Summary: 63 instances, 56 unique

Text	Count
// FIXME: we're losing the bool saying whether metrics were actually written.	1
// TODO: Maybe at some point include this in the checks we do when populating the MMDS via the API,	1
// TODO: generate bindings for these from the kernel sources.	1
// TODO: get rid of this when splitting dumbo into public and internal parts.	1
// TODO: auto-generate kvm related structures with PartialEq on.	1
// TODO: To be super strict, we should make sure there aren't additional MSS	1
// TODO: The following logic fully makes sense only for a passive open (which is what we	1
// TODO: notify backend driver to stop the device	1
// TODO: rename this field to adopt inclusive language once Linux updates it, too.	4
// TODO: This is currently IPv4 specific. Maybe change it to a more generic implementation.	1
// TODO: mention this in doc comment for function	1
// TODO: The description of this method is also a TODO in disguise.	1
// TODO: this doesn't seem very space-efficient.	1
// TODO: like receive_segment(), this function is specific in some ways to Connections	1
// TODO: Are we sure we want to finish running a vcpu upon	1
// TODO: replace with validation based on `has_cpuid()` when it becomes stable:	1
// TODO: any better way to set all these bytes to 0? Or is this optimized by the compiler?	1
// TODO continued: the guest soft lockup watchdog does not panic on Resume.	1
// TODO: Fix the duration below to match the required time to collect enough samples.	1
/// TODO: use `GuestMemory::write_to()` and `GuestMemory::read_from()` when they are stable	1
// TODO: We skip verifying the checksum, just in case the device model relies on offloading	2
// TODO: Maybe get this nicely from packet at some point.	1
// TODO: Are we sure we want to finish running a vcpu upon	1
// TODO: just now wondering ... is doing a push()/pop() thing better than just creating	1
//TODO crosvm uses 0 here, but IIRC virtio specified some other vendor id that should be used	1
// TODO: This is another strange (and potentially dangerous) situation, because	1
// TODO working with IPv4 only for now	1
// TODO: would a stronger ordering make a difference here?	1
// TODO: Migrate the VMM public-facing code (i.e. interface) to use stateless structures,	1
// TODO: The implementation of TcpSegment is IPv4 specific in regard to checksum computation. Maybe	1
// TODO: Maximum efficiency.	1
// TODO: The following test code is way more verbose than it should've been. Make it	1
// TODO: I guess this should be refactored at some point to also remove the endpoint if found.	1
// TODO: Check that header_len | 0b11 == 0 and the resulting data_offset is valid?	1
// TODO: Add negative test for this warning failure.	1
// TODO log address	1
// TODO: These are currently expressed in cycles. Normally, they would be the equivalent of a	1
// TODO: return the actual advance value here	1
// TODO: If/when we support active opens, we'll also have to accept RST/SYN segments,	1
// TODO: hard link instead of copy? This would save up disk space, but hard linking is	1
// TODO: I'll be honest: the tests here cover the situations most likely to be encountered, but are	1
// TODO: If we ever implement window scaling, change the part that computes	1
// TODO: Maybe we should enqueue multiple ACKs here (after making such a thing	1
// TODO: also check for errors. Pending high level discussions on how we want	3
// TODO: This is a strange situation, because the other endpoint is sending data	1
// TODO: Using first_not_sent() makes sense here because a connection is currently	1
// TODO: rename this constant to adopt inclusive language once Linux updates it, too.	2
// TODO: we should call `KVM_KVMCLOCK_CTRL` here to make sure	1
// TODO: support 802.1Q tags?! If so, don't forget to change the speculative_test_* functions	1
// TODO: Please note that, right now, the buckets become full after being updated.	1
// TODO: using something like std::mem::uninitialized could avoid the extra initialization,	1
// TODO: Detect / handle queue deadlock:	1
// TODO: Is u32 enough to prevent overflow for the code in this function? I think so, but it	1
// TODO: Is it an overreaction to reset here?	1
// TODO: We need a better name than app_metrics (something that says that these are the actual	1
// FIXME: use macro	1