Dmitrii Kuvaiskii
Dmitrii Kuvaiskii
@DL8 @lejunzhu @kailun-qin Does the above implementation make sense to you?
Ok, looks like there is strong opposition to allowing older CPUSVNs. Then I will not pursue that approach. I think there is nothing else we can do but go with...
> > Otherwise, as some random thoughts, will an offline migration tool (or a migration mode of Gramine, that only conduct the file migration for all encrypted files from a...
Someone (maybe me) needs to create a reproducer and show perf numbers (+ show perf numbers on a real workload like Nginx). This is low priority for now though.
There are informative discussions in the now-closed PR #1026, please see there. For now this is very low priority, so I'm assigning P3 to it.
@lejunzhu `nfds` can be arbitrarily large. What we could do is to have logic like this: ``` if (nfds < SOME_MAX_LIMIT) { // new logic PAL_HANDLE* pals = alloca(nfds *...
But also, if @lejunzhu wants to implement this, could you just re-structure that whole `_libos_syscall_poll()` function? It looks pretty ugly and complicated, it can surely be refactored without changing the...
@mkow I'd be happy but isn't Borys super-tied already with other tasks?
> So, I think even using 1K stack space for each malloc will help. Yes, this will be good! From my experience, typically non-network-bound applications use about 5-10-20 FDs in...
I'm closing this issue now as it became a duplicate of #853