July 2014 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

Re: [PATCH] checkpatch: Remove unnecessary + after {8,8}
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 07:09:30PM -0700, Joe Perches wrote: > There's a useless "+" use that needs to be removed as perl 5.20 > emits a "Useless use of greediness modifier '+'" message each > time it's hit. > > Reported-by: Greg KH <gregkh(a)linuxfoundation.org> > Signed-off-by: Joe Perches <joe(a)perches.com> > --- > On Fri, 2014-07-11 at 19:05 -0700, Greg KH wrote: > > Ok, with linux-next I get the same thing: > > Thanks Greg. Very nice, thanks for this: Tested-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>

1 0

Re: [PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 06:57:24PM -0700, Joe Perches wrote: > On Fri, 2014-07-11 at 18:53 -0700, Greg KH wrote: > > On Fri, Jul 11, 2014 at 06:21:27PM -0700, Joe Perches wrote: > > > A simple script to run checkpatch --fix for various types of > > > of cleanups. > [] > > Anyway, try running this script on > > drivers/staging/lustre/lnet/lnet/acceptor.c to see how this build fails. > > lustre doesn't use "normal" kernel makefiles. Well, it does, it just has a: subdir-ccflags-y := -I$(src)/include/ line in drivers/staging/lustre/Makefile that messes up making .o files in subdirectories. I'll go fix that include file mess up now to be able to remove that line so the the .o file building will work. thanks, greg k-h

1 0

Re: [PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 06:21:27PM -0700, Joe Perches wrote: > A simple script to run checkpatch --fix for various types of > of cleanups. > > This script is useful primarily for staging. > > This reformats code to a more CodingStyle conforming style, > compiles it, verifies that the object code hasn't changed, > and git commits it too. > > You must have the necessary development tools, git, and a > recent git tree. Ideally use Greg KH's staging-next, which > can be retrieved via these commands: > > git clone git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git > git checkout staging-next > > To use this script try a sequence of commands like: > > cd <linux_repository> > git checkout -b <your_branch> > make allyesconfig > mkdir patches > ./scripts/reformat_with_checkpatch.sh drivers/staging/<dir>/*.[ch] > git format-patch --cover-letter -o patches/<your_branch> staging-next > git send-email patches/<your_branch> > > Signed-off-by: Joe Perches <joe(a)perches.com> > --- > scripts/reformat_with_checkpatch.sh | 141 ++++++++++++++++++++++++++++++++++++ > 1 file changed, 141 insertions(+) > create mode 100755 scripts/reformat_with_checkpatch.sh > > diff --git a/scripts/reformat_with_checkpatch.sh b/scripts/reformat_with_checkpatch.sh > new file mode 100755 > index 0000000..5415a8e > --- /dev/null > +++ b/scripts/reformat_with_checkpatch.sh > @@ -0,0 +1,141 @@ > +#!/bin/bash > +# (c) 2014, Joe Perches <joe(a)perches.com> > +# > +# Automate checkpatch modifications and git commits to > +# neaten code that doesn't conform to CodingStyle. > +# Useful primarily for files in drivers/staging > +# > +# Licensed under the terms of the GNU GPL License version 2 > + > +declare -ar whitespace_types=( \ > + 'whitespace_neatening:spacing,space_before_tab,pointer_location,trailing_whitespace,bracket_space' \ > + 'remove_spaces_before_tabs:space_before_tab' \ > + 'fix_label_positions:indented_label' \ > + 'align_arguments_to_parenthesis:parenthesis_alignment' \ > +) > + > +declare -ar changecode_types=( \ > + 'fix_brace_positions:open_brace,braces,else_after_brace,while_after_brace' \ > + 'fix_blank_lines:line_spacing' \ > + 'use_standard_attributes:prefer_packed,prefer_aligned' \ > + 'remove_unnecessary_externs:avoid_externs' \ > + 'update_c90_comment_style:c99_comments' \ > +) > + > +checkpatch_update () > +{ > + file=$1 > + > + desc=$(echo $2 | cut -f1 -d: | sed 's/_/ /g') > + types=$(echo $2 | cut -f2- -d:) > + > + echo "file: <$file> description: <$desc> types:<$types>" > + > + ./scripts/checkpatch.pl --file --fix-inplace --strict --types="$types" $file > + > + checkpatch_fixes=$file.diff > + git diff --stat -p --exit-code $file > $checkpatch_fixes > + if [ $? == 0 ] ; then > + rm -f $checkpatch_fixes > + return > + fi > + > + basename=$(basename $file) > + if [ "${basename##*.}" == "c" ] ; then > + > + git checkout $file > + obj="$(echo $file | sed 's/\.c$/\.o/')" > + if [ -e $obj ] ; then > + rm -f $obj > + fi > + > + echo "Compiling original version..." > + > + ${CROSS_COMPILE}make $obj This fails for when a cflags option is set (like an include path). Now, I could argue that having an include path override in a kernel Makefile is a bug, but well, it's staging... Anyway, try running this script on drivers/staging/lustre/lnet/lnet/acceptor.c to see how this build fails. thanks, greg k-h

