On Sat, Jan 24, 2009 at 10:45:11AM -0600, Dave Jones wrote: > David L. Craig wrote: > > [snip.....] > > Sigh... We're running VM/ESA 2.2 which means CP doesn't > > know about the IEEE floating point hardware. As I see it, > > we can upgrade VM, tell the kernel to use emulation even > > though the hardware is there, or put Debian (and VM) into > > their own LPARs. Does crypto make heavy use of floating > > point? > No, it does not. It makes (heavy) use of integer arithmetic, > though. Note that a crypto processor is available for the > 7060, to offload this work from the CPU. > > > Is there a way to virtually network between LPARs? > > Not on a 7060, as far as I know. It does not support > hipersockets, so any network connections between LPARS will, > I think, require some sort of route through the OS/2 side or > some other external connection. > > > Does the kernel parameter support using emulation even > > when the hardware is available? If you can answer off the > > top of your head, let me know. Otherwise, I'll figure > > this out next week. > > If the Linux kernel detects the crypto hardware, it will use > it; if not, it uses emulation (all that integer arithmetic > mentioned above). > > If you need an encryption solution for the CMS environment, > drop me a note off list.
I looked at the code (don't you just love free software?) and it looks like I can get away with zapping off the AFP bit in the first set of control registers loaded by the kernel in head31.S shortly after bootstrapping completes; i.e., .Lctl: .long 0x04b50002 # cr0: various things becomes .Lctl: .long 0x00b50002 # cr0: various things When the same module tests for the IEEE feature by attempting an EFPC, it merely sets a flag in machine_Flags if no program check is incurred. The Program New PSW has been modified to merely point at the instruction following the OI, so the AFP bit is left alone (it ought to check that CP IEEE support is also functional if no program check occurs and it is in a virtual machine and, if so, skip the machine_flags change and reset the AFP bit in CR0, but since I'm probably the only person in the world that would care, I'll just zap the bit). Unless, of course, CP will panic when it sees the funny program check caused by attempting EFPC with the real AFP bit off. I guess I'd better check out if that's the case... -- May the LORD God bless you exceedingly abundantly! Dave Craig - - - - - - - - - - - - - - - - - - - - "'So the universe is not quite as you thought it was. You'd better rearrange your beliefs, then. Because you certainly can't rearrange the universe.'" --from _Nightfall_ by Asimov/Silverberg ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
