Message-ID: <20191031045242.GA3130@Gentoo>
Date:   Thu, 31 Oct 2019 10:22:45 +0530
From:   Bhaskar Chowdhury <unixbhaskar@...il.com>
To:     Randy Dunlap <rdunlap@...radead.org>
Cc:     bfields@...ldses.org, yamada.masahiro@...ionext.com,
        michal.lkml@...kovi.net, linux-kbuild@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] scripts: prune-kernel:remove old kernels and modules dir
 from system


Thank you so much Randy. I am awfully sorry about the ugly code ,
hopefully come up with a better one. If I understood your expectation.

I know we all are blessed with limited resources like patience.Thanks
for holding this long...little more and we will be over it.

:)

On 21:32 Wed 30 Oct 2019, Randy Dunlap wrote:
>On 10/30/19 8:37 PM, Bhaskar Chowdhury wrote:
>> Thank you Randy, my response are inline. Please look at it.I am
>> wondering , what else I could do get this damn! thing going??
>> Any clue??
>> 
>> On 19:33 Wed 30 Oct 2019, Randy Dunlap wrote:
>>> Hi,
>>>
>>> On 10/30/19 2:54 AM, Bhaskar Chowdhury wrote:
>>>> This patch will remove old kernels and modules directorey related
>>>> to that kernel from the system by interactively and silently.Here
>>>> are few interactions with the scripts
>>>>
>>>> 1)
>>>>
>>>> ✔ ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>> 14:52 $ ./scripts/prune-kernel -h
>>>> Usage: prune-kernel [ri]
>>>>
>>>>  -r | --remove kernel_ver modules_dir_name
>>>>
>>>>   -i | --interactive use as interactive way
>>>>   ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>>     14:52 $ ./scripts/prune-kernel --help
>>>>   Usage: prune-kernel [ri]
>>>
>>> That "[ri]" is confusing to me.
>> This are the options one has to pass with the script.Like below:
>
>I know that.  But it's missing '-', so it looks like
>$ prune-kernel r 5.2.5-foobar
>
>would work.
>
Will correct that.
>>>>
>>>>    -r | --remove kernel_ver modules_dir_na]
>>>>
>>>>     -i | --interactive use as interactive way
>>>>     2)
>>>>
>>>>  ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>>  14:52 $ ./scripts/prune-kernel -r 5.3.3
>>>>  You need to provide kernel version and modules dir name
>>>>  
>>>>  ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>>  14:53 $ ./scripts/prune-kernel -r
>>>>  You need to provide kernel version and modules dir name
>>>>  
>>>>  ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>>  14:54 $ ./scripts/prune-kernel -r 5.3.3 5.3.3-foo
>>>
>>> This one above didn't remove any kernel files.
>>> Needs more testing.
>> It does remove but silently, as you and Bruce asked for this feature.
>
>No, see the code below for -r...
>
Okay ...look like some some uniformity missing
>>>> 3)
>>>>
>>>> $ ./scripts/prune-kernel --remove
>>>> You need to provide kernel version and modules dir name
>>>>
>>>> ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>> 14:55 $ ./scripts/prune-kernel --remove 5.3.3
>>>> You need to provide kernel version and modules dir name
>>>>
>>>> ✘-1 ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>> 14:55 $ ./scripts/prune-kernel --remove 5.3.3 5.3.3-foo
>>>>
>>>>
>>>> 4)14:55 $ ./scripts/prune-kernel -i
>>>>
>>>> Enter kernel version to remove or blank/empty to exit:
>>>>
>>>>
>>>> 5)14:57 $ ./scripts/prune-kernel --interactive
>>>>
>>>> Enter kernel version to remove or blank/empty to exit:
>>>> ✔ ~/git-linux/linux-kbuild [master|AM 1/1 ↑·59|✔]
>>>>
>>>>
>>>> 6)14:59 $ ./scripts/prune-kernel --interactive
>>>>
>>>> Enter kernel version to remove or blank/empty to exit:5.3.3
>>>> Please give the full modules directory name to remove:5.3.3-foo
>>>>
>>>>
>>>>
>>>> Removed kernel version:5.3.3 and associated modules:5.3.3-foo ...Done.
>>>>
>>>>
>>>> 7)15:00 $ ./scripts/prune-kernel -i
>>>>
>>>> Enter kernel version to remove or blank/empty to exit:5.3.3
>>>> Please give the full modules directory name to remove:5.3.3-foo
>>>>
>>>>
>>>>
>>>> Removed kernel version:5.3.3 and associated modules:5.3.3-foo ...Done.
>>>>
>>>>
>>>> Signed-off-by: Bhaskar Chowdhury <unixbhaskar@...il.com>
>>>> ---
>>>>  scripts/prune-kernel | 63 ++++++++++++++++++++++++++++++++++++++++++++
>>>>  1 file changed, 63 insertions(+)
>>>>
>>>> diff --git a/scripts/prune-kernel b/scripts/prune-kernel
>>>> index a25aa2160d47..a91010d0e2af 100755
>>>> --- a/scripts/prune-kernel
>>>> +++ b/scripts/prune-kernel
>>>> @@ -1,3 +1,66 @@
>>>>  #!/bin/bash
>>>>  # SPDX-License-Identifier: GPL-2.0
>>>> +#This script will delete old kernels and modules directory related to it
>>>> +#-h with the script will show you the help
>>>> +#-r with the script take two parameter: kernel_ver and modules_dir_name
