Hello Bernhard, thanks for interrest in the FlexCAN emulation for QEMU.
On Sunday 30 of March 2025 12:05:45 Bernhard Beschow wrote: > Am 2. Oktober 2024 15:06:01 UTC schrieb Pavel Pisa <p...@fel.cvut.cz>: > >Dear Nikita and other, > > > >my student Matyas Bobek has chosen to work > >on QEMU CAN support in a frame of his study > >final project. > > > >We have identified iMX6 FlexCAN as the next interesting > >CAN controller for emulation. > > I'd be interested in emulating FlexCAN as well! Is there any repository for > a sneak preview? Yes, Matyas Bobek's repository with work in progress is there https://gitlab.fel.cvut.cz/bobekmat/qemu-flexcan I have tested his branch in January and it worked with QEMU iMX6 and Linux kernel driver. He is continuing the work and we have met last week when he did rebase to something around v10.0.0-rc1 version. I have tested with next config at that time and I have retested int again now after update ~/repo/qemu/qemu-flexcan-build/qemu-system-arm \ -m 1G -M sabrelite -smp 4 \ -kernel kernel-imx6q-sabrelite \ -dtb imx6q-sabrelite.dtb \ -initrd ramdisk.cpio \ -serial mon:stdio \ -nographic \ -object can-bus,id=canbus0-bus \ -machine canbus0=canbus0-bus -machine canbus1=canbus0-bus \ -object can-host-socketcan,if=can0,canbus=canbus0-bus,id=canbus0-socketcan \ There is lot of prints and debug etc... But basic functionality, enough to allow Linux driver to work is in place. > >One of our industrial partners > >is using iMX6 based system with CAN and iMX6 SabreLite > >is supported by QEMU. iMX6 is often used in industry > >still and FlexCAN has use even for iMX8 emulation > >if it is added into QEMU in the future. > > As luck would have it, QEMU will support i.MX 8M Plus in the upcoming > release: > <https://lore.kernel.org/qemu-devel/20250225180510.1318207-32-peter.maydell >@linaro.org/> I'm more than happy to integrate FlexCAN emulation! Great. At this moment the thesis target has been only FlexCAN for iMX6 (that is without FD) but we have discussed that code should be designed with extension to more FlexCAN controller in the mind. Please, discuss with colleague what he sees as realistic at this time. He has to submit thesis text in May... But I would be happy if project continues at CTU or elsewhere. It would be fair to find at least some basic funding if the student continues on the project over holidays. I have received no money for all my QEMU work for years nor I expect that I find funding for that at CTU or Skoda Auto or others using our services. I can find some small money from my reserve from other projects. DO you have some option for funding on your side? Or the option is GSOC, QEMU is on the list as I see. I am already mentor at NuttX and take such role for RTEMS probably as well, I have lot of other duties but I can accept co-mentor, consultant role for QEMU GSoC if there is interrest. As for QEMU core CAN subsystem and CAN FD, it is prepared and work well for CTU CAN FD emulation and even AMD Versal which builds on base of our work.. > >In the preparatory work, we want to setup > >shared directory between host and guest > >kernel using 9P FS on PCI virtio. > > > >The development environment is based on current > >mainline QEMU and mainline Linux kernel sources. > > > >We use standard QEMU options > > > > -virtfs local,path=shareddir,security_model=none,mount_tag=shareddir > > > >and the 1af4:1009 is seen by monitor command "info pci" > >but on the guest kernel side only "16c3:abcd" DWC_usb3 / PCIe bridge > >is seen but none from the mapped devices, virtio (1af4:1009), > >CTU CAN FD, Kvaser CAN. > > Should be fixed in master / QEMU 10.0 by > <https://lore.kernel.org/qemu-devel/20250225180510.1318207-30-peter.maydell >@linaro.org/> When I try QEMU with -virtfs local,path=shareddir,security_model=none,mount_tag=shareddir on iMX6 sabrelite now, I can see the device on PCI/PCIe lspci 00:00.0 Class 0604: 16c3:abcd 01:00.0 Class 0002: 1af4:1049 I can see interrupts cat /proc/interrupts 290: 0 0 0 0 PCI-MSI 524288 Edge virtio0-config 291: 0 0 0 0 PCI-MSI 524289 Edge virtio0-requests but when I try mount -t 9p -o trans=virtio shareddir /mnt/shareddir it stucks and no interrupts seems to arrive cat /proc/interrupts 290: 0 0 0 0 PCI-MSI 524288 Edge virtio0-config 291: 0 0 0 0 PCI-MSI 524289 Edge virtio0-requests I have tried even CTU CAN FD PCIe emulation. I can see the device on the lspci, the first send frame passes but then it stucks. Receive does not see any mesaage. I expect that at least on sabrelite there is something missing. May it be in kernel devicetree or on QEMU side in PCI/PCIe interrupts. But that is not so critical at this moment. By the way, I would like to discuss QEMU interrupts for generic FPGA mapping of our CTU CAN FD IP core (in addition to PCIe) https://github.com/ppisa/qemu/blob/net-can-ctucanfd-platform/hw/net/can/ctucan_mm.c#L144 where I have looked for help in past. It would be very usesfull for us to have memory-mapped CTU CAN FD IP core emulation with commad line attaching to Zynq and PolarFire... > >The imx6q-sabrelite DTS is used with standard device-tree > > > >https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/ar > >ch/arm/boot/dts/nxp/imx/imx6q-sabrelite.dts > > > >We hope that all required drivers are enabled > >in the kernel. But it is possible that something > >has been overlooked. > > > >Do you have some hint or some kernel > >and QEMU working example for iMX6 PCIe > >device mapping? > > When adding a PCI device via command line, add `bus=dw-pcie` as an > additional option as a workaround for earlier QEMUs. Best wishes, Pavel Pavel Pisa phone: +420 603531357 e-mail: p...@cmp.felk.cvut.cz Department of Control Engineering FEE CVUT Karlovo namesti 13, 121 35, Prague 2 university: http://control.fel.cvut.cz/ personal: http://cmp.felk.cvut.cz/~pisa social: https://social.kernel.org/ppisa projects: https://www.openhub.net/accounts/ppisa CAN related:http://canbus.pages.fel.cvut.cz/ RISC-V education: https://comparch.edu.cvut.cz/ Open Technologies Research Education and Exchange Services https://gitlab.fel.cvut.cz/otrees/org/-/wikis/home
