News for hackers. Stuff that matters.
Re:Glibc and Dt_gnu_hash by Jasper_ on Monday August 22, 2022 @02:47AM (#32548129)
One fun thing worth noting is that apparently the Linux version of Shovel Knight broke. As a developer of that game, I looked through our source code and was unable to find any reliance on libdl.so.2 or DT_HASH. I am unsure where the breakage occurred, it's possible it's in one of our third party middleware libraries, but it is exhausting that it did happen.
Re:Glibc and Dt_gnu_hash by timdoug on Sunday August 21, 2022 @10:19PM (#32546593)
For a different perspective of the same issue: https://news.ycombinator.com/item?id=32471624

(don't be fooled by the article's title, I regret I was the first time around)

Re:Glibc and Dt_gnu_hash by MaskRay on Sunday August 21, 2022 @07:43PM (#32545651)
The subject should be "glibc and DT_GNU_HASH" :) Hope this gives some background information for "Easy Anti-Cheat" users.
Re:Glibc and Dt_gnu_hash by CaliforniaKarl on Sunday August 21, 2022 @08:34PM (#32545940)
HN automatically redoes the capitalization of post titles, and it's easy forget to check it after posting.
Re:Glibc and Dt_gnu_hash by JonChesterfield on Monday August 22, 2022 @05:18AM (#32549132)
This seems fine to me. The newer table format is better. It'll break a bunch of stuff that can't be rebuilt from source. Not sympathetic to proprietary software dynamically linked against glibc.

Contrast with rpath, which was repurposed without change of name to set runpath, with obnoxiously different semantics and a new linker flag for 'by rpath I actually meant rpath, not runpath' as something that actively wastes my time on a roughly monthly basis.