
#include <linux/percpu.h>

The patch itself looks okay; but should this fix option not be limited to lines where content is added?

Hi,
I think the warning itself is emitted for lines which are added. So I think the fix option even now will change those added lines only.

Quoted 3 lines

Further we need to investigate:
- is this still a valid rule?
- is this rule and the rationale documented somewhere?

I think the rule is valid. I was going through the documentation files and in several examples I saw that the linux headers are preferred.

Like in Documentation/core-api/local_ops.rst:

91:
#include <linux/percpu.h>
#include <asm/local.h>

In Documentation/trace/kprobes.rst:

362:
#include <linux/kprobes.h>
int register_kprobe(struct kprobe *kp);

400:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int pre_handler(struct kprobe *p, struct pt_regs *regs);

410:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
void post_handler(struct kprobe *p, struct pt_regs *regs,
unsigned long flags);

and so on...

I didn't find any explicit mention about absolutely preffering linux headers over asm headers whenever it exists.

Quoted 2 lines

- how many rule violations currently exist in the kernel repository?
- are there known exceptions for some files in some directories?

There are exceptions to this.
Like in include/linux/percpu.h,
#include <asm/percpu.h> needs to be included. as linux/percpu wraps over asm/percpu. So these kinds are exempted from the check and checkpatch doesn't report any warning.

I did find some violations like in arch/csky/kernel/probes/decode-insn.h:
CHECK: Consider using #include <linux/kprobes.h> instead of <asm/kprobes.h> #7: FILE: arch/csky/kernel/probes/decode-insn.h:7:
+#include <asm/kprobes.h>

I found one commit which fixed one such case in the kernel:
commit ca5999fde0a17 ("mm: introduce include/linux/pgtable.h") It says "The include/linux/pgtable.h is going to be the home of generic page table
manipulation functions."

So I think it means that future additings should use <linux/pgtable.h> instead of
<asm/pgtable.h>

I think the ARCH_INCLUDE_LINUX also deserves a --fix option similarly.

What do you think? Should I continue with this?

Thanks,
Dwaipayan.
_______________________________________________
Linux-kernel-mentees mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees

#3 lore raw
Quoted 78 lines · Dwaipayan Ray
Quoted 9 lines · Lukas Bulwahn
Quoted 7 lines · Dwaipayan Ray

Provide fix option to INCLUDE_LINUX check to replace asm includes.

Macros of type:
#include <asm/percpu.h>

Quoted 7 lines
Quoted 3 lines

are corrected to:
#include <linux/percpu.h>

The patch itself looks okay; but should this fix option not be limited to lines where content is added?

Hi,
I think the warning itself is emitted for lines which are added. So I think the fix option even now will change those added lines only.

Quoted 3 lines

Further we need to investigate:
- is this still a valid rule?
- is this rule and the rationale documented somewhere?

I think the rule is valid. I was going through the documentation files and in several examples I saw that the linux headers are preferred.

Like in Documentation/core-api/local_ops.rst:

91:

   #include <linux/percpu.h>
   #include <asm/local.h>

In Documentation/trace/kprobes.rst:

362:
#include <linux/kprobes.h>
int register_kprobe(struct kprobe *kp);

400:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int pre_handler(struct kprobe *p, struct pt_regs *regs);

410:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
void post_handler(struct kprobe *p, struct pt_regs *regs,
unsigned long flags);

and so on...

I didn't find any explicit mention about absolutely preffering linux headers over asm headers whenever it exists.

Quoted 2 lines

- how many rule violations currently exist in the kernel repository?
- are there known exceptions for some files in some directories?

There are exceptions to this.
Like in include/linux/percpu.h,
#include <asm/percpu.h> needs to be included. as linux/percpu wraps over asm/percpu. So these kinds are exempted from the check and checkpatch doesn't report any warning.

I did find some violations like in arch/csky/kernel/probes/decode-insn.h:
CHECK: Consider using #include <linux/kprobes.h> instead of <asm/kprobes.h> #7: FILE: arch/csky/kernel/probes/decode-insn.h:7:
+#include <asm/kprobes.h>

I found one commit which fixed one such case in the kernel:
commit ca5999fde0a17 ("mm: introduce include/linux/pgtable.h") It says "The include/linux/pgtable.h is going to be the home of generic page table
manipulation functions."

Quoted 8 lines

So I think it means that future additings should use <linux/pgtable.h> instead of
<asm/pgtable.h>

I think the ARCH_INCLUDE_LINUX also deserves a --fix option similarly.

What do you think? Should I continue with this?

Yes, that is good. Thanks for the investigation.

You can consider to:

- also fix the cases where this rule is violated and send patches to the maintainers.
- add some documentation on this rule.

Lukas
_______________________________________________
Linux-kernel-mentees mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees

#4 lore raw
Quoted 89 lines · Lukas Bulwahn
Quoted 78 lines · Dwaipayan Ray
Quoted 9 lines · Lukas Bulwahn
Quoted 7 lines · Dwaipayan Ray

Provide fix option to INCLUDE_LINUX check to replace asm includes.

Macros of type:
#include <asm/percpu.h>

Quoted 7 lines
Quoted 3 lines

are corrected to:
#include <linux/percpu.h>

The patch itself looks okay; but should this fix option not be limited to lines where content is added?

Hi,
I think the warning itself is emitted for lines which are added. So I think the fix option even now will change those added lines only.

Quoted 3 lines

Further we need to investigate:
- is this still a valid rule?
- is this rule and the rationale documented somewhere?

I think the rule is valid. I was going through the documentation files and in several examples I saw that the linux headers are preferred.

Like in Documentation/core-api/local_ops.rst:

91:

   #include <linux/percpu.h>
   #include <asm/local.h>

In Documentation/trace/kprobes.rst:

362:
#include <linux/kprobes.h>
int register_kprobe(struct kprobe *kp);

400:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
int pre_handler(struct kprobe *p, struct pt_regs *regs);

410:
#include <linux/kprobes.h>
#include <linux/ptrace.h>
void post_handler(struct kprobe *p, struct pt_regs *regs,
unsigned long flags);

and so on...

I didn't find any explicit mention about absolutely preffering linux headers over asm headers whenever it exists.

Quoted 2 lines

- how many rule violations currently exist in the kernel repository?
- are there known exceptions for some files in some directories?

There are exceptions to this.
Like in include/linux/percpu.h,
#include <asm/percpu.h> needs to be included. as linux/percpu wraps over asm/percpu. So these kinds are exempted from the check and checkpatch doesn't report any warning.

I did find some violations like in arch/csky/kernel/probes/decode-insn.h:
CHECK: Consider using #include <linux/kprobes.h> instead of <asm/kprobes.h> #7: FILE: arch/csky/kernel/probes/decode-insn.h:7:
+#include <asm/kprobes.h>

I found one commit which fixed one such case in the kernel:
commit ca5999fde0a17 ("mm: introduce include/linux/pgtable.h") It says "The include/linux/pgtable.h is going to be the home of generic page table
manipulation functions."

Quoted 8 lines

So I think it means that future additings should use <linux/pgtable.h> instead of
<asm/pgtable.h>

I think the ARCH_INCLUDE_LINUX also deserves a --fix option similarly.

What do you think? Should I continue with this?

Quoted 8 lines

Yes, that is good. Thanks for the investigation.

You can consider to:

- also fix the cases where this rule is violated and send patches to the maintainers.
- add some documentation on this rule.

Okay sure I will try doing that.

I am just sending this patch unchanged to Joe and LKML to have their review. If it is ok, I will get on with the next task.

Thanks,
Dwaipayan.
_______________________________________________
Linux-kernel-mentees mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees