Skip to contentSkip to conversation

add a trivial patch style checker

PATCH In review Andy Whitcroft 13 participants 38 messages May 27, 2007 – Jun 6, 2007

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 add a trivial patch style checker

We are seeing increasing levels of minor patch style violations in submissions to the mailing lists as well as making it into the tree. These detract from the quality of the submission and cause unnessary work for reviewers.

As a first step package up the current state of the patch style checker and include it in the kernel tree. Add instructions suggesting running it on submissions. This adds version v0.01 of the patch-trivia-detector.

---

As discussed here is the updated version of the patchstylecheckemail.pl tool. This version incorporates all of Randy Dunlap and my updates. I have also cleaned it up significantly to aid supporting it, overall indent and the like.

I am proposing we handle the support of this tool overall outside kernel tree so we can maintain the regression test suite and surrounding infrastructure for the proposed
automated style robot. Obviously with updates to the
checker itself flowing from there to the kernel tree. To this end I have added a version number to the checker.

Also if either Joel or Randy want to be on on the MAINTAINERS entry yell and we'll get it updated, wouldn't want to list anyone without permission.

I have the infrastructure in place to for the automated robot if and when its deemed ready.

Documentation/SubmittingPatches +28 -10
index a417b25..23637e8 100644
@@ -118,7 +118,21 @@ then only post say 15 or so at a time and wait for review and integration.
 
 
 
-4) Select e-mail destination.
+4) Style check your changes.
+
+Check your patch for basic style violations, details of which can be
+found in Documentation/CodingStyle.  Failure to do so simply wastes
+the reviewers time and will get your patch rejected, probabally
+without even being read.
+
+At a minimum you should check your patches with the patch style
+checker prior to submission (scripts/patch-trivia-detector.pl).
+You should be able to justify all violations that remain in your
+patch.
+
+
+
+5) Select e-mail destination.
 
 Look through the MAINTAINERS file and the source code, and determine
 if your change applies to a specific subsystem of the kernel, with
@@ -146,7 +160,7 @@ discussed should the patch then be submitted to Linus.
 
 
 
-5) Select your CC (e-mail carbon copy) list.
+6) Select your CC (e-mail carbon copy) list.
 
 Unless you have a reason NOT to do so, CC [email protected].
 
@@ -187,8 +201,7 @@ URL: <http://www.kernel.org/pub/linux/kernel/people/bunk/trivial/>
 
 
 
-
-6) No MIME, no links, no compression, no attachments.  Just plain text.
+7) No MIME, no links, no compression, no attachments.  Just plain text.
 
 Linus and other kernel developers need to be able to read and comment
 on the changes you are submitting.  It is important for a kernel
@@ -223,7 +236,7 @@ pref("mailnews.display.disable_format_flowed_support", true);
 
 
 
-7) E-mail size.
+8) E-mail size.
 
 When sending patches to Linus, always follow step #6.
 
@@ -234,7 +247,7 @@ server, and provide instead a URL (link) pointing to your patch.
 
 
 
-8) Name your kernel version.
+9) Name your kernel version.
 
 It is important to note, either in the subject line or in the patch
 description, the kernel version to which this patch applies.
@@ -244,7 +257,7 @@ Linus will not apply it.
 
 
 
-9) Don't get discouraged.  Re-submit.
+10) Don't get discouraged.  Re-submit.
 
 After you have submitted your change, be patient and wait.  If Linus
 likes your change and applies it, it will appear in the next version
@@ -270,7 +283,7 @@ When in doubt, solicit comments on linux-kernel mailing list.
 
 
 
-10) Include PATCH in the subject
+11) Include PATCH in the subject
 
 Due to high e-mail traffic to Linus, and to linux-kernel, it is common
 convention to prefix your subject line with [PATCH].  This lets Linus
@@ -279,7 +292,7 @@ e-mail discussions.
 
 
 
-11) Sign your work
+12) Sign your work
 
 To improve tracking of who did what, especially with patches that can
 percolate to their final resting place in the kernel through several
@@ -328,7 +341,8 @@ now, but you can do this to mark internal company procedures or just
 point out some special detail about the sign-off. 
 
 
-12) The canonical patch format
+
+13) The canonical patch format
 
 The canonical patch subject line is:
 
@@ -427,6 +441,10 @@ section Linus Computer Science 101.
 Nuff said.  If your code deviates too much from this, it is likely
 to be rejected without further review, and without comment.
 
+Check your patches with the patch style checker prior to submission
+(scripts/patch-trivia-detector.pl).  You should be able to justify
+all violations that remain in your patch.
+
 
 
 2) #ifdefs are ugly
MAINTAINERS +10 -2
index d7533dd..b8a5611 100644
@@ -30,8 +30,11 @@ trivial patch so apply some common sense.
 	job the maintainers (and especially Linus) do is to keep things
 	looking the same. Sometimes this means that the clever hack in
 	your driver to get around a problem actually needs to become a
-	generalized kernel feature ready for next time. See
-	Documentation/CodingStyle for guidance here.
+	generalized kernel feature ready for next time.
+
+	PLEASE check your patch with the automated style checker
+	(scripts/patch-trivia-detector.pl) to catch trival style
+	violations.  See Documentation/CodingStyle for guidance here.
 
 	PLEASE try to include any credit lines you want added with the
 	patch. It avoids people being missed off by mistake and makes
@@ -2724,6 +2727,11 @@ L:	[email protected]
 L:	[email protected]
 S:	Supported
 
+PATCH TRIVIA DETECTOR
+P:	Andy Whitcroft
+M:	[email protected]
+S:	Supported
+
 PC87360 HARDWARE MONITORING DRIVER
 P:	Jim Cromie
 M:	[email protected]
scripts/patch-trivia-detector.pl new file +553 -0

553 lines not on this page · raw message

#1 lore raw
Quoted 3 lines · On Sun, 27 May 2007 18:11:25 +0100 Andy Whitcroft wrote:

Also if either Joel or Randy want to be on on the MAINTAINERS entry yell and we'll get it updated, wouldn't want to list anyone without permission.

Yes, please.

Documentation/SubmittingPatches

Quoted 7 lines

I have the infrastructure in place to for the automated robot if and when its deemed ready.

Documentation/SubmittingPatches +0 -0
index a417b25..23637e8 100644

It needs to be added to Documentation/SubmitChecklist also... (but I'm working in yard/house this holiday-not weekend).

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

#2 lore raw
Quoted 11 lines · Randy Dunlap
Quoted 3 lines · On Sun, 27 May 2007 18:11:25 +0100 Andy Whitcroft wrote:

Also if either Joel or Randy want to be on on the MAINTAINERS entry yell and we'll get it updated, wouldn't want to list anyone without permission.

Yes, please.

Quoted 3 lines

I have the infrastructure in place to for the automated robot if and when its deemed ready.

Quoted 7 lines

Documentation/SubmittingPatches

Quoted 4 lines
Documentation/SubmittingPatches +0 -0
index a417b25..23637e8 100644

It needs to be added to Documentation/SubmitChecklist also... (but I'm working in yard/house this holiday-not weekend).

So true.

I've sorted this one out, but there seems to be a number of little changes trickling in so I will roll them all up and send up a V2 pretty soon.

-apw

#3 lore raw
In reply to Randy Dunlap
Quoted 7 lines · Randy Dunlap
Quoted 3 lines · On Sun, 27 May 2007 18:11:25 +0100 Andy Whitcroft wrote:

Also if either Joel or Randy want to be on on the MAINTAINERS entry yell and we'll get it updated, wouldn't want to list anyone without permission.

Yes, please.

Yes. Add me as well.

#4 lore raw
In reply to Andy Whitcroft

Documentation/SubmittingPatches pref("mailnews.display.disable_format_flowed_support", true); · line 241

Quoted 8 lines · Andy Whitcroft
Documentation/SubmittingPatches +1 -1
@@ -223,7 +236,7 @@ pref("mailnews.display.disable_format_flowed_support", true);
 
 
 
-7) E-mail size.
+8) E-mail size.
 