>>>> +#-i with the script allow you do the removing interactive way
>>>>
>>>> +flag=$1
>>>> +kernel_ver=$2
>>>> +modules_dir_name=$3
>>>> +boot_dir=/boot
>>>> +modules_dir=/lib/modules
>>>> +
>>>> +remove_old_kernel() {
>>>> +    cd $boot_dir
>>>> +    rm -If vmlinuz-$kernel_version System.map-$kernel_version config-$kernel_version
>>>> +    return 0
>>>> +}
>>>> +
>>>> +remove_old_modules_dir() {
>>>> +    cd $modules_dir
>>>> +    rm -rf $modules_version
>>>> +    return 0
>>>> +}
>>>> +
>>>> +usage() {
>>>> +    printf "Usage: $(basename $0) [ri] \n"
>>>> +    printf "\n -r | --remove kernel_ver modules_dir_name \n"
>>>> +    printf "\n -i | --interactive use as interactive way \n"
>>>> +}
>>>> +
>>>> +for arg in "$@"
>>>
>>> what is the purpose (use) of "arg" here?
>> 
>> This variable is used in case statement below.
>
>I can't find any use of 'arg' anywhere else in the script.
>Please show me where it is.
My bad and apologies for overlooking.
>
>>> what is the purpose of the for loop?
>>>
>> It scan through all the parameters pass .
>
>What is this script supposed (expected) to do with multiple arg parameters?
>
It uses multiple parameter
>>> Is any 'shift' needed to consume (or discard) the first 3 positional
>>> command line arguments?
>> Nope, that is not required. And I haven't use any.
>>>
>>>> +do
>>>> +    case "$flag" in
>>>> +        -i | --interactive)
>>>> +            printf "\nEnter kernel version to remove or blank/empty to exit:%s"
>>>> +            read kernel_version
>>>> +            if [[ $kernel_version != "" ]]; then
>>>> +                remove_old_kernel
>>>> +                printf "Please give the full modules directory name to remove:%s"
>>>> +                read modules_version
>>>> +                if [[ $modules_version != "" ]]; then
>>>> +                    remove_old_modules_dir
>>>> +                    printf "\n\n\n Removed kernel version:$kernel_version and associated modules:$modules_version ...Done. \n"
>>>
>>> This message is only printed if $modules_version is non-empty.  If it is empty,
>>> remove_old_kernel() has silently removed some kernel files (if they existed).
>> it will fail to remove anything if the kernel_version or modules_version
>> are empty and importantly exit.
>>>
>>>> +                else
>>>> +                    exit 1
>>>> +                fi
>>>> +            fi
>>>> +            ;;
>>>> +        -h | --help)
>>>> +            usage
>>>> +            exit 1
>>>> +            ;;
>>>> +        -r | --remove)
>>>> +            if [[ $# -ne 3 ]]; then
>>>> +                printf "You need to provide kernel version and modules dir name\n"
>>>> +                exit 1
>>>> +            else
>>>> +                cd $boot_dir
>>>> +                rm -f $kernel_ver
>>>
>>> That 'rm' doesn't remove any files.  Compare what remove_old_kernel() does.
>> No,it is not using that function rather take the parameter from the
>> commandline and get into boot dir match with it and remove it.
>
>But it doesn't do that.  I tested it.  It should be more like what
>rmeove_old_kernel() does:
>
>		rm -If vmlinuz-$kernel_ver System.map-$kernel_ver config-$kernel_ver
>
>and if not, please explain why not.
Okay, again some uniformity missing in the code, I would like to your
suggested method,i.e call remove_old_kernel to do the job instead of 
depending on individual kernel.
>
>
>>>> +                cd $modules_dir
>>>> +                rm -rf $modules_dir_name
>>>> +            fi
>>>> +            ;;
>>>> +    esac
>>>> +done
>>>> -- 
>>>
>>> The script, after this patch is applied, still contains the old script's for-loop
>>> at the end of the "new" prune-kernel script.
>> 
>> Amazing! now it needs some explanation how I did...you probably want
>> that ..here are the steps....
>> 1)fetch that prune-kernel file from repos , which contains Bruce's code
>> in it.
>> 2) get inot it by editior, remove all except first two lines i.e bash
>> interpreter and PSDX .
>> 3)Save and commit it locally.
>> 4) Write my own code
>> 5) save it and commit it locally.
>> 6) go one level up use checkpatch to see anything bad creeps in
>> 7) Fixed the damn things if it reports.
>> 8) create the patch
>> 9) test it
>> 10) Send it.
>> 
>> Now, how the heck , that for loop is getting staying there is a mystry
>> to me!! Look like that is ruin all the work.
>> irk...
>
>I don't know.  I just know that it's not working AFAICT.
Thank you, will be more vigilant in next iteration.
>
>-- 
>~Randy
>

Download attachment "signature.asc" of type "application/pgp-signature" (489 bytes)

Powered by blists - more mailing lists