| 30 Dec 2024 |
 dirbaio | for example given a 1024b buffer the expected throughput is 1024b/100ms=10240b/s (at zero loss). and it checks out exactly :D | 10:56:29 |
| dirbaio | and then throughput goes down with loss | 10:56:41 |
| dirbaio | I dunno why the lines get more random with the higher buffer sizes. intuitively i'd say more packets means more chances for the randomness to even out so it's the lower lines that shold be more random, not the highest. dunno if this indicates a bug or what | 10:57:22 |
 smoltcp-bot | New PR: Several TCP fixes, improvements. | 11:40:12 |
| smoltcp-bot | PR closed without merging: tcp: set ack_all when remote acks bytes sent in earlier retransmissions. | 11:42:51 |
| smoltcp-bot | New PR: tcp: add netsim test | 11:59:43 |
| 2 Jan 2025 |
| dirbaio | thoughts on this? any concerns about using insta, or any better ideas? | 12:40:30 |
| 3 Jan 2025 |
 Catherine | i'll try to look after 7th (once my vacation ends) | 08:24:15 |
| smoltcp-bot | New PR: Fix RawHardwareAddress panic when parsing to specific link layer address | 10:50:52 |
| smoltcp-bot | PR merged: Fix RawHardwareAddress panic when parsing to specific link layer address | 11:35:24 |
| smoltcp-bot | New PR: Fix DHCPv4 panic when T1 < T2 < lease duration is not respected | 13:43:27 |
 Thibaut Vandervelden | Download NGIR Smoltcp penetration test report 2024 1.0.pdf | 14:26:06 |
| Thibaut Vandervelden | This is the report from Radically Open Security. | 14:26:28 |
| Thibaut Vandervelden | * This is the penetration test report of smoltcp from Radically Open Security. | 14:27:17 |
| Thibaut Vandervelden | They found 2 panics. One which is fixed in https://github.com/smoltcp-rs/smoltcp/pull/1027 and I opened https://github.com/smoltcp-rs/smoltcp/pull/1029 for the DHCP panic. However, I'm not sure if that is the correct approach. | 14:29:04 |
| dirbaio | nice! | 14:54:45 |
| Catherine | only two easily fixed DoS, nice | 14:54:56 |
| Catherine | (... in code that I didn't write 🫢) | 14:55:11 |
| dirbaio | the RFC says nothing of what to do if the server sends us invalid values yeah... reverting to the defaults seems reasonable | 14:55:24 |
| Catherine | writing TCP/IP stacks in a memory safe language is so wonderfully boring | 14:56:11 |
| dirbaio | still lots of room for exciting tcp logic bugs tho :D | 14:57:03 |
| Catherine | for sure | 14:58:58 |
| Catherine | I'm disappointed in how many I missed... | 14:59:11 |
| Catherine | but then I started with the original TCP RFC and the damn thing is like a napkin sketch | 14:59:29 |
| Catherine | In reply to @thvdveld:rubdos.be
sent a file. should we have things like 802.15.4 enabled by default? DoS isn't ideal and this code is tested less and needed less often | 15:00:25 |
| Catherine | (also, do we have DHCP in the fuzzer rotation?) | 15:00:53 |
| dirbaio | nop | 15:03:02 |
| dirbaio | * nope | 15:03:04 |
| Catherine | we should | 15:03:12 |
| Catherine | (am still totally on vacation 🙈) | 15:03:30 |