 When sending patches to Linus, always follow step #6.

That's step #7 now.

Andreas.

Signature · 4 lines
Andreas Schwab, SuSE Labs, [email protected]
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."
#6 lore raw
In reply to Andy Whitcroft

scripts/patch-trivia-detector.pl line 426

Quoted 6 lines · On Sun, 27 May 2007 18:11:25 +0100 Andy Whitcroft <[email protected]> wrote:
scripts/patch-trivia-detector.pl new file +6 -0
+#gotos aren't indented
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

This should be "labels". Plus a lot of people do indent the labels by a single space to avoid confusing `diff -p', which seems a reasonable thing to not complain about.

#7 lore raw
Quoted 12 lines · Andrew Morton

scripts/patch-trivia-detector.pl line 426

Quoted 6 lines · On Sun, 27 May 2007 18:11:25 +0100 Andy Whitcroft <[email protected]> wrote:
scripts/patch-trivia-detector.pl new file +6 -0
+#gotos aren't indented
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

This should be "labels". Plus a lot of people do indent the labels by a single space to avoid confusing `diff -p', which seems a reasonable thing to not complain about.

Yep makes sense. Done.

-apw

#8 lore raw
In reply to Andy Whitcroft

scripts/patch-trivia-detector.pl

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +0 -0
new file mode 100755
index 0000000..e71c4ef

Could we please name this checkpatch.pl to follow naming style in scripts/ The namespace check* and clean* and mk* is well established so it is preferred that we stick to it.

The patch-trivia-detector was a nice working name but as it enters the kernel it is better that it follows the surroundings.

Sam

#9 lore raw
In reply to Andy Whitcroft
Quoted 8 lines · Andy Whitcroft
scripts/patch-trivia-detector.pl new file +7 -0
+
+my $P = $0;
+
+my $V = '0.01';
+
+use Getopt::Long qw(:config no_auto_abbrev);
[...]
+sub top_of_kernel_tree {

scripts/patch-trivia-detector.pl line 62

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +9 -0
+	if ((-f "COPYING") && (-f "CREDITS") && (-f "Kbuild") &&
+	    (-f "MAINTAINERS") && (-f "Makefile") && (-f "README") &&
+	    (-d "Documentation") && (-d "arch") && (-d "include") &&
+	    (-d "drivers") && (-d "fs") && (-d "init") && (-d "ipc") &&
+	    (-d "kernel") && (-d "lib") && (-d "scripts")) {
+		return 1;
+	}
+	return 0;
+}
[...]

The consistent style quite looses after this point, esp. space around operators ;-)

Quoted 13 lines
scripts/patch-trivia-detector.pl new file +13 -0
+sub process {
+	my $filename = shift;
+	my @lines = @_;
+
+	my $linenr=0;
+	my $prevline="";
+	my $stashline="";
+
+	my $lineforcounting='';
+	my $indent;
+	my $previndent=0;
+	my $stashindent=0;
+

scripts/patch-trivia-detector.pl line 116

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+	my $clean = 1;
+	my $signoff = 0;
+
+	# Trace the real file/line as we go.
+	my $realfile = '';
+	my $realline = 0;
+	my $realcnt = 0;
+	my $here = '';
+	my $in_comment = 0;
+	my $first_line = 0;

Like in the kernel/C (as far as .bss is concerned), they usually do not need initialization. (They will be different from 0 though, namely undef, but doing ++$x where x is undefined will also DTRT.)

scripts/patch-trivia-detector.pl line 122

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+
+	foreach my $line (@lines) {
+		$linenr++;
+
+#extract the filename as it passes
+		if($line=~/^\+\+\+\s+(\S+)/) {

/^\+{3}\s+(\S+)/ "there's always more than one way to do it in Perl" - meh

scripts/patch-trivia-detector.pl line 128

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+			$realfile=$1;
+			$in_comment = 0;
+			next;
+		}
+#extract the line range in the file after the patch is applied
+		if($line=~/^@@ -\d+,\d+ \+(\d+)(,(\d+))? @@/) {

The @ should be escaped afaicr
/^\@\@\s+-\d+,\d+\s+\+(\d+)(,(\d+))? \@\@/

scripts/patch-trivia-detector.pl line 140

Quoted 12 lines
scripts/patch-trivia-detector.pl new file +12 -0
+			$first_line = 1;
+			$in_comment = 0;
+			$realline=$1-1;
+			if (defined $2) {
+				$realcnt=$3+1;
+			} else {
+				$realcnt=1+1;
+			}
+			next;
+		}
+#check the patch for a signoff:
+		if ($line =~ /^[[:space:]]*Signed-off-by:/i) {

[[:space:]] = \s
Perhaps require a \s after : and get rid of /i? ;-)

scripts/patch-trivia-detector.pl line 144

Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+			$signoff++;
+		}
+#track the line number as we move through the hunk
+		if($line=~/^( |\+)/) {

Or ... /^([ \+])/ (mostly depends on taste)

scripts/patch-trivia-detector.pl line 146

Quoted 2 lines
scripts/patch-trivia-detector.pl new file +2 -0
+			$realline++;
+			$realcnt-- if ($realcnt);

That's a bit ambiguous at first, since it usually means (I think) if(defined($realcnt) || $realcnt != 0).
Using if ($realcnt == 0) could help.

scripts/patch-trivia-detector.pl line 157

Quoted 11 lines
scripts/patch-trivia-detector.pl new file +11 -0
+
+			# track any sort of multi-line comment.  Obviously if
+			# the added text or context do not include the whole
+			# comment we will not see it. Such is life.
+			#
+			# Guestimate if this is a continuing comment.  If this
+			# is the start of a diff block and this line starts
+			# ' *' then it is very likely a comment.
+			if ($first_line and $line =~ m@^.\s*\*@) {
+				$in_comment = 1;
+			}

m@@ and and needed? Could be just

	if ($first_line == 0 && $line =~ /^.\s*\*/)

scripts/patch-trivia-detector.pl line 163

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+			if ($line =~ m@/\*@) {
+				$in_comment = 1;
+			}
+			if ($line =~ m@\*/@) {
+				$in_comment = 0;
+			}

Although not necessary here, I'd like to point out that I've had more luck using paired characters as boundaries, i.e. m{} or s{}{}, which means less escaping for the inner regex in some cases.

scripts/patch-trivia-detector.pl line 189

Quoted 2 lines
scripts/patch-trivia-detector.pl new file +2 -0
+# check we are in a valid source file *.[hc] if not then ignore this hunk
+		next if ($realfile !~ /\.[hc]$/);

Oh I think trailing whitespace and 80 column limit should also be applied to .S and .s files.

scripts/patch-trivia-detector.pl line 199

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+
+#trailing whitespace
+		if($line=~/\S\s+$/) {
+			my $herevet = "$here\n" . cat_vet($line) . "\n\n";
+			print ("trailing whitespace\n");
+			print "$herevet";
+			$clean = 0;
+		}
+#80 column limit
+		if(!($prevline=~/\/\*\*/) && length($lineforcounting) > 80){

Actually, I think this should be "> 79" (after stripping a .diff's control column), since the cursor may move to the 81th column when editing an 80-col line - which is what we want to avoid, no?

scripts/patch-trivia-detector.pl line 207

Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+			print "line over 80 characters\n";
+			print "$herecurr";
+			$clean = 0;
+		}
+
+# at the beginning of a line any tabs must come first and anything
+# more than 8 must use tabs.
+		if($line=~/^\+\s* \t\s*\S/ or $line=~/^\+\s*        \s*/) {

|| will do instead of or.
Somehow this does not catch "Only spaces" lines (/^ +\S/), because there's no \t in them. (Talk about regex fun.)

