h1. PeriPeri-2018-02 http://muistio.tieke.fi/p/periperi-20180202 | Dates | 2018/02/01 - 2018/02/02 | | Place | see 'Activities' below | |/12. Member | Geert | 02/02 - 04/02 (night @Home) | | Jacopo | 28/01 - 05/02 | | Kieran | 28/01 - 05/02 | | Laurent | 28/01 - 05/02 | | -Magnus- | | Marek | | -Morimoto- | | -Ito- | | Niklas | 28/01 - 05/02 | | Simon | 02/02 - 04/02 | | Ulrich | 01/02 - 05/02 | | Wolfram | 01/02 - 05/02 | h2. Activities in Brussels | Date | Time | Activity | Location | | Monday 2018-01-29 | 09:00 - 18:00 | GMSL Code Camp | Antoine Dansaert 24, 1000 Brussels | | Tuesday 2018-01-30 | 09:00 - 18:00 | GMSL Code Camp | Antoine Dansaert 24, 1000 Brussels | | | 19:00 - 21:00 | San Sablon | | | Wednesday 2018-01-31 | 09:00 - 13:00 | GMSL Code Camp | Antoine Dansaert 24, 1000 Brussels | | Thursday 2018-02-01 | 14:30 - 18:00 | Multimedia Meeting | Antoine Dansaert 24, 1000 Brussels | | Thursday 2018-02-01 | 19:00 | Dinner | https://www.bonsoirclara.com/en/ | | Friday 2018-02-02 | 09:00 - 19:00 | Core, I/O Group, Virtualization BOF? | http://www.hotelbloom.com/ | | Friday 2018-02-02 | 19:30 | Dinner | http://www.kokob.be/ | | Saturday 2018-02-03 | 09:30 - 19:00 | FOSDEM | https://fosdem.org/2018/ | | Sunday 2018-02-04 | 09:00 - 18:00 | FOSDEM | https://fosdem.org/2018/ | h2. List of material for Multimedia code camp |Who | What | |/4. Jacopo | Scope (with probes) | | Multimeter | | Network switch | | Power strip | |/3. Kieran | Salvator-X H3 + Expansion board | | 8x RDACM20 | | 8x RDACM21 | |/3. Laurent | Kingfisher (including H3SK and MAX9286 add-on boards) | | Multimeter | | Power strip | |/3. Niklas | V3M + Expansion board | | 8x RDACM20 | | 8x RDACM21 | h2. Agenda for Friday Meeting | 09:00 - 09:30 | Welcome Coffee | |/2. 09:30 - 10:30 | I/O Group Meeting (incl. future tasks) | | - i2c mux | | 10:30 - 11:00 | Coffee Break | |/3. 11:00 - 12:30 | Core Group Meeting (incl. future tasks) (60') | | - R-Car Gen2 Watchdog Timer | | Improving renesas-drivers (30') | | 12:30 - 13:30 | Lunch | |/4. 13:30 - 15:30 | Virtualization BoF (120') | | - Status reports | | - Issues | | - Next steps | | 15:30 - 16:00 | Coffee Break | |/2. 16:00 - 19:00 | Board Farm, Test automation, Remote access (60') | | Hacking (Eagle, Draak, ...)? (120') | h2. Meeting minutes (thanks Kieran!) h1. I/O Group meeting h2. RAVB : 2K Header limits * Niklas has an idea to increase this limit. * Difference between transmitting and receiving. h2. SPECTRE * Wolfram raised that he is Interested in SPECTRE * 3 Variants of SPECTRE, v2 is the worst - and it's unknown if we can fix it. * Needs updated kernel (not available for ARM yet) and userspace rebuild. * Userspace rebuilds (retpoline) may have a performance penalty. * RMK has reportedly got PoC's which are even more effective at utilising the vulnerabilities. * Meltdown (PTI) should be an easier fix. h2. Handling Security Issues in a broad context: * Laurent raised that a bug in linux-media to read and write kernel memory. h2. SDHI * Rasied by Simon in absence * Next steps for HS400 * Driver and H3/M3-W enablement patches are pending review * Second upport attempt seemed much easier than first as BSP seems closer to upstream now * Would like to review other SDHI patches in BSP v3.6.0 with Wolfram, possibly on IRC some time in February ** Yes. WSA, SH to review. h2. I2C demux * Raised by Simon in absence * Understands that consensus from San Sebastian meeting was to accept changes despite OOpses in some subsystems when switching cores. Not entirely happy but willing to accept that consensus. * Would appreciate a rebast/repost of the dts patches ** Wolfram will repost patches. Ack, thanks! * Any other issues h2. MAX9286 GPIO Controller Regulator Loop * Circular dependency on GPIO regulator for cameras on Eagle-V3M * MFD