<GitHub89>
[smoltcp] whitequark commented on pull request #190 91fdc1f: This isn't enough. You need to at least match the ident field with the source IP, because otherwise you're introducing a vulnerability: anyone who knows that there is a flow of fragmented packets has an 1/65536 chance of arbitrarily altering their contents. This is a 100% viable attack vector and very similar attacks (utilizing the port number, but the c
<GitHub91>
[smoltcp] whitequark commented on pull request #190 91fdc1f: Why is it generic over `T`? We'll never want to defragment anything except `u8`. This just results in silly `T: Clone` bounds on functions. https://github.com/m-labs/smoltcp/pull/190#discussion_r180947397
<GitHub84>
[smoltcp] whitequark commented on pull request #190 91fdc1f: This doesn't belong in this pull request. Please submit it separately and we'll have a discussion about it; in particular, you're counting 802.1Q, but not QinQ or QinQinQ.... https://github.com/m-labs/smoltcp/pull/190#discussion_r180948714
<GitHub-m-labs>
[artiq] sbourdeauducq commented on issue #854: That media converter causes some obscure bug with the PHY that results in a few percents of packets being lost. The transceiver that works correctly is a HP 453578-001 (@jbqubit I sent you the picture last week).... https://github.com/m-labs/artiq/issues/854#issuecomment-380760416
<GitHub-m-labs>
[artiq] sbourdeauducq commented on issue #854: That media converter causes some obscure bug with the PHY that results in a few percents of the packets being lost. The transceiver that works correctly is a HP 453578-001 (@jbqubit I sent you the picture last week).... https://github.com/m-labs/artiq/issues/854#issuecomment-380760416
rohitksingh_wor1 has joined #m-labs
rohitksingh_work has quit [Ping timeout: 240 seconds]
rohitksingh_wor1 has quit [Ping timeout: 240 seconds]
<GitHub-m-labs>
[artiq] marmeladapk commented on issue #854: @sbourdeauducq I wanted to run your ethernet yak-shaving tools. I synthesied both (renamed eth_rgmii to eth in sayma_transmitter.py), loaded them using vivado and connected sayma <-> sata <-> sfp <-> ethernet <->kc705 rj45 on fp.... https://github.com/m-labs/artiq/issues/854#issuecomment-380784251
rohitksingh_work has quit [Read error: Connection reset by peer]
greni has quit [Quit: Page closed]
larsc has joined #m-labs
rohitksingh has joined #m-labs
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined #m-labs
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined #m-labs
hartytp has joined #m-labs
<hartytp>
is there anything special I need to do to get access to a CSR from a kernel?
<hartytp>
I'm currently getting "[ 10.173307s] ERROR(runtime::session): session aborted: invalid kernel CPU pointer 0x0"
<hartytp>
(doing what I discussed with whitequark last night)
<hartytp>
I guess I use register_kernel_cpu_csrdevice instead of csr_devices.append()
<hartytp>
and add an entry to mem_map
rohitksingh has quit [Quit: Leaving.]
hartytp has quit [Quit: Page closed]
<GitHub-m-labs>
[artiq] marmeladapk commented on issue #854: @sbourdeauducq I wanted to run your ethernet yak-shaving tools. I synthesied both (renamed eth_rgmii to eth in sayma_transmitter.py), loaded them using vivado and connected sayma <-> sata <-> sfp <-> ethernet <->kc705 rj45 on fp.... https://github.com/m-labs/artiq/issues/854#issuecomment-380784251
Ultrasauce_ has quit [Remote host closed the connection]
Ultrasauce has joined #m-labs
rohitksingh has joined #m-labs
rohitksingh has quit [Quit: Leaving.]
lars__ has joined #m-labs
<GitHub185>
[smoltcp] podhrmic commented on pull request #190 91fdc1f: Done. On a side note, fragmentation is currently a part of the default features. If I remove it, it makes building the server_fragmented example really verbose (you have to enumerate all features). Is there some better way how to do that? https://github.com/m-labs/smoltcp/pull/190#discussion_r181172981
<GitHub22>
[smoltcp] podhrmic commented on pull request #190 91fdc1f: Yes, this one was actually quite tricky. I am happy to explain in more detail, but in short, the problem was that to access fragments from `self` I needed a mutable borrow of self, but the following processing (such as `process_udp`) uses a non-mutable borrow. And because the reassembled packet is points to `mut self`'s buffer it didn't pass the borrow che
<GitHub-m-labs>
picam/master 991aebb Robert Jordens: clean up test.py
<GitHub-m-labs>
picam/master 02f7ce6 Robert Jordens: remove interceptor binary
<GitHub-m-labs>
picam/master 69d2cd8 Robert Jordens: gcc -O2 interceptor
<GitHub167>
[smoltcp] podhrmic commented on pull request #190 91fdc1f: But I need to include `FragmentSet` because interface now contains `fragments: Option<FragmentSet<'e>>`, except without enabled fragmentation it compiles away as it is always None https://github.com/m-labs/smoltcp/pull/190#discussion_r181178474
<GitHub68>
[smoltcp] podhrmic commented on pull request #190 91fdc1f: This is needed to get the fragmentation to work. In Linux MTU refers to the max length of the ethernet payload. But smoltcp counts MTU as max length of the ethernet frame. Hence without this added ethernet length, all fragmented packets are truncated. I am happy to fix is more meaningfully later, but without this fix the fragmentation reassembly doesn't wo
<GitHub-m-labs>
[artiq] jbqubit commented on issue #854: > By default they work in 1000Base-X mode and usually work, the only negotiation they may need is Pause. This is one of the things that need to be implemented in MMC - the PHY has several associated register that need to be updated.... https://github.com/m-labs/artiq/issues/854#issuecomment-380946319
hartytp has joined #m-labs
<hartytp>
okay, added CSR using self.submodules.my_csr_module = MyCSRModule(platform)
<GitHub-m-labs>
[artiq] hartytp commented on issue #981: @gkasprow I have shorted pins 1 and 4 of the programming connector and connected pin 3 of T5 to the AMC 3V3 supply via a 1k resistor (using the 3V3 PG LED).... https://github.com/m-labs/artiq/issues/981#issuecomment-380961921
<GitHub-m-labs>
[artiq] hartytp commented on issue #981: @gkasprow I have shorted pins 1 and 4 of the programming connector and connected pin 3 of T5 to the AMC 3V3 supply via a 1k resistor (using the 3V3 PG LED).... https://github.com/m-labs/artiq/issues/981#issuecomment-380961921