I'd say split and simplify:

	if ($line =~ /^\s* {8,}/) {
		print "Make that big space a \\t\n";
	}
	if ($line =~ /^ +\t/) {
		print "Inadvertent space at line start, or make it \\t\n";
	}
Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+			my $herevet = "$here\n" . cat_vet($line) . "\n\n";
+			print ("use tabs not spaces\n");
+			print "$herevet";
+			$clean = 0;
+		}
+
+		#
+		# The rest of our checks refer specifically to C style

scripts/patch-trivia-detector.pl line 225

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+		# only apply those _outside_ comments.
+		#
+		next if ($in_comment);
+
+# no C99 // comments
+		if ($line =~ qr|//|) {
+			print "do not use C99 // comments\n";
+			print "$herecurr";
+			$clean = 0;
+		}

(Now we're using qr{}, what about being consistent with m{}?) It may break on rare occassions:

	printk("Hello // World\n");
Quoted 25 lines
scripts/patch-trivia-detector.pl new file +25 -0
+		# Remove comments from the line before processing.
+		$line =~ s@/\*.*\*/@@g;
+		$line =~ s@/\*.*@@;
+		$line =~ s@.*\*/@@;
+		$line =~ s@//.*@@;
+
+#EXPORT_SYMBOL should immediately follow its function closing }.
+		if (($line =~ /EXPORT_SYMBOL.*\(.*\)/) ||
+		    ($line =~ /EXPORT_UNUSED_SYMBOL.*\(.*\)/)) {
+			if (($prevline !~ /^}/) &&
+			   ($prevline !~ /^\+}/) &&
+			   ($prevline !~ /^ }/)) {
+				print "EXPORT_SYMBOL(func); should immediately follow its function\n";
+				print "$herecurr";
+				$clean = 0;
+			}
+		}
+
+		# check for static initialisers.
+		if ($line=~/\s*static\s.*=\s+(0|NULL);/) {
+			print "do not initialise statics to 0 or NULL\n";
+			print "$herecurr";
+			$clean = 0;
+		}
+

scripts/patch-trivia-detector.pl line 261

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+		# check for new typedefs.
+		if ($line=~/\s*typedef\s/) {
+			print "do not add new typedefs\n";
+			print "$herecurr";
+			$clean = 0;
+		}
+
+# * goes on variable not on type
+		if($line=~/[A-Za-z\d_]+\* [A-Za-z\d_]+/) {
+			print ("\"foo* bar\" should be \"foo *bar\"\n"); 

print does not need parentheses in most cases.

Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+			print "$herecurr";
+			$clean = 0;
+		}
+
+# no BUG() or BUG_ON()
+		if ($line =~ /\b(BUG|BUG_ON)\b/) {
+			print "Try to use WARN_ON & Recovery code rather than BUG() or BUG_ON()\n";
+			print "$herecurr";

scripts/patch-trivia-detector.pl line 279

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+			$clean = 0;
+		}
+
+# printk should use KERN_* levels
+		if (($line =~ /\bprintk\b/) &&
+		    ($line !~ /KERN_/)) {
+			print "printk() should include KERN_ facility level\n";
+			print "$herecurr";
+			$clean = 0;
+		}

if ($line =~ /\bprintk\((?!KERN_)/)

Attention span ran out. :)

Jan

#10 lore raw
Quoted 14 lines · Jan Engelhardt
Quoted 13 lines · Andy Whitcroft
scripts/patch-trivia-detector.pl new file +12 -0
+
+my $P = $0;
+
+my $V = '0.01';
+
+use Getopt::Long qw(:config no_auto_abbrev);
[...]
+sub top_of_kernel_tree {
+	if ((-f "COPYING") && (-f "CREDITS") && (-f "Kbuild") &&
+	    (-f "MAINTAINERS") && (-f "Makefile") && (-f "README") &&
+	    (-d "Documentation") && (-d "arch") && (-d "include") &&
+	    (-d "drivers") && (-d "fs") && (-d "init") && (-d "ipc") &&
+	    (-d "kernel") && (-d "lib") && (-d "scripts")) {
Quoted 8 lines

scripts/patch-trivia-detector.pl line 62

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +4 -0
+		return 1;
+	}
+	return 0;
+}
[...]

The consistent style quite looses after this point, esp. space around operators ;-)

Yep thats on my list for cleanup ...

Quoted 19 lines
Quoted 19 lines
scripts/patch-trivia-detector.pl new file +19 -0
+sub process {
+	my $filename = shift;
+	my @lines = @_;
+
+	my $linenr=0;
+	my $prevline="";
+	my $stashline="";
+
+	my $lineforcounting='';
+	my $indent;
+	my $previndent=0;
+	my $stashindent=0;
+
+	my $clean = 1;
+	my $signoff = 0;
+
+	# Trace the real file/line as we go.
+	my $realfile = '';
+	my $realline = 0;
Quoted 8 lines

scripts/patch-trivia-detector.pl line 116

Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+	my $realcnt = 0;
+	my $here = '';
+	my $in_comment = 0;
+	my $first_line = 0;

Like in the kernel/C (as far as .bss is concerned), they usually do not need initialization. (They will be different from 0 though, namely undef, but doing ++$x where x is undefined will also DTRT.)

Well I tend to prefer initialisers as they are the only up front hint as to the notional type of the thing.

Quoted 9 lines

scripts/patch-trivia-detector.pl line 122

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+
+	foreach my $line (@lines) {
+		$linenr++;
+
+#extract the filename as it passes
+		if($line=~/^\+\+\+\s+(\S+)/) {

/^\+{3}\s+(\S+)/ "there's always more than one way to do it in Perl" - meh

Quoted 9 lines

scripts/patch-trivia-detector.pl line 128

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+			$realfile=$1;
+			$in_comment = 0;
+			next;
+		}
+#extract the line range in the file after the patch is applied
+		if($line=~/^@@ -\d+,\d+ \+(\d+)(,(\d+))? @@/) {

The @ should be escaped afaicr
/^\@\@\s+-\d+,\d+\s+\+(\d+)(,(\d+))? \@\@/

The @'s need quoting yes. The spaces are officially spaces though.

Quoted 7 lines
Quoted 7 lines
scripts/patch-trivia-detector.pl new file +7 -0
+			$first_line = 1;
+			$in_comment = 0;
+			$realline=$1-1;
+			if (defined $2) {
+				$realcnt=$3+1;
+			} else {
+				$realcnt=1+1;
Quoted 8 lines

scripts/patch-trivia-detector.pl line 140

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +5 -0
+			}
+			next;
+		}
+#check the patch for a signoff:
+		if ($line =~ /^[[:space:]]*Signed-off-by:/i) {

[[:space:]] = \s
Perhaps require a \s after : and get rid of /i? ;-)

Yeah, split this into a couple of checks implementing the recommended capitalisation and checking for a space after the : etc.

Quoted 6 lines

scripts/patch-trivia-detector.pl line 144

Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+			$signoff++;
+		}
+#track the line number as we move through the hunk
+		if($line=~/^( |\+)/) {

Or ... /^([ \+])/ (mostly depends on taste)

We generally use [] so switched to that.

Quoted 6 lines

scripts/patch-trivia-detector.pl line 146

Quoted 2 lines
scripts/patch-trivia-detector.pl new file +2 -0
+			$realline++;
+			$realcnt-- if ($realcnt);

That's a bit ambiguous at first, since it usually means (I think) if(defined($realcnt) || $realcnt != 0).
Using if ($realcnt == 0) could help.

if ($realcnt != 0); but yes

Quoted 13 lines

scripts/patch-trivia-detector.pl line 157

Quoted 11 lines
scripts/patch-trivia-detector.pl new file +11 -0
+
+			# track any sort of multi-line comment.  Obviously if
+			# the added text or context do not include the whole
+			# comment we will not see it. Such is life.
+			#
+			# Guestimate if this is a continuing comment.  If this
+			# is the start of a diff block and this line starts
+			# ' *' then it is very likely a comment.
+			if ($first_line and $line =~ m@^.\s*\*@) {
+				$in_comment = 1;
+			}

m@@ and and needed? Could be just

Quoted 2 lines
	if ($first_line == 0 && $line =~ /^.\s*\*/)

This one is being consistent with the ones around it.