is a lot of over head - but it the only 'correct' solution. h2. IO Group: General Status * Short meeting style is good for all. (and taken on by other groups) * Wolfram reprhases summary to make sure he understands it in his own words. * Wolfram likes Quarterly scheduling of tasks. * IO tasks agreed by mid-january ** raised externally that this agreement is at least 1 month late however ** high confidence by end of December h2. I2C Mux * Can we leave mux channels open to improve performance (reduce 'deselect' calls) * We can have multiple devices with the same 'default' address, and there must be a negotiation stage to re-configure each address on the child bus ** Currently handled by the device driver * Should core be more involved in this? * WSA to think about mux topic :-D h1. Core Group Meeting * Status updates round the table * We're expecting a new Salvator-XS M3-N board potentially sometime in February ? * Stout board (H2) getting improved support ... * Can we add elinux-wiki links to cover letters when we send relevant patch sets for additional tasks. h2. RCar Gen2 Watchdog Timer * We have a patchset (with 25 patches) from Fabrizio Casteu, and we need to test them. * Geert to try on Goelsch * WSA to try on Lager * Marek to try on Stout. h2. Future Tasks * With the upcoming complex Virtualised use cases, adding support for JTAG debuggers would assist in debugging complex issues of the virtualised environments at the core level. * Function enablement on boards as we get them. * New platforms share lots of support, and with the shared dtsi infrastructure board bringup should be rapid. * Expand IPMMU IO Vspace to more than 32 bits (neg) * Blocked on work from Magnus h3. Renesas Drivers * Improving Integration for multimedia * Magnus would like a central place for latest multimedia work. * it easier to use especially from a developer's point of view h3. Goals * "Renesas-next" but with WIP * Testing -> Continuous testing * Shouldn't be a development base * Doing development on renesas-drivers, means you have to rebase on main before submission. * Consumption by the BSP ** Releasing a custom kernel -> But we are not a distribution with the resources to test each release. * Can we have a 'latest topic' branch to continue automatic integration into renesas-drivers * standardise the feature branches and tag naming scheme * Dependency handling * mainline * subsystem.next * topic branch * If you have complex / stale dependencies - then it's fine to leave them out of Renesas-Drivers h3. Topic naming scheme bikeshedding: * topic/name-version * subsystem/next/drivers/feature/ * subsystem/next/core/feature/ * topic/subsystems/?/feature/ * subsystem/topic/? h3. Preference: * Still to be considered ... h3. Update for magnus: * Standardize on a naming scheme, to define the topic prefix. From there Geert will integrate the latest version based on that prefix. h1. Virtualisation h2. QEMU running on an arm64 guest * Plain QEMU if host is amd64 or arm64 * KVM if host is arm64 (R-CAR H3 or M3-W) * KVM needs ARM trusted firmware with HYP support ** https://elinux.org/R-Car/Virtualization h2. Libvirt * virsh (on amd64, or arm64) * virt-manager (on amd64) ** https://elinux.org/R-Car/Virtualization/Libvirt * Provides an abstracted interface common description for multiple virtualisation options. h3. Virtualisation Status * Walk trhough of GPIO pass through from Geert * Configuration of host to enable GPIO passthrough h3. Guest side (QEMU) * vfio-platform support not yet supported but patches were posted h2. Next Steps * Reset driver, using CPG/MSSR * Interrupts * IOMMU groups h1. Board Farms * BayLibre ACME is sold out. Can we make or obtain alternatives? * Prototyping with a Teensy 3.2 (freescale cortex-M4) * Kit: (Search aliexpress for 'travel router usb')