Chris,
On Fri, Jun 15, 2018 at 12:56:05PM -0400, Chris Rothrock wrote:
- In a previous email in this chain you mentioned trying something more
simplistic, such as run/log. Here are the results of this with and without the SABLE bootloader - this just hangs on handing off from NOVA to Genode, not even any errors. The SABLE code will be going open source soon and I will be able to provide this to you as needed in the effort to trace down the problem.
Honestly, with this little information I see no chance to help. I tend to blame SABLE for breaking boot compatibility because (like you write) it works without SABLE. I suggest you investigate if SABLE is able to boot multiboot OSes in a standard compliant way. Have a look at hypervisor and image.elf and ask the original SABLE developers if they identify something SABLE may not support / never was designed to support (e.g., multiple loaded segments). I'm not able to identify anything Genode core may do wrong if even the first log message is missing. Maybe you could add more diagnostic messages?
- The only working binary I have from the 17.05 that I can get SABLE to
boot properly is Genode with seL4. I am trying to be able to reproduce this build but when I build and run Genode 17.05 using seL4 kernel, it doesn't boot into the demo. The nitpicker and/or framebuffer seems to not load properly when it does on the binary I use for testing SABLE. Below is the output from that 17.05 (without using SABLE). Any ideas why this build is no longer working?
[...]
Warning: void Genode::Rpc_cap_factory::free(Genode::Native_capability) not implemented - resources leaked: 0x20 Warning: PD (init -> acpi_drv) cap limit (used=17, limit=19) exceeded during transfer_quota(3) [init -> acpi_drv] resource_request: cap_quota=3 [init -> nitpicker_config] Warning: top-level node <xray> missing in input ROM xray [init -> nitpicker_config] Warning: could not obtain input value for input xray_enabled [init -> acpi_drv] used before freeing emergency=864K [init -> acpi_drv] used after freeing emergency=848K [init] child "acpi_drv" requests resources: cap_quota=3
I doubt that this ever worked as the logged CAP quota shortage is a fatal error.
Sincerely