Quoted 10 lines

scripts/patch-trivia-detector.pl line 163

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+			if ($line =~ m@/\*@) {
+				$in_comment = 1;
+			}
+			if ($line =~ m@\*/@) {
+				$in_comment = 0;
+			}

Although not necessary here, I'd like to point out that I've had more luck using paired characters as boundaries, i.e. m{} or s{}{}, which means less escaping for the inner regex in some cases.

Some consistancy would help for sure. Will try and move to some common form.

Quoted 5 lines

scripts/patch-trivia-detector.pl line 189

Quoted 2 lines
scripts/patch-trivia-detector.pl new file +2 -0
+# check we are in a valid source file *.[hc] if not then ignore this hunk
+		next if ($realfile !~ /\.[hc]$/);

Oh I think trailing whitespace and 80 column limit should also be applied to .S and .s files.

Seems sane.

Quoted 6 lines
Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+
+#trailing whitespace
+		if($line=~/\S\s+$/) {
+			my $herevet = "$here\n" . cat_vet($line) . "\n\n";
+			print ("trailing whitespace\n");
+			print "$herevet";
Quoted 8 lines

scripts/patch-trivia-detector.pl line 199

Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+			$clean = 0;
+		}
+#80 column limit
+		if(!($prevline=~/\/\*\*/) && length($lineforcounting) > 80){

Actually, I think this should be "> 79" (after stripping a .diff's control column), since the cursor may move to the 81th column when editing an 80-col line - which is what we want to avoid, no?

80 tends to work for me because of that "if on 80 then don't wrap until there is another character" behaviour of most terminals. Anyone else with a firm opinion.

Quoted 4 lines
Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+			print "line over 80 characters\n";
+			print "$herecurr";
+			$clean = 0;
+		}
Quoted 8 lines

scripts/patch-trivia-detector.pl line 207

Quoted 4 lines
scripts/patch-trivia-detector.pl new file +4 -0
+
+# at the beginning of a line any tabs must come first and anything
+# more than 8 must use tabs.
+		if($line=~/^\+\s* \t\s*\S/ or $line=~/^\+\s*        \s*/) {

|| will do instead of or.
Somehow this does not catch "Only spaces" lines (/^ +\S/), because there's no \t in them. (Talk about regex fun.)

Well this is designed to pick up spaces before tabs and any sets of spaces which are long enough to be a tab. The only thing it would not pick up as bad is things indented less than 8 which would be possibly frowned upon but would be correct with spaces.

Quoted 30 lines

I'd say split and simplify:

	if ($line =~ /^\s* {8,}/) {
		print "Make that big space a \\t\n";
	}
	if ($line =~ /^ +\t/) {
		print "Inadvertent space at line start, or make it \\t\n";
	}
Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+			my $herevet = "$here\n" . cat_vet($line) . "\n\n";
+			print ("use tabs not spaces\n");
+			print "$herevet";
+			$clean = 0;
+		}
+
+		#
+		# The rest of our checks refer specifically to C style

scripts/patch-trivia-detector.pl line 225

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+		# only apply those _outside_ comments.
+		#
+		next if ($in_comment);
+
+# no C99 // comments
+		if ($line =~ qr|//|) {
+			print "do not use C99 // comments\n";
+			print "$herecurr";
+			$clean = 0;
+		}

(Now we're using qr{}, what about being consistent with m{}?) It may break on rare occassions:

Quoted 1 line
	printk("Hello // World\n");

Yep, tighened this up to catch this one.

Quoted 29 lines
Quoted 29 lines
scripts/patch-trivia-detector.pl new file +29 -0
+		# Remove comments from the line before processing.
+		$line =~ s@/\*.*\*/@@g;
+		$line =~ s@/\*.*@@;
+		$line =~ s@.*\*/@@;
+		$line =~ s@//.*@@;
+
+#EXPORT_SYMBOL should immediately follow its function closing }.
+		if (($line =~ /EXPORT_SYMBOL.*\(.*\)/) ||
+		    ($line =~ /EXPORT_UNUSED_SYMBOL.*\(.*\)/)) {
+			if (($prevline !~ /^}/) &&
+			   ($prevline !~ /^\+}/) &&
+			   ($prevline !~ /^ }/)) {
+				print "EXPORT_SYMBOL(func); should immediately follow its function\n";
+				print "$herecurr";
+				$clean = 0;
+			}
+		}
+
+		# check for static initialisers.
+		if ($line=~/\s*static\s.*=\s+(0|NULL);/) {
+			print "do not initialise statics to 0 or NULL\n";
+			print "$herecurr";
+			$clean = 0;
+		}
+
+		# check for new typedefs.
+		if ($line=~/\s*typedef\s/) {
+			print "do not add new typedefs\n";
+			print "$herecurr";
Quoted 8 lines

scripts/patch-trivia-detector.pl line 261

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+			$clean = 0;
+		}
+
+# * goes on variable not on type
+		if($line=~/[A-Za-z\d_]+\* [A-Za-z\d_]+/) {
+			print ("\"foo* bar\" should be \"foo *bar\"\n"); 

print does not need parentheses in most cases.

Indeed, removed.

Quoted 12 lines
Quoted 12 lines
scripts/patch-trivia-detector.pl new file +12 -0
+			print "$herecurr";
+			$clean = 0;
+		}
+
+# no BUG() or BUG_ON()
+		if ($line =~ /\b(BUG|BUG_ON)\b/) {
+			print "Try to use WARN_ON & Recovery code rather than BUG() or BUG_ON()\n";
+			print "$herecurr";
+			$clean = 0;
+		}
+
+# printk should use KERN_* levels
Quoted 9 lines

scripts/patch-trivia-detector.pl line 279

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+		if (($line =~ /\bprintk\b/) &&
+		    ($line !~ /KERN_/)) {
+			print "printk() should include KERN_ facility level\n";
+			print "$herecurr";
+			$clean = 0;
+		}

if ($line =~ /\bprintk\((?!KERN_)/)

Yep, nicer, changed.

Quoted 1 line

Attention span ran out. :)

Thanks for bothering :) Any review is appreciated.

-apw

#11 lore raw
Quoted 8 lines
Quoted 4 lines

scripts/patch-trivia-detector.pl line 199

Quoted 1 line
scripts/patch-trivia-detector.pl new file +1 -0
+		if(!($prevline=~/\/\*\*/) && length($lineforcounting) > 80){

Actually, I think this should be "> 79" (after stripping a .diff's control column), since the cursor may move to the 81th column when editing an 80-col line - which is what we want to avoid, no?

80 tends to work for me because of that "if on 80 then don't wrap until there is another character" behaviour of most terminals. Anyone else with a firm opinion.

I think 80 is good. What the specific number is does not matter much, we all have screens wider than 80 characters. The point is just to have a number that prevents really long lines and prevents people from indenting too many levels past our minds ability to keep up. We've already all been coding to 80, and it happens to be a nice round number we can all remember and love. The only reason I see to select 79 is that prime numbers are generally cooler than other numbers.

-Joel

#12 lore raw
Quoted 9 lines · Joel Schopp
Quoted 7 lines
Quoted 4 lines

scripts/patch-trivia-detector.pl line 199

Quoted 1 line
scripts/patch-trivia-detector.pl new file +1 -0
+		if(!($prevline=~/\/\*\*/) && length($lineforcounting) > 80){

Actually, I think this should be "> 79" (after stripping a .diff's control column), since the cursor may move to the 81th column when editing an 80-col line - which is what we want to avoid, no?

80 tends to work for me because of that "if on 80 then don't wrap until there is another character" behaviour of most terminals. Anyone else with a firm opinion.

I think 80 is good. What the specific number is does not matter much, we all have screens wider than 80 characters. The point is just to have a number that prevents really long lines and prevents people from indenting too many levels past our minds ability to keep up. We've already all been coding to 80, and it happens to be a nice round number we can all remember and love. The only reason I see to select 79 is that prime numbers are generally cooler than other numbers.

