Re: [Bug-xorriso] building libisofs-1.4.4, libburn-1.4.4, libisoburn-1.4.4 on OpenBSD

[Thomas Schmitt]

Hi,

thanks for the many tests. I planned to get to them in detail today
but it is already getting late night here ...

SASANO Takayoshi wrote:
> |-- linux
> |   |-- ahci-linux-cdrw.log           xorriso/Linux/AHCI: works good
> |   |-- ahci-linux-wodim-vv.log       wodim/Linux/AHCI: works good
> |   |-- pciide-linux-cdrw.log         xorriso/Linux/PCI-IDE: works good
> |   `-- pciide-linux-wodim-vv.log     wodim/Linux/PCI-IDE: works good
> `-- openbsd
>     |-- ahci-cdrecord-vv-2.log        cdrecord/OpenBSD/AHCI: not working
>     |-- ahci-cdrw-2.log               xorriso/OpenBSD/AHCI: not working
>     |-- pciide-cdrecord-vv-2.log      cdrecord/OpenBSD/PCI-IDE: works good
>     `-- pciide-cdrw-2.log             xorriso/OpenBSD/PCI-IDE: not working

Looks like we have to hunt for a Sense Data defacing piece of code
in the OpenBSD kernel.

Do you know anybody who could give hints where to look ?

Are you willing to insert print statements into suspicious code and
run the modified kernel ?

Normally i would now begin to prepare a qemu VM. But it is tricky to
replay real CD hardware situations with kernels on virtual machines.
E.g. i am able to operate a real CD drive from a virtual Linux via
virtio, but the kernel does not treat the virtio device like it
would operate a real CD device. ioctl(SG_IO) does not get routed
through sr.c.
Do you have a proposal how i could lend a DVD burner of Debian 8 amd64
to a virtual OpenBSD ? (Best on plain qemu-i386 or qemu-x86_64, else on
a Debian provided VM system.)


I will look into the logs tomorrow. Maybe i find out why xorriso failed
with CD-RW where cdrecord succeeded and how exactly both fail with AHCI.


Have a nice day :)

Thomas