Re: [PATCH v3 1/2] common: Move exit related functions to a common/exit

[Ritesh Harjani (IBM) <ritesh.list@gmail.com>]

Zorro Lang <zlang@redhat.com> writes:

> On Thu, May 01, 2025 at 08:47:46AM +0530, Ritesh Harjani wrote:
>> "Nirjhar Roy (IBM)" <nirjhar.roy.lists@gmail.com> writes:
>> 
>> > Introduce a new file common/exit that will contain all the exit
>> > related functions. This will remove the dependencies these functions
>> > have on other non-related helper files and they can be indepedently
>> > sourced. This was suggested by Dave Chinner[1].
>> > While moving the exit related functions, remove _die() and die_now()
>> > and replace die_now with _fatal(). It is of no use to keep the
>> > unnecessary wrappers.
>> >
>> > [1] https://lore.kernel.org/linux-xfs/Z_UJ7XcpmtkPRhTr@dread.disaster.area/
>> > Suggested-by: Dave Chinner <david@fromorbit.com>
>> > Signed-off-by: Nirjhar Roy (IBM) <nirjhar.roy.lists@gmail.com>
>> > ---
>> >  check           |  2 ++
>> >  common/config   | 17 -----------------
>> >  common/exit     | 39 +++++++++++++++++++++++++++++++++++++++
>> >  common/preamble |  3 +++
>> >  common/punch    | 39 +++++++++++++++++----------------------
>> >  common/rc       | 28 ----------------------------
>> >  6 files changed, 61 insertions(+), 67 deletions(-)
>> >  create mode 100644 common/exit
>> >
>> > diff --git a/check b/check
>> > index 9451c350..bd84f213 100755
>> > --- a/check
>> > +++ b/check
>> > @@ -46,6 +46,8 @@ export DIFF_LENGTH=${DIFF_LENGTH:=10}
>> >  
>> >  # by default don't output timestamps
>> >  timestamp=${TIMESTAMP:=false}
>> > +. common/exit
>> > +. common/test_names
>> 
>> So this gets sourced at the beginning of check script here.
>> 
>> >  
>> >  rm -f $tmp.list $tmp.tmp $tmp.grep $here/$iam.out $tmp.report.* $tmp.arglist
>> >  
>> <...>
>> > diff --git a/common/preamble b/common/preamble
>> > index ba029a34..51d03396 100644
>> > --- a/common/preamble
>> > +++ b/common/preamble
>> > @@ -33,6 +33,9 @@ _register_cleanup()
>> >  # explicitly as a member of the 'all' group.
>> >  _begin_fstest()
>> >  {
>> > +	. common/exit
>> > +	. common/test_names
>> > +
>> 
>> Why do we need to source these files here again? 
>> Isn't check script already sourcing both of this in the beginning
>> itself?
>
> The _begin_fstest is called at the beginning of each test case (e.g. generic/001).
> And "check" run each test cases likes:
>
>   cmd="generic/001"
>   ./$cmd
>
> So the imported things (by "check") can't help sub-case running

aah right. Each testcase is inoked by "exec ./$seq" and it won't have
the function definitions sourced from the previous shell process. So we
will need to source the necessary files again within the test execution.

-ritesh