We do have screens wider than 80 characters, but almost all the time I spend in terminal windows, they are set to 24x80 (the default size). It is a matter of habit, and I bet I'm not alone. Hence, 80 is "annoying" not only because patches will wrap, but also because in some editors the 80th character will also wrap.

Just my 2 cents,

Signature · 1 line
Julio M. Merino Vidal <[email protected]>
#13 lore raw
In reply to Andy Whitcroft
Quoted 5 lines · Andy Whitcroft
Quoted 3 lines

The consistent style quite looses after this point, esp. space around operators ;-)

Yep thats on my list for cleanup ...

Hey, you might even try running patch-trivia-detector.pl through itself. :)

Quoted 3 lines
Quoted 1 line

Attention span ran out. :)

Thanks for bothering. Any review is appreciated.

Followup comes..

Jan

#14 lore raw
In reply to Andy Whitcroft

scripts/patch-trivia-detector.pl line 268

Quoted 4 lines · Andy Whitcroft
scripts/patch-trivia-detector.pl new file +4 -0
+
+# no BUG() or BUG_ON()
+		if ($line =~ /\b(BUG|BUG_ON)\b/) {
+			print "Try to use WARN_ON & Recovery code rather than BUG() or BUG_ON()\n";

Just outlawing BUG_ON doesn't seem like a good idea to me. We'll just end up with lots of untested and likely buggy recovery code or no asserts. Both would be bad.

scripts/patch-trivia-detector.pl line 421

Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+#need space before brace following if, while, etc
+		if($line=~/\(.*\){/) {
+			print ("need a space before the brace\n");
+			print "$herecurr";
+			$clean = 0;		
+		}
+
+#gotos aren't indented

You mean goto labels? Surely goto statements are to be indented. Confusing message

scripts/patch-trivia-detector.pl line 426

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Quoted 1 line

+# don't include <linux/video_decoder.h>

It would be probably better to define some syntax that makes it possible to auto extract those from feature-removal-schedule.txt. Otherwise long term this will become messy.

Possible further checks that might make sense:
- panic() anywhere in drivers/*
- externs in .c files without asmlinkage
- general checking that everything in a fully visible {} block is the right indentation

-Andi

#15 lore raw
Quoted 9 lines · Andi Kleen

scripts/patch-trivia-detector.pl line 268

Quoted 4 lines · Andy Whitcroft
scripts/patch-trivia-detector.pl new file +4 -0
+
+# no BUG() or BUG_ON()
+		if ($line =~ /\b(BUG|BUG_ON)\b/) {
+			print "Try to use WARN_ON & Recovery code rather than BUG() or BUG_ON()\n";

Just outlawing BUG_ON doesn't seem like a good idea to me. We'll just end up with lots of untested and likely buggy recovery code or no asserts. Both would be bad.

Thats not an unreasonable position. And I tend to agree with it. Either we try and have two levels Warnings and Errors, or we just drop this one for now.

Anyone got an oppinion, so we can get a consensus.

Quoted 11 lines

scripts/patch-trivia-detector.pl line 421

Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+#need space before brace following if, while, etc
+		if($line=~/\(.*\){/) {
+			print ("need a space before the brace\n");
+			print "$herecurr";
+			$clean = 0;		
+		}
+
+#gotos aren't indented

You mean goto labels? Surely goto statements are to be indented. Confusing message

Yes, changed to labels

Quoted 8 lines

scripts/patch-trivia-detector.pl line 426

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Yep, we also now allow one space something to do with diff -p not getting confused...

Quoted 5 lines
Quoted 1 line

+# don't include <linux/video_decoder.h>

It would be probably better to define some syntax that makes it possible to auto extract those from feature-removal-schedule.txt. Otherwise long term this will become messy.

Yeah, that is a very sensible idea.

Quoted 7 lines

Possible further checks that might make sense:
- panic() anywhere in drivers/*
- externs in .c files without asmlinkage
- general checking that everything in a fully visible {} block is the right indentation

-Andi

Thanks.

-apw

#16 lore raw
Quoted 10 lines · Andy Whitcroft
Quoted 9 lines

scripts/patch-trivia-detector.pl line 426

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Quoted 2 lines

Yep, we also now allow one space something to do with diff -p not getting confused...

Not only diff and emacs. At least joe (joe.sf.net) is another editor whose %y tag or so also takes lines with

no space at the front and properly paired parens,braces,etc.

as key lines for %y. Since usually, functions are the only thing at the beginning of a line, this works well. That's why labels should be allowed to be indented by one.

Jan

#17 lore raw
Quoted 15 lines · On Tue, 29 May 2007 22:22:43 +0200 (MEST) Jan Engelhardt wrote:
Quoted 10 lines · Andy Whitcroft
Quoted 9 lines

scripts/patch-trivia-detector.pl line 426

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Quoted 2 lines

Yep, we also now allow one space something to do with diff -p not getting confused...

Quoted 8 lines

Not only diff and emacs. At least joe (joe.sf.net) is another editor whose %y tag or so also takes lines with

no space at the front and properly paired parens,braces,etc.

as key lines for %y. Since usually, functions are the only thing at the beginning of a line, this works well. That's why labels should be allowed to be indented by one.

That's just a joe bug then.

diff no longer requires a beginning space, at least in the testing that I did it worked with no leading space on a label:.

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

#18 lore raw
Quoted 13 lines · Randy Dunlap
Quoted 13 lines
Quoted 11 lines
Quoted 8 lines

scripts/patch-trivia-detector.pl line 426

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Yep, we also now allow one space something to do with diff -p not getting confused...

Not only diff and emacs. At least joe (joe.sf.net) is another editor

[...]

Quoted 5 lines

That's just a joe bug then.

diff no longer requires a beginning space, at least in the testing that I did it worked with no leading space on a label:.

But not everyone runs the latest and greatest. diffutils 2.8.7 takes /^label:/ as a marker for the @@ line.

Jan

#19 lore raw
Quoted 11 lines · Jan Engelhardt
Quoted 10 lines · Randy Dunlap
Quoted 10 lines
Quoted 9 lines
Quoted 7 lines

scripts/patch-trivia-detector.pl line 426

Quoted 5 lines
scripts/patch-trivia-detector.pl new file +5 -0
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

emacs generates one space label in front of a goto label. I wouldn't outlaw this.

Yep, we also now allow one space something to do with diff -p not getting confused...

Not only diff and emacs. At least joe (joe.sf.net) is another editor

Quoted 8 lines

[...]

Quoted 4 lines

That's just a joe bug then.

diff no longer requires a beginning space, at least in the testing that I did it worked with no leading space on a label:.

But not everyone runs the latest and greatest. diffutils 2.8.7 takes /^label:/ as a marker for the @@ line.

OK. What does "marker for the @@ line" mean?

Signature · 2 lines
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
#20 lore raw
Quoted 9 lines · Randy Dunlap
Quoted 7 lines
Quoted 4 lines

That's just a joe bug then.

diff no longer requires a beginning space, at least in the testing that I did it worked with no leading space on a label:.

But not everyone runs the latest and greatest. diffutils 2.8.7 takes /^label:/ as a marker for the @@ line.

OK. What does "marker for the @@ line" mean?

if ($_ !~ /^[\(\[\{]/ && $_ =~ /^(\S+)/) {
    print "@@ -a,b +c,d @@ $1\n"
}

So
label:
will be the 'mark[er]' for the @@ line

Just try yourself:
<<< foo.c <<<
int main(void)
{
x
y
z
label:
a
b
c
d
e
f
}

Quoted 1 line
Quoted 1 line
Quoted 1 line

<<< foo2.c <<<
int main(void)
{
x
y
z
label:

	a
	b
	c
	d
	d+
	e
	f
}
Quoted 1 line
Quoted 1 line
Quoted 1 line

<<< diff -dpru foo.c foo2.c <<<

foo2.c 2007-05-30 18:03:43.061471495 +0200 renamed from foo.c 2007-05-30 10:33:45.714222656 +0200 +1 -0
@@ -8,6 +8,7 @@ label:
 	b
 	c
 	d
+	d+
 	e
 	f
 }
Quoted 1 line
Quoted 1 line
Quoted 1 line

Jan

#21 lore raw

Re: [PATCH] add a trivial patch style checker II

In reply to Andy Whitcroft
Quoted 8 lines

Yeah, that is a very sensible idea.