2 1

Re: [PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 06:21:27PM -0700, Joe Perches wrote: > A simple script to run checkpatch --fix for various types of > of cleanups. > > This script is useful primarily for staging. > > This reformats code to a more CodingStyle conforming style, > compiles it, verifies that the object code hasn't changed, > and git commits it too. > > You must have the necessary development tools, git, and a > recent git tree. Ideally use Greg KH's staging-next, which > can be retrieved via these commands: > > git clone git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git > git checkout staging-next > > To use this script try a sequence of commands like: > > cd <linux_repository> > git checkout -b <your_branch> > make allyesconfig > mkdir patches > ./scripts/reformat_with_checkpatch.sh drivers/staging/<dir>/*.[ch] > git format-patch --cover-letter -o patches/<your_branch> staging-next > git send-email patches/<your_branch> > > Signed-off-by: Joe Perches <joe(a)perches.com> > --- > scripts/reformat_with_checkpatch.sh | 141 ++++++++++++++++++++++++++++++++++++ > 1 file changed, 141 insertions(+) > create mode 100755 scripts/reformat_with_checkpatch.sh > > diff --git a/scripts/reformat_with_checkpatch.sh b/scripts/reformat_with_checkpatch.sh > new file mode 100755 > index 0000000..5415a8e > --- /dev/null > +++ b/scripts/reformat_with_checkpatch.sh > @@ -0,0 +1,141 @@ > +#!/bin/bash > +# (c) 2014, Joe Perches <joe(a)perches.com> > +# > +# Automate checkpatch modifications and git commits to > +# neaten code that doesn't conform to CodingStyle. > +# Useful primarily for files in drivers/staging > +# > +# Licensed under the terms of the GNU GPL License version 2 > + > +declare -ar whitespace_types=( \ > + 'whitespace_neatening:spacing,space_before_tab,pointer_location,trailing_whitespace,bracket_space' \ > + 'remove_spaces_before_tabs:space_before_tab' \ > + 'fix_label_positions:indented_label' \ > + 'align_arguments_to_parenthesis:parenthesis_alignment' \ > +) > + > +declare -ar changecode_types=( \ > + 'fix_brace_positions:open_brace,braces,else_after_brace,while_after_brace' \ > + 'fix_blank_lines:line_spacing' \ > + 'use_standard_attributes:prefer_packed,prefer_aligned' \ > + 'remove_unnecessary_externs:avoid_externs' \ > + 'update_c90_comment_style:c99_comments' \ > +) > + > +checkpatch_update () > +{ > + file=$1 > + > + desc=$(echo $2 | cut -f1 -d: | sed 's/_/ /g') > + types=$(echo $2 | cut -f2- -d:) > + > + echo "file: <$file> description: <$desc> types:<$types>" > + > + ./scripts/checkpatch.pl --file --fix-inplace --strict --types="$types" $file > + > + checkpatch_fixes=$file.diff > + git diff --stat -p --exit-code $file > $checkpatch_fixes > + if [ $? == 0 ] ; then > + rm -f $checkpatch_fixes > + return > + fi > + > + basename=$(basename $file) > + if [ "${basename##*.}" == "c" ] ; then > + > + git checkout $file > + obj="$(echo $file | sed 's/\.c$/\.o/')" > + if [ -e $obj ] ; then > + rm -f $obj > + fi > + > + echo "Compiling original version..." > + > + ${CROSS_COMPILE}make $obj > + > + ${CROSS_COMPILE}objdump -D $obj | \ > + sed "s/^[[:space:]]\+[0-9a-f]\+//" > $obj.old > + > + patch -p1 < $checkpatch_fixes > + > + echo "Compiling modified version..." > + > + ${CROSS_COMPILE}make $obj > + > + ${CROSS_COMPILE}objdump -D $obj | \ > + sed "s/^[[:space:]]\+[0-9a-f]\+//" > $obj.new > + > + echo "Comparing objects..." > + diff -Nurd $obj.new $obj.old > + if [ $? -ne 0 ] ; then > + echo "Object differences exist! - Verify changes before commit!" > + read -s -p "Press the 'enter' key to continue: " > + else > + echo "No object differences found" > + fi > + rm -f $obj.old > + rm -f $obj.new > + fi > + > + echo "running checkpatch on possible checkpatch fixes..." > + > + ./scripts/checkpatch.pl --no-summary --no-signoff $checkpatch_fixes > + rm -f $checkpatch_fixes > + > + echo "Verify checkpatch output and make any necessary changes" > + echo "Edit '$file' if appropriate" > + read -s -p "Press the 'enter' key to continue: " > + > + commit_log_file=$(mktemp git_commit.XXXXXX) > + > + if [ -e $commit_log_file ] ; then > + rm -f $commit_log_file > + fi > + > + if [[ $file =~ ^drivers/staging/ ]] ; then > + echo -n "staging: " >> $commit_log_file > + fi > + echo "$(basename $(dirname $file)): checkpatch cleanup: $desc" >> $commit_log_file If I pick drivers/staging/lustre/include/linux/lnet/types.h, then I get: staging: lnet: checkpatch cleanup: whitespace neatening and no 'types.h' here, is that intentional? If so, why? And this is fun, I'm going to let this rip on the lustre code... thanks, greg k-h

