Anastasios Andronidis

Thanks a lot! Much appreciated!

Can we also upgrade Go to 1.7 now that kubernetes/kubernetes#28742 is merged? Some tests are failing under 1.6.

Going even further down the rabbit hole, when the `%fs` register switches we also need to update the `thread_local` variables in SaBRe. **But why do we need these vars to...

> Thanks, @andronat. I haven't looked at the details, but I think the TLS switching should be separate from adding support for mimalloc. In other words, SaBRe should work the...

To throw another wrench in the mix, TLS stores its variables (i.e. `thread_local` vars like those [here](https://github.com/srg-imperial/SaBRe/pull/54/files#diff-d43c18fe4c200f231a84690c421a2338R201-R202)) in fixed offsets at compile-time. As an example, you can verify this by...

> Before the TLS switch you can take a pointer to the address of the thread local variable and access it through there as mentioned in GCC’s documentation: Hmm, I...

An update on my side: - @daniel-grumberg sorry for the delayed response, I didn't expect to take me that long... For what it's worth, I implemented a global hash table...

Just another update on my side, I was able to run [zboxfs](https://github.com/zboxfs/zbox) as a new plugin with the current PR. To make it run though I had to solve another...

@parras > Thanks for your great endeavour trying to make this TLS scheme work. Unfortunately you're now running into the kind of intricacies that I absolutely wanted to avoid. Yea,...

> I don't get this. Since you call pthread_create, libc will handle everything properly on SaBRe side. Of course, you will get a different TID since it's a different thread....