Re: [dpdk-dev] [PATCH] buildtools: fix build with meson 0.60

[Bruce Richardson <bruce.richardson@intel.com>]

On Tue, Oct 26, 2021 at 10:32:39PM +0300, Dmitry Kozlyuk wrote:
> Meson 0.60 switched the format of uninstalled static libraries
> to thin archives, that is, they contain only paths to object files,
> not the files themselves. Files cannot be extracted in this case,
> resulting in build errors:
> 
>     ar: `x' cannot be used on thin archives.
> 
> Handle thin archives when invoking pmdinfogen
> by directly using the files referenced in the archive.
> 
> Bugzilla ID: 836
> Fixes: e6e9730c7066 ("buildtools: support object file extraction for Windows")
> Cc: stable@dpdk.org
> 
> Reported-by: Michal Berger <michallinuxstuff@gmail.com>
> Signed-off-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>

Thanks for looking into this and proposing a fix. Some thoughts inline
below.

/Bruce

> ---
>  buildtools/gen-pmdinfo-cfile.py | 23 ++++++++++++++---------
>  1 file changed, 14 insertions(+), 9 deletions(-)
> 
> diff --git a/buildtools/gen-pmdinfo-cfile.py b/buildtools/gen-pmdinfo-cfile.py
> index 58fe3ad152..3453e5b4b9 100644
> --- a/buildtools/gen-pmdinfo-cfile.py
> +++ b/buildtools/gen-pmdinfo-cfile.py
> @@ -8,13 +8,18 @@
>  import tempfile
>  
>  _, tmp_root, ar, archive, output, *pmdinfogen = sys.argv
> -with tempfile.TemporaryDirectory(dir=tmp_root) as temp:
> -    run_ar = lambda command: subprocess.run(
> -        [ar, command, os.path.abspath(archive)],
> -        stdout=subprocess.PIPE, check=True, cwd=temp
> -    )
> -    # Don't use "ar p", because its output is corrupted on Windows.
> -    run_ar("x")
> -    names = run_ar("t").stdout.decode().splitlines()
> -    paths = [os.path.join(temp, name) for name in names]
> +archive = os.path.abspath(archive)
> +names = subprocess.run([ar, "t", archive],
> +    stdout=subprocess.PIPE, check=True).stdout.decode().splitlines()
> +with open(archive, "rb") as f:
> +    is_thin = f.read(7) == b"!<thin>"

This part seems overly low-level, and a bit nasty to me, compared to the
rest of the script.

Since the thin archive files already exist, I wonder if we could work on a
file by file basis here, and ignore generally the type of the ".a" file.
For example, something like:

    o_files = []
    for n in names:
        if os.path.exists(n):
            o_files += n
        else
            <extract file>
            o_files += <extracted_path>

WDYT?