2 ms·
There's a bit of a question about what "fixed" means here. It seems like most of these packages have applied the prefix-and-suffix mitigation measure: namely, w
by geofft 12y ago
There's a bit of a question about what "fixed" means here. It seems like most of these packages have applied the prefix-and-suffix mitigation measure: namely, while the function importer code is still present, it only works on environment variables with specific, uncommon names. It's basically the case that in any context where you can set environment variables with arbitrary names, you can do other sorts of evil.
Several distros have not released patches for the later parser vulnerabilities _now that that codepath isn't reachable_ unless you're able to make an environment variable with the specific sort of odd name it requires. The test program here does specifically check for all the known patches with custom variable names, and maybe it's good to fix them, but it seems slightly disingenuous to suggest this is "unfixed". Remote attackers shouldn't be able to create environment variables with these names anyway.
I think I'd prefer a bash that has applied the prefix-and-suffix hardening and not the patches for the lcamtuf bugs, than one that's fixed all the parser bugs we know about but still allows function imports without the prefix-and-suffix requirement.... there are probably parser bugs we don't know about.