2 1

Re: [PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 06:21:27PM -0700, Joe Perches wrote: > A simple script to run checkpatch --fix for various types of > of cleanups. > > This script is useful primarily for staging. > > This reformats code to a more CodingStyle conforming style, > compiles it, verifies that the object code hasn't changed, > and git commits it too. > > You must have the necessary development tools, git, and a > recent git tree. Ideally use Greg KH's staging-next, which > can be retrieved via these commands: > > git clone git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git > git checkout staging-next > > To use this script try a sequence of commands like: > > cd <linux_repository> > git checkout -b <your_branch> > make allyesconfig > mkdir patches > ./scripts/reformat_with_checkpatch.sh drivers/staging/<dir>/*.[ch] > git format-patch --cover-letter -o patches/<your_branch> staging-next > git send-email patches/<your_branch> > > Signed-off-by: Joe Perches <joe(a)perches.com> > --- > scripts/reformat_with_checkpatch.sh | 141 ++++++++++++++++++++++++++++++++++++ > 1 file changed, 141 insertions(+) > create mode 100755 scripts/reformat_with_checkpatch.sh No --help option? How do I run this thing? I ran it on a file that had no problems and got a mess, all I think due to something odd in checkpatch itself right now: $ ./scripts/checkpatch.pl --file drivers/staging/lustre/include/linux/lnet/api.h Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 220 lines checked What's with that 'Useless use..." error? perl 5.20 in use, is that the issue? Anyway, running this script on the file gave me this: $ ./reformat_with_checkpatch drivers/staging/lustre/include/linux/lnet/api.h file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <whitespace neatening> types:<spacing,space_before_tab,pointer_location,trailing_whitespace,bracket_space> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: BRACKET_SPACE POINTER_LOCATION SPACE_BEFORE_TAB SPACING TRAILING_WHITESPACE drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <remove spaces before tabs> types:<space_before_tab> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: SPACE_BEFORE_TAB drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <fix label positions> types:<indented_label> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: INDENTED_LABEL drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <align arguments to parenthesis> types:<parenthesis_alignment> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: PARENTHESIS_ALIGNMENT drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <fix brace positions> types:<open_brace,braces,else_after_brace,while_after_brace> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: BRACES ELSE_AFTER_BRACE OPEN_BRACE WHILE_AFTER_BRACE drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <fix blank lines> types:<line_spacing> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: LINE_SPACING drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <use standard attributes> types:<prefer_packed,prefer_aligned> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: PREFER_ALIGNED PREFER_PACKED drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <remove unnecessary externs> types:<avoid_externs> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: AVOID_EXTERNS drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. file: <drivers/staging/lustre/include/linux/lnet/api.h> description: <update c90 comment style> types:<c99_comments> Useless use of greediness modifier '+' in regex; marked by <-- HERE in m/(^\+.*) {8,8}+ <-- HERE \t/ at ./scripts/checkpatch.pl line 2217. total: 0 errors, 0 warnings, 0 checks, 220 lines checked NOTE: Used message types: C99_COMMENTS drivers/staging/lustre/include/linux/lnet/api.h has no obvious style problems and is ready for submission. Is that expected? thanks, greg k-h

2 1

Re: [PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Greg KH July 12, 2014

July 12, 2014

On Fri, Jul 11, 2014 at 06:21:27PM -0700, Joe Perches wrote: > A simple script to run checkpatch --fix for various types of > of cleanups. > > This script is useful primarily for staging. > > This reformats code to a more CodingStyle conforming style, > compiles it, verifies that the object code hasn't changed, > and git commits it too. And 'git commits' it? Heh, I should just run this myself to clean up staging and beat everyone else to it... I know some people already have private versions of these things, might as well make it public for all to abuse :) thanks, greg k-h

