Sean McBride

>So to summarize, is this correct? I think so. >The code never calls usbi_reallocf() with 0, but if it would, this could cause a double-free on Windows? Correct. On *nix,...

It's not ***needed***, but it's beneficial. If we remove it, this code: ```c priv->descriptors = usbi_reallocf(priv->descriptors, alloc_len); if (!priv->descriptors) { if (fd != wrapped_fd) close(fd); return LIBUSB_ERROR_NO_MEM; } ``` would...

@dlmiles did you try making them const? Does it still compile? i.e. is libusb actually modifying those `struct timeval`s? These could be made conditionally const with `LIBUSB_API_VERSION` checks. Something like:...

I wouldn't be surprised if this isn't all the same root cause as TSan reports with the [stress_mt](https://github.com/libusb/libusb/issues/1615#issuecomment-2731264493)...

>From other side, there is nothing important or serious to hurry up with the next release. Isn't there? Supporting newer OpenSSH should be kinda urgent, no? i.e. https://github.com/fail2ban/fail2ban/issues/3785 https://github.com/fail2ban/fail2ban/issues/4028

> 1. Please don't confuse release of fail2ban here with the release of fail2ban for your distribution. I understand. > 2. I spoke about basic fail2ban functionality (API, etc), not...

> it is fixed in upstream I know, and that's great! > a patch is a git diff or cherry-pick that can be applied in distro repo and released as...

I'd be interested to try this in our app when you think it's in shape...

Glad to see this work. Again, when it's in better shape, I'll give it a try in our app, which has various tests of its own that might exercise some...

@hjmjohnson @lassoan these days, many OSes, like macOS in my case, sandboxes processes from the filesystem and allows them only to access files the user has granted permission to access....