Quoted 6 lines

Possible further checks that might make sense:
- panic() anywhere in drivers/*
- externs in .c files without asmlinkage
- general checking that everything in a fully visible {} block is the right indentation

Here are some more warnings I would like to see:

- Warning for any spinlock/mutex definition that doesn't have a comment nearby (all locks ought to be documented)
- Keep an ifdef count and give minus points for too many
- Warn about any architecture ifdefs (__i386__ etc.)
While not 100% illegal this is definitely something that needs to be justified

-Andi

#22 lore raw

Re: [PATCH] add a trivial patch style checker II

Quoted 9 lines · Andi Kleen
Quoted 8 lines

Yeah, that is a very sensible idea.

Quoted 6 lines

Possible further checks that might make sense:
- panic() anywhere in drivers/*
- externs in .c files without asmlinkage
- general checking that everything in a fully visible {} block is the right indentation

Quoted 4 lines

Here are some more warnings I would like to see:

- Warning for any spinlock/mutex definition that doesn't have a comment nearby (all locks ought to be documented)

Also barriers. (Probably even moreso).

Dave

Signature · 1 line
http://www.codemonkey.org.uk
#23 lore raw

Re: [PATCH] add a trivial patch style checker II

Quoted 10 lines · Dave Jones
Quoted 9 lines · Andi Kleen
Quoted 8 lines

Yeah, that is a very sensible idea.

Quoted 6 lines

Possible further checks that might make sense:
- panic() anywhere in drivers/*
- externs in .c files without asmlinkage
- general checking that everything in a fully visible {} block is the right indentation

Quoted 6 lines
Quoted 4 lines

Here are some more warnings I would like to see:

- Warning for any spinlock/mutex definition that doesn't have a comment nearby (all locks ought to be documented)

Also barriers. (Probably even moreso).

Both of these seem a pretty good idea. Should be in version 0.03 which I'll try and get to Andrew over the weekend. Example reports from files in 2.6.22-rc2-mm1 below.

-apw

spinlock_t definition without comment
FILE: lib/statistic.c:243:
+ spinlock_t lock;

struct mutex definition without comment
FILE: include/linux/kernelcapi.h:67:
+ struct mutex recv_mtx;

memory barrier without comment
FILE: fs/ext2/balloc.c:1250:
+ smp_rmb();

#24 lore raw
In reply to Andi Kleen
Quoted 2 lines · Andi Kleen

Possible further checks that might make sense:
- panic() anywhere in drivers/*

A driver should be allowed to panic. E.g. if it detects that due to a firmware or driver bug memory corruption happened. IMHO the best thing to do then is panic.

#25 lore raw
Quoted 7 lines · Heiko Carstens
Quoted 2 lines · Andi Kleen

Possible further checks that might make sense:
- panic() anywhere in drivers/*

A driver should be allowed to panic. E.g. if it detects that due to a firmware or driver bug memory corruption happened. IMHO the best thing to do then is panic.

That is not how Linux normally operates. A BUG() doesn't panic() by default either.

And on systems with IOMMU that is exactly the wrong thing to do.

Besides the problem is that bad drivers tend to badly abuse it (e.g. see some particular BSD derviced SCSI drivers). We definitely don't want any more of such code.

-Andi

#26 lore raw
Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

Possible further checks that might make sense:
- panic() anywhere in drivers/*

A driver should be allowed to panic. E.g. if it detects that due to a firmware or driver bug memory corruption happened. IMHO the best thing to do then is panic.

Quoted 8 lines

That is not how Linux normally operates. A BUG() doesn't panic() by default either.

And on systems with IOMMU that is exactly the wrong thing to do.

Besides the problem is that bad drivers tend to badly abuse it (e.g. see some particular BSD derviced SCSI drivers). We definitely don't want any more of such code.

So you prefer random data corruption over an emergency stop? That doesn't make much sense to me...

#27 lore raw
Quoted 1 line

So you prefer random data corruption over an emergency stop?

With an oops you can at least recover the system and actually look at the problem. On a machine with a panic you're just dead and the probability of actually being able to do something about the problem is much lower. On x86 systems you typically don't even get
any message out.

And I'm not convinced drivers are in a good position to decide if memory was likely corrupted or not anyways. At least the panics I see in driver sources seem to be just random logic bugs from someone not familiar with BUG().

Also they typically don't make much attempt to figure out if there might have been data corruption.

If you're really worried about memory corruption in drivers you should just use an IOMMU.

Quoted 1 line

That doesn't make much sense to me...

So you're always setting panic_on_oops?

-Andi

#28 lore raw
Quoted 7 lines · Andi Kleen
Quoted 1 line

So you prefer random data corruption over an emergency stop?

With an oops you can at least recover the system and actually look at the problem. On a machine with a panic you're just dead and the probability of actually being able to do something about the problem is much lower. On x86 systems you typically don't even get
any message out.

Ok, that's the different approach of analyzing problems. On s390 we prefer dumps of a crashed system, since that is much easier to debug than a kernel that just printed some lines and then went on as if nothing happened. Besides the nice side effects that a BUG() statement has of course.

Quoted 7 lines

And I'm not convinced drivers are in a good position to decide if memory was likely corrupted or not anyways. At least the panics I see in driver sources seem to be just random logic bugs from someone not familiar with BUG().

Also they typically don't make much attempt to figure out if there might have been data corruption.

I'm talking of a specific problem where we just added a panic to the zfcp device driver. If that panic ever triggers we know for sure that memory corruption happened.
So I'm just asking to not say in general that panic() in a device driver is a bad thing.

Quoted 2 lines

If you're really worried about memory corruption in drivers you should just use an IOMMU.

IOMMU and s390 don't fit together.

Quoted 2 lines
Quoted 1 line

That doesn't make much sense to me...

So you're always setting panic_on_oops?

On our internal tests, yes. Otherwise it's of course up to the distros. Btw. the default implementation for BUG() is panic(). See asm-generic/bug.h.

#29 lore raw
Quoted 6 lines · Heiko Carstens

I'm talking of a specific problem where we just added a panic to the zfcp device driver. If that panic ever triggers we know for sure that memory corruption happened.
So I'm just asking to not say in general that panic() in a device driver is a bad thing.

Ok there might be exceptions, but in general I still think it's true. It probably fits the "you must be able to justify any warnings" rule. I would still like this check to be added.

-Andi

#30 lore raw
In reply to Andi Kleen
Quoted 5 lines · Andi Kleen
Quoted 1 line

+# don't include <linux/video_decoder.h>

It would be probably better to define some syntax that makes it possible to auto extract those from feature-removal-schedule.txt. Otherwise long term this will become messy.

Turns out that we already have Files: entries in f-r-s.txt and so I've enhanced things to pick them out if they start include/ and check them.
Works well for the two that exists as of today.

-apw

#31 lore raw
In reply to Andy Whitcroft

On 28/05/07, Andy Whitcroft <[email protected]> wrote:

...

Quoted 6 lines
Documentation/SubmittingPatches +0 -0
index a417b25..23637e8 100644
@@ -118,7 +118,21 @@ then only post say 15 or so at a time and wait for review and integration.
 

...

Quoted 5 lines

-8) Name your kernel version.
+9) Name your kernel version.

It is important to note, either in the subject line or in the patch description, the kernel version to which this patch applies.

If omitted, we could assume it default to the latest linus git tree.

Quoted 5 lines

-10) Include PATCH in the subject
+11) Include PATCH in the subject

Due to high e-mail traffic to Linus, and to linux-kernel, it is common convention to prefix your subject line with [PATCH]. This lets Linus

or use "patch", to be easier to read.