2 1

[PATCH] reformat_with_checkpatch: Add automation to checkpatch
by Joe Perches July 12, 2014

July 12, 2014

A simple script to run checkpatch --fix for various types of of cleanups. This script is useful primarily for staging. This reformats code to a more CodingStyle conforming style, compiles it, verifies that the object code hasn't changed, and git commits it too. You must have the necessary development tools, git, and a recent git tree. Ideally use Greg KH's staging-next, which can be retrieved via these commands: git clone git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git git checkout staging-next To use this script try a sequence of commands like: cd <linux_repository> git checkout -b <your_branch> make allyesconfig mkdir patches ./scripts/reformat_with_checkpatch.sh drivers/staging/<dir>/*.[ch] git format-patch --cover-letter -o patches/<your_branch> staging-next git send-email patches/<your_branch> Signed-off-by: Joe Perches <joe(a)perches.com> --- scripts/reformat_with_checkpatch.sh | 141 ++++++++++++++++++++++++++++++++++++ 1 file changed, 141 insertions(+) create mode 100755 scripts/reformat_with_checkpatch.sh diff --git a/scripts/reformat_with_checkpatch.sh b/scripts/reformat_with_checkpatch.sh new file mode 100755 index 0000000..5415a8e --- /dev/null +++ b/scripts/reformat_with_checkpatch.sh @@ -0,0 +1,141 @@ +#!/bin/bash +# (c) 2014, Joe Perches <joe(a)perches.com> +# +# Automate checkpatch modifications and git commits to +# neaten code that doesn't conform to CodingStyle. +# Useful primarily for files in drivers/staging +# +# Licensed under the terms of the GNU GPL License version 2 + +declare -ar whitespace_types=( \ + 'whitespace_neatening:spacing,space_before_tab,pointer_location,trailing_whitespace,bracket_space' \ + 'remove_spaces_before_tabs:space_before_tab' \ + 'fix_label_positions:indented_label' \ + 'align_arguments_to_parenthesis:parenthesis_alignment' \ +) + +declare -ar changecode_types=( \ + 'fix_brace_positions:open_brace,braces,else_after_brace,while_after_brace' \ + 'fix_blank_lines:line_spacing' \ + 'use_standard_attributes:prefer_packed,prefer_aligned' \ + 'remove_unnecessary_externs:avoid_externs' \ + 'update_c90_comment_style:c99_comments' \ +) + +checkpatch_update () +{ + file=$1 + + desc=$(echo $2 | cut -f1 -d: | sed 's/_/ /g') + types=$(echo $2 | cut -f2- -d:) + + echo "file: <$file> description: <$desc> types:<$types>" + + ./scripts/checkpatch.pl --file --fix-inplace --strict --types="$types" $file + + checkpatch_fixes=$file.diff + git diff --stat -p --exit-code $file > $checkpatch_fixes + if [ $? == 0 ] ; then + rm -f $checkpatch_fixes + return + fi + + basename=$(basename $file) + if [ "${basename##*.}" == "c" ] ; then + + git checkout $file + obj="$(echo $file | sed 's/\.c$/\.o/')" + if [ -e $obj ] ; then + rm -f $obj + fi + + echo "Compiling original version..." + + ${CROSS_COMPILE}make $obj + + ${CROSS_COMPILE}objdump -D $obj | \ + sed "s/^[[:space:]]\+[0-9a-f]\+//" > $obj.old + + patch -p1 < $checkpatch_fixes + + echo "Compiling modified version..." + + ${CROSS_COMPILE}make $obj + + ${CROSS_COMPILE}objdump -D $obj | \ + sed "s/^[[:space:]]\+[0-9a-f]\+//" > $obj.new + + echo "Comparing objects..." + diff -Nurd $obj.new $obj.old + if [ $? -ne 0 ] ; then + echo "Object differences exist! - Verify changes before commit!" + read -s -p "Press the 'enter' key to continue: " + else + echo "No object differences found" + fi + rm -f $obj.old + rm -f $obj.new + fi + + echo "running checkpatch on possible checkpatch fixes..." + + ./scripts/checkpatch.pl --no-summary --no-signoff $checkpatch_fixes + rm -f $checkpatch_fixes + + echo "Verify checkpatch output and make any necessary changes" + echo "Edit '$file' if appropriate" + read -s -p "Press the 'enter' key to continue: " + + commit_log_file=$(mktemp git_commit.XXXXXX) + + if [ -e $commit_log_file ] ; then + rm -f $commit_log_file + fi + + if [[ $file =~ ^drivers/staging/ ]] ; then + echo -n "staging: " >> $commit_log_file + fi + echo "$(basename $(dirname $file)): checkpatch cleanup: $desc" >> $commit_log_file + echo "" >> $commit_log_file + git diff --exit-code -w $file > /dev/null + if [ $? == 0 ] ; then + echo "whitespace changes only - git diff -w shows no difference" >> $commit_log_file + fi + + git diff $file | cat + + echo "Ready to commit - First verify all diffs and make any necessary changes" + echo "" + read -r -p "Would you like to commit these changes <y>: " response + response=${response,,} + if [[ $response =~ ^(yes|y|)$ ]] ; then + git commit -s -F $commit_log_file -e $file + else + git checkout $file + fi + + rm -f $commit_log_file +} + +which git > /dev/null +if [ $? -ne 0 ] || [ ! -e .git ] ; then + echo "git or git directory not found - run this from the top of the source tree" + exit 1 +fi + +for file in "$@" ; do + + if [ ! -f $file ] ; then + echo "Argument '$file' is not a file" + continue + fi + + for type in "${whitespace_types[@]}" ; do + checkpatch_update $file $type + done + + for type in "${changecode_types[@]}" ; do + checkpatch_update $file $type + done + +done -- 1.8.1.2.459.gbcd45b4.dirty

