Re: [seL4] Problems with tk1 nonsecure

Hi Mike, I'm not sure what changes in the last 6 months would have changed your build from a working to broken state. The below commits[1] are the only changes that I am aware would change the behavior of smmu permissions. ID: 98? corresponds to sdmmc3 (there is a table in the tk1 reference manual, around page 719 [2]), and this corresponds to a vm.smmu id of 22, which is absent from the TK1INSECURE configuration. Adding that ID made my VM crash---I tracked the cause down to a iommu page table initialisation bug here: https://github.com/SEL4PROJ/seL4_projects_libs/commit/396542fc8b7d0ab41bf92d... Fixing that allowed my vm to boot with the extra vm.smmu id of 22. (Note: I boot over tftpboot and not using an sdcard, so it also boots without id 22 for me, however the ID: 98 error you are getting very likely corresponds to id 22 being missing). The latest checkout that I tested is this one: https://github.com/SEL4PROJ/camkes-arm-vm-manifest/commit/8f83660cdd98c51bda... Also note that yesterday we merged some refactors that migrated the project to use a new CMake based build system. running the following: mkdir build && cd build && ../init-build.sh -DAARCH32=TRUE -DTk1Insecure=True -DCAMKES_VM_APP=tk1_vm && ninja should build you an equivalent image to what I tested. Kent. PS: That bug I mentioned was causing corrupted vspace shadow state and could have potentially been the cause of that hypercall issue you were experiencing about a year ago. (Sorry I dropped the email thread about trying to solve it back then :-( ) [1] https://github.com/seL4/seL4/commit/80d32b160989c55f9fdc33fce1bf605ac1b42f1e https://github.com/SEL4PROJ/camkes-arm-vm/commit/f3948cf41978bee33179ac7a814... [2] https://developer.nvidia.com/embedded/tegra-k1-reference ________________________________ From: Devel on behalf of Mike Clark Sent: Wednesday, June 20, 2018 10:35 PM To: Danis, Adrian (Data61, Kensington NSW) Cc: devel@sel4.systems Subject: Re: [seL4] Problems with tk1 nonsecure Anyone else running into this issue with the latest clone? I had this all working just fine previously (maybe 6 months ago), but something must have been updated that breaks this. Not sure why. Mike On Tue, Jun 12, 2018 at 8:45 AM, Mike Clark mailto:undefinedspace@gmail.com> wrote: Adrian, I am doing `make menuconfig`, then going to "Applications" and checking the box that says "Insecure: Give all devices to VM component". That is the only change I do, then recompile, copy the new image to my SD card, and boot. Mike On Mon, Jun 11, 2018 at 9:20 PM, mailto:Adrian.Danis@data61.csiro.au> wrote: Hi Mike, In order to be acting as a hypervisor in the default configuration that you say booted fine seL4 will already be setup to run in nonsecure hypervisor mode. Given this can you elaborate on what 'the unsecure mode' is and how you are switching to it? Adrian On Sun 10-Jun-2018 10:08 PM, Mike Clark wrote:

I just checked out a fresh copy of https://github.com/ SEL4PROJ/camkes-arm-vm-manifest and built an image for the TK1 using the docker build environment. I copied the resulting image to an SD card and it boots fine.

If I switch to the unsecure mode, I get an error on boot.

SMMU Address translation error: ID: 98 address: 0xae5be000 type: 6 direction: 0x0 IOPT permission: read 0x0 write 0x0 nonsecure 0x0 [ 10.204777] mmc1: ADMA error

I have not made any changes, so my vm.smmu settings are all default.

Any thoughts or help?

Mike

_______________________________________________ Devel mailing list Devel@sel4.systems https://sel4.systems/lists/listinfo/devel