Signature · 1 line
Qi Yong
#32 lore raw

PATCH 1/1 add a trivial patch style checker v2

In reply to Andy Whitcroft

We are seeing increasing levels of minor patch style violations in submissions to the mailing lists as well as making it into the tree. These detract from the quality of the submission and cause unnessary work for reviewers.

As a first step package up the current state of the patch style checker and include it in the kernel tree. Add instructions suggesting running it on submissions. This adds version v0.01 of the checkpatch.pl script.

---
Ok, here is V2 of the checker. A lot of feedback has been
incorporated including changing its name to better fit with the current naming of other things in the scripts directory.

Changelog:
o new name checkpatch.pl
o also mention in Documentation/SubmitChecklist
o whitespace cleanups
o if () { spacing cleanups
o improves signoff handling
o goto label handling improved and message clarified
o handle deprecated headers using feature-removal-schedule.txt
o fixed reference to section in Documentation/SubmittingPatches
---

Documentation/SubmitChecklist +6 -0
index 3af3e65..6ebffb5 100644
@@ -84,3 +84,9 @@ kernel patches.
 24: Avoid whitespace damage such as indenting with spaces or whitespace
     at the end of lines.  You can test this by feeding the patch to
     "git apply --check --whitespace=error-all"
+
+25: Check your patch for general style as detailed in
+    Documentation/CodingStyle.  Check for trivial violations with the
+    patch style checker prior to submission (scripts/checkpatch.pl).
+    You should be able to justify all violations that remain in
+    your patch.
Documentation/SubmittingPatches +28 -11
index a417b25..d91125a 100644
@@ -118,7 +118,20 @@ then only post say 15 or so at a time and wait for review and integration.
 
 
 
-4) Select e-mail destination.
+4) Style check your changes.
+
+Check your patch for basic style violations, details of which can be
+found in Documentation/CodingStyle.  Failure to do so simply wastes
+the reviewers time and will get your patch rejected, probabally
+without even being read.
+
+At a minimum you should check your patches with the patch style
+checker prior to submission (scripts/patchcheck.pl).  You should
+be able to justify all violations that remain in your patch.
+
+
+
+5) Select e-mail destination.
 
 Look through the MAINTAINERS file and the source code, and determine
 if your change applies to a specific subsystem of the kernel, with
@@ -146,7 +159,7 @@ discussed should the patch then be submitted to Linus.
 
 
 
-5) Select your CC (e-mail carbon copy) list.
+6) Select your CC (e-mail carbon copy) list.
 
 Unless you have a reason NOT to do so, CC [email protected].
 
@@ -187,8 +200,7 @@ URL: <http://www.kernel.org/pub/linux/kernel/people/bunk/trivial/>
 
 
 
-
-6) No MIME, no links, no compression, no attachments.  Just plain text.
+7) No MIME, no links, no compression, no attachments.  Just plain text.
 
 Linus and other kernel developers need to be able to read and comment
 on the changes you are submitting.  It is important for a kernel
@@ -223,9 +235,9 @@ pref("mailnews.display.disable_format_flowed_support", true);
 
 
 
-7) E-mail size.
+8) E-mail size.
 
-When sending patches to Linus, always follow step #6.
+When sending patches to Linus, always follow step #7.
 
 Large changes are not appropriate for mailing lists, and some
 maintainers.  If your patch, uncompressed, exceeds 40 kB in size,
@@ -234,7 +246,7 @@ server, and provide instead a URL (link) pointing to your patch.
 
 
 
-8) Name your kernel version.
+9) Name your kernel version.
 
 It is important to note, either in the subject line or in the patch
 description, the kernel version to which this patch applies.
@@ -244,7 +256,7 @@ Linus will not apply it.
 
 
 
-9) Don't get discouraged.  Re-submit.
+10) Don't get discouraged.  Re-submit.
 
 After you have submitted your change, be patient and wait.  If Linus
 likes your change and applies it, it will appear in the next version
@@ -270,7 +282,7 @@ When in doubt, solicit comments on linux-kernel mailing list.
 
 
 
-10) Include PATCH in the subject
+11) Include PATCH in the subject
 
 Due to high e-mail traffic to Linus, and to linux-kernel, it is common
 convention to prefix your subject line with [PATCH].  This lets Linus
@@ -279,7 +291,7 @@ e-mail discussions.
 
 
 
-11) Sign your work
+12) Sign your work
 
 To improve tracking of who did what, especially with patches that can
 percolate to their final resting place in the kernel through several
@@ -328,7 +340,8 @@ now, but you can do this to mark internal company procedures or just
 point out some special detail about the sign-off. 
 
 
-12) The canonical patch format
+
+13) The canonical patch format
 
 The canonical patch subject line is:
 
@@ -427,6 +440,10 @@ section Linus Computer Science 101.
 Nuff said.  If your code deviates too much from this, it is likely
 to be rejected without further review, and without comment.
 
+Check your patches with the patch style checker prior to submission
+(scripts/checkpatch.pl).  You should be able to justify all
+violations that remain in your patch.
+
 
 
 2) #ifdefs are ugly
Documentation/feature-removal-schedule.txt +1 -0
index 368063e..6b6e778 100644
@@ -70,6 +70,7 @@ Who:	David Miller <[email protected]>
 
 What:	Video4Linux API 1 ioctls and video_decoder.h from Video devices.
 When:	December 2006
+Files:	include/linux/video_decoder.h
 Why:	V4L1 AP1 was replaced by V4L2 API. during migration from 2.4 to 2.6
 	series. The old API have lots of drawbacks and don't provide enough
 	means to work with all video and audio standards. The newer API is
MAINTAINERS +14 -2
index d7533dd..c53f01f 100644
@@ -30,8 +30,11 @@ trivial patch so apply some common sense.
 	job the maintainers (and especially Linus) do is to keep things
 	looking the same. Sometimes this means that the clever hack in
 	your driver to get around a problem actually needs to become a
-	generalized kernel feature ready for next time. See
-	Documentation/CodingStyle for guidance here.
+	generalized kernel feature ready for next time.
+
+	PLEASE check your patch with the automated style checker
+	(scripts/checkpatch.pl) to catch trival style violations.
+	See Documentation/CodingStyle for guidance here.
 
 	PLEASE try to include any credit lines you want added with the
 	patch. It avoids people being missed off by mistake and makes
@@ -955,6 +958,15 @@ M:	[email protected]
 L:	[email protected]
 S:	Maintained
 
+CHECKPATCH
+P:	Andy Whitcroft
+M:	[email protected]
+P:	Randy Dunlap
+M:	[email protected]
+P:	Joel Schopp
+M:	[email protected]
+S:	Supported
+
 COMMON INTERNET FILE SYSTEM (CIFS)
 P:	Steve French
 M:	[email protected]
scripts/checkpatch.pl new file +595 -0

595 lines not on this page · raw message

#34 lore raw

Re: [PATCH] add a trivial patch style checker v2

In reply to Andy Whitcroft
Quoted 3 lines · Andy Whitcroft

As a first step package up the current state of the patch style checker and include it in the kernel tree. Add instructions suggesting running it on submissions. This adds version v0.01 of

                                                           ^^^^^
Quoted 5 lines

the checkpatch.pl script.

---
Ok, here is V2 of the checker. A lot of feedback has been

                    ^^
Quoted 4 lines
scripts/checkpatch.pl new file +1 -0
@@ -0,0 +1,595 @@
+#!/usr/bin/perl -w

scripts/checkpatch.pl line 11

Quoted 10 lines
scripts/checkpatch.pl new file +10 -0
+# (c) 2001, Dave Jones. <[email protected]> (the file handling bit)
+# (c) 2005, Joel Scohpp <[email protected]> (the ugly bit)
+# (c) 2007, Andy Whitcroft <[email protected]> (new conditions, test suite, etc)
+# Licensed under the terms of the GNU GPL License version 2
+
+use strict;
+
+my $P = $0;
+
+my $V = '0.01';
            ^^^^

So, what's the correct version number? ;-)

Will http://www.codemonkey.org.uk/projects/checkpatch/checkpatch.git/ (hey, a project with the same name started by the same person???) be merged in?

