On 1.8.2012 1:02, Martin Decky wrote: > I don't know how Jakub progresses with #461 and I don't want to minimize > this issue (it affects not only Jiri's machine, but a large class of > machines). But, on the other hand, a proper solution would be to > reimplement all the interrupt controller drivers as first-class DDF > citizens supporting all the non-trivial configurations there is, Thus, > postponing #461 is not fatal.
Still, it would be interesting to find out whether exporting the IRQ2PIN map to the uspace driver will fix this problem. A relatively easy first step is looking at whether at least the SMP kernel on Jiri's machine can normally receive interrupts (for example by booting a kernel-only image and trying to issue some kconsole commands) or whether the problem is somewhere else. > Regarding #452: If I remember correctly, Jiri said during our last > meeting that he was not entirely sure whether register information in > amd64 cores had been implemented yet. Again, postponing #452 to the > 0.5.1 development cycle won't kill us. Looks like it has been implemented by virtue of implementing amd64 version of istate_to_elf_regs(), but probably not tested well enough. Or something must have silently changed in the test environment which breaks this feature for us now. According to the comments about how the Linux kernel screws up the alignment on 64-bits, one would think that this piece of code had worked at least once. Jakub _______________________________________________ HelenOS-devel mailing list [email protected] http://lists.modry.cz/cgi-bin/listinfo/helenos-devel