1 0

Eudyptula Challenge
by Javier July 11, 2014

July 11, 2014

Hello all, Is it convenient to work on our main kernel on this projects? Or, would you suggest using a different partition kernel to practice this? I am afraid of messing up my work computer. Thank you. -- Javier Pajuelo

3 4

Practical way to rename kernel-release on custom kernel
by Safarin July 11, 2014

July 11, 2014

Hi Guys, This question might be very easy for all of you. But seem this is kernelnewbies mailing list. I need to ask this question. I in the middle of learning to build custom kernel and my question is how to have custom name for kernel version. $ uname -r I can see custom kernel print message. >From what I read, we can change inside .config CONFIG_LOCALVERSION="-ARCH" will produce the kernel release naming something like this. 3.16.0-rc3-ARCH But somehow, when I try to install the modules $make modules_install I can see makefile create folder inside the lib/modules with naming 3.16.0-rc3-ARCH-00570-g110e430 110e430 is actually commit id and I check that naming is generated by #define UTS_RELEASE "3.16.0-rc3-ARCH-00570-g110e430" in the "include/generated/utsrelease.h" header which will overwrite "include/config/kernel.release". Back to the question, what is the practical way to naming the kernel release for custom kernel? with additional question for understanding more from where that utsrelease came from? Thanks, Regards, Safarin

4 5

Re: Do per cpu variables need to be atomic?
by Pranith Kumar July 11, 2014

July 11, 2014

On Thu, Jul 10, 2014 at 4:55 PM, Christoph Lameter <cl(a)gentwo.org> wrote: > On Thu, 10 Jul 2014, Pranith Kumar wrote: > >> > Per cpu data structures are for the use of one cpu exclusively. If you use >> > the variable as intended then there is no atomic_t required. If you want >> > to update the percpu variable from other processors then atomic_t is >> > required. But then this is not a proper percpu variable anymroe. >> > >> >> That also makes using barriers while updating such variables >> meaningless I guess? like > > Correct. I understand one of the advantage of using these per cpu variables is avoiding disabling interrupts. But what happens when this cpu is preempted? The new processor will be working on the other cpu variables, but without barriers will it see the latest values? > >> >> struct data *p = this_cpu_ptr(&datap); >> > >> > This converts the offset into an address that can be generally used as >> > a pointer (but no longer with this_cpu operations). >> > >> >> So we can pass around this pointer to other cpus but not the previous >> __percpu pointer? > > You can pass this pointer around to other cpus. Caveat: the other cpus > will be accessing the percpu data from another cpu if the actually use > this pointer. Not recommended. > > A __percpu offset (dont call it pointer) is always relative to the current > cpu. So if the offset is used then the instance of the percpu variable for > the current cpu is used and not the instance for another processor. If you > pass the offset to other processors then this_cpu_ptr() will create the > address for the currrent instance and this_cpu ops will always refer to > the instance associated with the currently executing processor. > > Use of __percpu offsets is recommended. OK. > >> I think adding this information to the documentation would be a great >> idea. Especially the "__percpu" annotation. Is it ok if I try? > > Sure. But there is already some documentation in > linux/Documentation/this_cpu_ops.txt. > -- Pranith

2 1