Rob Jansen

I think in `main.c` (or `test_escape_shadow.c` in your 2nd example) you just need to add extern bool shim_swapAllowNativeSyscalls(bool new); and then you don't need the "fake shim" stuff. But then...

> I think ... you just need to add Yeah, actually, I'm wrong. The reason is that we no longer LD_PRELOAD the shim directly, because we don't want to allow...

Maybe we need to build the shim lib with `-fvisibility=hidden`?

>I believe that closing a socket sends the FIN, and while waiting for the FIN-ACK it will continue re-transmitting packets which haven't been acknowledged. This should be the case even...

If you would like support for these common UDP functions, it would help if you could do the following: - explain the expected behavior from the kernel for these functions...

Here is a list of common ones that we don't support yet: - [ ] sendmmsg - [ ] recvmmsg - [x] sendmsg - [x] recvmsg

Attached is a test case using a `linkat` procedure to link a temp file into the filesystem, which should work as described in `man open` but does not work in...

Since this issue was created, we've learned more about the performance cost of switching between Shadow and the managed processes. Switching cost is much less significant than we originally thought...

Is there a reason you are using `sudo`? Shadow does not require root privileges and we strongly advise against using `sudo` because it is an unnecessary security risk and often...

Thanks for reporting! It appears that some valid memory information is being logged before the error occurs. This option is not on by default, and I must admit that we...