Below is a small fix for the wrong hardcoded script name

---
Subject: [PATCH] Fix checkpatch.pl name in usage template

Fix checkpatch.pl name in usage template:
- Don't use wrong hardcoded script name
- Strip dirname from command name

---
 scripts/checkpatch.pl |    3 ++-
 1 files changed, 2 insertions(+), 1 deletion(-)
scripts/checkpatch.pl +2 -1
@@ -7,6 +7,7 @@
 use strict;
 
 my $P = $0;
+$P =~ s@.*/@@g;
 
 my $V = '0.01';
 
@@ -26,7 +27,7 @@ GetOptions(
 my $exit = 0;
 
 if ($#ARGV < 0) {
-	print "usage: patchstylecheckemail.pl [options] patchfile\n";
+	print "usage: $P [options] patchfile\n";
 	print "version: $V\n";
 	print "options: -q           => quiet\n";
 	print "         --no-tree    => run without a kernel tree\n";

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- Sony Network and Software Technology Center Europe (NSCE) [email protected] ------- The Corporate Village, Da Vincilaan 7-D1 Voice +32-2-7008453 Fax +32-2-7008622 ---------------- B-1935 Zaventem, Belgium

#35 lore raw

Re: [PATCH] add a trivial patch style checker v2

Quoted 2 lines · Geert Uytterhoeven

Will http://www.codemonkey.org.uk/projects/checkpatch/checkpatch.git/ (hey, a project with the same name started by the same person???) be merged in?

Andy and I were working on this in parallel with the same goal. When his started to get traction, I stopped working on my variant. Any regexps in my version that aren't in Andy's should of course be merged.

Dave

Signature · 1 line
http://www.codemonkey.org.uk
#36 lore raw

Re: [PATCH] add a trivial patch style checker v2

In reply to Geert Uytterhoeven
Quoted 26 lines · Geert Uytterhoeven
Quoted 3 lines · Andy Whitcroft

As a first step package up the current state of the patch style checker and include it in the kernel tree. Add instructions suggesting running it on submissions. This adds version v0.01 of

                                                           ^^^^^
Quoted 5 lines

the checkpatch.pl script.

---
Ok, here is V2 of the checker. A lot of feedback has been

                    ^^
Quoted 4 lines
scripts/checkpatch.pl new file +1 -0
@@ -0,0 +1,595 @@
+#!/usr/bin/perl -w

scripts/checkpatch.pl line 11

Quoted 10 lines
scripts/checkpatch.pl new file +10 -0
+# (c) 2001, Dave Jones. <[email protected]> (the file handling bit)
+# (c) 2005, Joel Scohpp <[email protected]> (the ugly bit)
+# (c) 2007, Andy Whitcroft <[email protected]> (new conditions, test suite, etc)
+# Licensed under the terms of the GNU GPL License version 2
+
+use strict;
+
+my $P = $0;
+
+my $V = '0.01';
            ^^^^
Quoted 2 lines

So, what's the correct version number? ;-)

Yep that is terribly confusing. Internally I have actually tagged the V2 version 0.02, and will be using those numbers from now on.

Quoted 4 lines

Will http://www.codemonkey.org.uk/projects/checkpatch/checkpatch.git/ (hey, a project with the same name started by the same person???) be merged in?

Below is a small fix for the wrong hardcoded script name

Yep this looks good. I've squished this into my tree.

Quoted 24 lines

---
Subject: [PATCH] Fix checkpatch.pl name in usage template

Fix checkpatch.pl name in usage template:
- Don't use wrong hardcoded script name
- Strip dirname from command name

---
 scripts/checkpatch.pl |    3 ++-
 1 files changed, 2 insertions(+), 1 deletion(-)
scripts/checkpatch.pl +1 -0
@@ -7,6 +7,7 @@
 use strict;
 
 my $P = $0;
+$P =~ s@.*/@@g;
 
 my $V = '0.01';
 

scripts/checkpatch.pl GetOptions( · line 33

Quoted 8 lines
scripts/checkpatch.pl +1 -1
@@ -26,7 +27,7 @@ GetOptions(
 my $exit = 0;
 
 if ($#ARGV < 0) {
-	print "usage: patchstylecheckemail.pl [options] patchfile\n";
+	print "usage: $P [options] patchfile\n";
 	print "version: $V\n";
 	print "options: -q           => quiet\n";
 	print "         --no-tree    => run without a kernel tree\n";

-apw

#37 lore raw
In reply to Andy Whitcroft

scripts/patch-trivia-detector.pl line 62

Quoted 11 lines
scripts/patch-trivia-detector.pl new file +10 -0
+sub top_of_kernel_tree {
+	if ((-f "COPYING") && (-f "CREDITS") && (-f "Kbuild") &&
+	    (-f "MAINTAINERS") && (-f "Makefile") && (-f "README") &&
+	    (-d "Documentation") && (-d "arch") && (-d "include") &&
+	    (-d "drivers") && (-d "fs") && (-d "init") && (-d "ipc") &&
+	    (-d "kernel") && (-d "lib") && (-d "scripts")) {
+		return 1;
+	}
+	return 0;
+}

Looks like some redundant parentheses.

scripts/patch-trivia-detector.pl line 426

Quoted 6 lines
scripts/patch-trivia-detector.pl new file +6 -0
+#gotos aren't indented
+		if($line=~/^\s*[A-Za-z\d_]+:/ and !($line=~/^\s*default:/)){
+			print "Gotos should not be indented\n";
+			print "$herecurr";
+			$clean = 0;
+		}

I think this was discussed already - some indent (1 space?) should be allowed.

Review

Quoted 2 lines
(quoted diff) +2 -0
+#studly caps, commented out until figure out how to distinguish between use
+#of existing and adding new

Yeah that's a big problem, given that linux's mm/ directory has quite a lot of camel case function names.

scripts/patch-trivia-detector.pl line 470

Quoted 8 lines
scripts/patch-trivia-detector.pl new file +8 -0
+#		if(($line=~/[\w_][a-z\d]+[A-Z]/) and !($line=~/print/)) {
+#		    print ("No studly caps, use _\n");
+#		    print "$herecurr";
+#		    $clean = 0;
+#		}
+
+#no spaces allowed after \ in define
+		if($line=~/\#define.*\\\s$/){

Usually, #s do _not_ need to be quoted (in contrast to @). I am at stake to be wrong, anyone know more? :)

scripts/patch-trivia-detector.pl line 487

Quoted 2 lines
scripts/patch-trivia-detector.pl new file +2 -0
+#if/while/etc brace do not go on next line, unless #defining a do while loop, or if that brace on the next line is for something else
+		if ($prevline=~/(if|while|for|switch)\s*\(/) {

/(?:if|while|.../

I don't see $1 being captured/used, so..

Quoted 11 lines
scripts/patch-trivia-detector.pl new file +11 -0
+			my @opened = $prevline=~/\(/g;
+			my @closed = $prevline=~/\)/g;
+			my $nr_line = $linenr;
+			my $remaining = $realcnt;
+			my $next_line = $line;
+			my $extra_lines = 0;
+			my $display_segment = $prevline;
+
+			while ($remaining > 0 && scalar @opened > scalar @closed) {
+				$prevline .= $next_line;
+				$display_segment .= "\n" . $next_line;

scripts/patch-trivia-detector.pl line 508

Quoted 10 lines
scripts/patch-trivia-detector.pl new file +10 -0
+				$next_line = $lines[$nr_line];
+				$nr_line++;
+				$remaining--;
+
+				@opened = $prevline=~/\(/g;
+				@closed = $prevline=~/\)/g;
+			}
+
+			if(($prevline=~/(if|while|for|switch)\s*\(.*\)\s*$/) and ($next_line=~/{/) and
+			   !($next_line=~/(if|while|for)/) and !($next_line=~/\#define.*do.*while/)) {

Same.

Quoted 1 line

+# don't include <linux/video_decoder.h>

Who does that?

Jan