[Buildroot] [PATCH 3/4] support/scripts/check-host-bins: skip RPATH check for missing libraries

[Thomas Petazzoni]

On Wed, Sep 09, 2026 at 07:39:24AM -0500, Matthew Weber wrote:
> An ELF with only non-toolchain dependencies absent from HOST_DIR cannot use an
> RPATH to find them. Report those missing dependencies, but skip the misleading
> RPATH diagnostic.
> 
> Assisted-by: GitHub Copilot [VS Code]
> Signed-off-by: Matthew Weber <matt at thewebers.ws>
> ---
> Depends on v3 of the check-host-bins series:
> https://patchwork.ozlabs.org/project/buildroot/list/?series=520240
> 
> The test script is available at:
> https://gist.github.com/matthew-l-weber/34fae15a4319a81462961eadf0e31164
> 
> While exercising v3's missing-library check, an ELF with only external
> non-toolchain dependencies was reported as both missing a library and missing
> an RPATH. Since no relevant library exists in HOST_DIR, the latter diagnostic
> is misleading.
> 
> Test 3h places /bin/ls in a synthetic HOST_DIR without its non-toolchain
> dependencies. It verifies that check-host-bins reports the missing library
> but does not also report an RPATH error when there is no HOST_DIR library to
> locate.

Hu? Why do we care?

If an ELF has non-toolchain library dependencies, then those must be
in HOST_DIR, and the binary must have an RPATH.

> 
> Test 3l builds a binary with its non-toolchain dependency installed in the
> synthetic HOST_DIR/lib but an invalid RPATH. It verifies that this valid
> RPATH-check case still reports the RPATH diagnostic and not a missing-library
> error.

My understanding is that this is the nominal case and this commit
doesn't change anything about this.

Honestly, AI generated stuff is meh. Takes a massive amount of brain
energy because the commit log is meh, all to figure out that the
change is useless (if I'm not wrong, of course).

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com