Bug 940217 – dev-util/pahole-1.27-r1 - [ncurses-6.5] [gcc-15] tar: .../BAD_TAR_INVOCATION: Cannot open: Not a directory

• Product: Gentoo Linux
• Status: RESOLVED FIXED

[Description Toralf Förster 2024-09-25 07:22:46 UTC]

 * USE:        abi_x86_64 amd64 elibc_glibc kernel_linux python_single_target_python3_12 verify-sig
 * FEATURES:   network-sandbox preserve-libs sandbox userpriv usersandbox
 * Using python3.12 to build
>>> Unpacking source...
 * Unpacking dwarves-1.27.tar.xz ...
 * Verifying (stdin) ...
tar: /dev/null/BAD_TAR_INVOCATION: Cannot open: Not a directory
tar: Error is not recoverable: exiting now
[  ERROR] OpenPGP verification failed for <_io.BufferedReader name='<stdin>'> (sig in /var/tmp/portage/dev-util/pahole-1.27-r1/distdir/dwarves-1.27.tar.sign):

  -------------------------------------------------------------------

  This is an unstable amd64 chroot image at a tinderbox (==build bot)
  name: 23.0_desktop-20240924-140001

  UNMASKED:
    Please re-assign to toolchain@ if you get a test failure in C, C++, or Fortran code which makes no sense.
  /etc/portage/package.unmask/60gcc:<sys-devel/gcc-15.0.9999:15
    Requested by sam
  /etc/portage/package.unmask/50unstable:>=sys-libs/ncurses-6.5

  The attached etc.portage.tar.xz has all details.
  -------------------------------------------------------------------

gcc-config -l:
 [1] x86_64-pc-linux-gnu-15 *
clang/llvm (if any):
Python 3.12.6
Available Ruby profiles:
  (none found)
Available Rust versions:
  [1]   rust-bin-1.81.0 *
php cli (if any):

  HEAD of ::gentoo
commit 3303a33103217d83731520fac64a62f1d517bce5
Author: Repository mirror & CI <[e-mail]>
Date:   Tue Sep 24 18:19:00 2024 +0000

    2024-09-24 18:18:59 UTC

emerge -qpvO =dev-util/pahole-1.27-r1
[ebuild   R   ] dev-util/pahole-1.27-r1  USE="verify-sig* -debug" PYTHON_SINGLE_TARGET="python3_12 -python3_10 -python3_11"

[Comment 1 Toralf Förster 2024-09-25 07:22:47 UTC]

Created attachment 903676 [details]
emerge-info.txt

[Comment 2 Toralf Förster 2024-09-25 07:22:48 UTC]

Created attachment 903677 [details]
dev-util:pahole-1.27-r1:20240924-203556.log

[Comment 3 Toralf Förster 2024-09-25 07:22:49 UTC]

Created attachment 903678 [details]
emerge-history.txt

[Comment 4 Toralf Förster 2024-09-25 07:22:50 UTC]

Created attachment 903679 [details]
environment

[Comment 5 Toralf Förster 2024-09-25 07:22:51 UTC]

Created attachment 903680 [details]
etc.clang.tar.xz

[Comment 6 Toralf Förster 2024-09-25 07:22:51 UTC]

Created attachment 903681 [details]
etc.portage.tar.xz

[Comment 7 Toralf Förster 2024-09-25 07:22:52 UTC]

Created attachment 903682 [details]
logs.tar.xz

[Comment 8 Toralf Förster 2024-09-25 07:22:53 UTC]

Created attachment 903683 [details]
qlist-info.txt

[Comment 9 Matthias Schwarzott 2024-09-25 15:14:53 UTC]

I found this commit touching pahole-ebuild (and others):

commit 48c7150d63793ba6f5a8a8c5429b3de929920b48
Author: Mike Gilbert <[e-mail]>
Date:   Tue Sep 24 15:56:03 2024 -0400

    */*: replace "tar -x" with "tar -xf -"
    
    Closes: https://bugs.gentoo.org/940201
    Signed-off-by: Mike Gilbert <[e-mail]>
[...]
 25 files changed, 25 insertions(+), 25 deletions(-)

Change:
--- a/dev-util/pahole/pahole-1.27-r1.ebuild
+++ b/dev-util/pahole/pahole-1.27-r1.ebuild
@@ -59,7 +59,7 @@ src_unpack() {
        if use verify-sig; then
                einfo "Unpacking ${MY_P}.tar.xz ..."
                verify-sig_verify_detached - "${DISTDIR}"/${MY_P}.tar.sign \
-                       < <(xz -cd "${DISTDIR}"/${MY_P}.tar.xz | tee >(tar -x))
+                       < <(xz -cd "${DISTDIR}"/${MY_P}.tar.xz | tee >(tar -xf -))
                assert "Unpack failed"
        else
                default

Not sure why it can fail in that way. Is the tar version on that tinderbox special?

[Comment 10 Matthias Schwarzott 2024-09-25 15:20:09 UTC]

I wonder why such a complicated commandline is needed at all.

Why not just 3 simple steps:
1. uncompress
2. verify
3. unpack

[Comment 11 Sam James 2024-09-25 15:21:10 UTC]

For the tinderbox q: https://bugs.gentoo.org/903921#c0.

For the complicated q: Yeah, floppym and I discussed this a bit on IRC. I don't actually care other than having it be consistent between kernel.org packages with verify-sig. mgorny changed a few of them and I wanted it to be the same across such packages.

[Comment 12 Matthias Schwarzott 2024-09-25 15:31:53 UTC]

(In reply to Sam James from comment #11)
> For the tinderbox q: https://bugs.gentoo.org/903921#c0.

Now I see. The found commit is not the cause but the fix of this bug.
floppym fixed the bug before it was reported :-)

@toralf: Ok to resolve this bug?

> 
> For the complicated q: Yeah, floppym and I discussed this a bit on IRC. I
> don't actually care other than having it be consistent between kernel.org
> packages with verify-sig. mgorny changed a few of them and I wanted it to be
> the same across such packages.

As long as it works I am fine. But in general pipelines can loose exit-codes of commands.

[Comment 13 Toralf Förster 2024-09-25 15:34:23 UTC]

(In reply to Matthias Schwarzott from comment #12)
> @toralf: Ok to resolve this bug?

sure, if it is solved.