proc-macro-srv: drop unnecessary usage of RTLD_DEEPBIND

[Fabian-Gruenbichler]

the constant is wrong on some platforms (e.g., on mips64el it's 0x10, and 0x8 is RTLD_NOLOAD which makes all this functionality broken), and it is no longer needed because #60593 got fixed by #99944 in the meantime.

[rustbot]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @Mark-Simulacrum (or someone else) some time within the next two weeks.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

• @rustbot author: the review is finished, PR author should check the comments and take action accordingly
• @rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

[rustbot]

rust-analyzer is developed in its own repository. If possible, consider making this change to rust-lang/rust-analyzer instead.

cc @rust-lang/rust-analyzer

[Fabian-Gruenbichler]

further references:

failing mips64el build in Debian caused by this: https://buildd.debian.org/status/fetch.php?pkg=rustc&arch=mips64el&ver=1.84.0%2Bdfsg1-1~exp1&stamp=1736831677&raw=1

alternatively, it would also be possible to have a hardcoded list of arch-specific constants instead..

[bjorn3]

#99944 is not really a fix. It is still possible for a proc-macro to dynamically link against libstd.so, which would expose the issue again as the proc-macro would import functions from librustc_driver.so's copy of libstd without RTLD_DEEPBIND rather than the libstd.so that the proc-macro should actually link against.

$ echo | rustc - --crate-type dylib --crate-name foo -Cprefer-dynamic
$ echo 'extern crate foo;' | rustc - --crate-type proc-macro --extern foo=libfoo.so
$ ldd librust_out.so
        linux-vdso.so.1 (0x00007fffbd9fb000)
        libstd-22be60875a4ac8d7.so => not found

[Fabian-Gruenbichler]

ack, then I'll open another MR that properly conditionalizes RTLD_DEEPBIND based on the target..

[bjorn3]

If you do, you may be able to use the RTLD_DEEPBIND constant from the libc crate at the same time to handle differences between architectures. You did still need conditional compilation based on the used libc implementation though.

[Fabian-Gruenbichler]

glibc and uclibc seem to agree (0x8 for everything but mips, 0x10 for mips), freebsd's is 0x04000, musl and bionic don't seem to have it at all.. I think deferring to libc and letting them differentiate further if needed makes the most sense?

[bjorn3]

The libc crate will not define RTLD_DEEPBIND if the libc implementation for the target doesn't support it. As such you need conditional compilation based on the target to only use RTLD_DEEPBIND when the libc implementation supports it.

[Fabian-Gruenbichler]

closed in favor of #135591 , hope that's about what you requested ;)