cvs commit: src/share/man/man4 Makefile cpuctl.4 src/sys/amd64/amd64 support.S src/sys/amd64/conf NOTES src/sys/amd64/include cpufunc.h specialreg.h src/sys/conf files.amd64 files.i386 src/sys/dev/cpuctl cpuctl.c ...

[Peter Wemm]

On Fri, Aug 8, 2008 at 11:59 AM, John Baldwin <jhb at freebsd.org> wrote:
> On Friday 08 August 2008 02:51:33 pm Kostik Belousov wrote:
>> On Fri, Aug 08, 2008 at 02:27:42PM -0400, John Baldwin wrote:
>> > On Friday 08 August 2008 02:10:52 pm Kostik Belousov wrote:
>> > > On Fri, Aug 08, 2008 at 12:51:17PM -0400, John Baldwin wrote:
>> > > > On Friday 08 August 2008 12:26:53 pm Stanislav Sedov wrote:
>> > > > > stas        2008-08-08 16:26:53 UTC
>> > > > >
>> > > > >   FreeBSD src repository
>> > > > >
>> > > > >   Modified files:
>> > > > >     share/man/man4       Makefile
>> > > > >     sys/amd64/amd64      support.S
>> > > > >     sys/amd64/conf       NOTES
>> > > > >     sys/amd64/include    cpufunc.h specialreg.h
>> > > > >     sys/conf             files.amd64 files.i386
>> > > > >     sys/i386/conf        NOTES
>> > > > >     sys/i386/i386        support.s
>> > > > >     sys/i386/include     cpufunc.h specialreg.h
>> > > > >     sys/modules          Makefile
>> > > > >     sys/sys              priv.h
>> > > > >     usr.sbin             Makefile
>> > > > >   Added files:
>> > > > >     share/man/man4       cpuctl.4
>> > > > >     sys/dev/cpuctl       cpuctl.c
>> > > > >     sys/modules/cpuctl   Makefile
>> > > > >     sys/sys              cpuctl.h
>> > > > >     usr.sbin/cpucontrol  Makefile amd.c amd.h cpucontrol.8
>> > > > >                          cpucontrol.c cpucontrol.h intel.c intel.h
>> > > > >   Log:
>> > > > >   SVN rev 181430 on 2008-08-08 16:26:53Z by stas
>> > > > >
>> > > > >   - Add cpuctl(4) pseudo-device driver to provide access to some
>> > low-level
>> > > > >     features of CPUs like reading/writing machine-specific
> registers,
>> > > > >     retrieving cpuid data, and updating microcode.
>> > > > >   - Add cpucontrol(8) utility, that provides userland access to
>> > > > >     the features of cpuctl(4).
>> > > > >   - Add subsequent manpages.
>> > > > >
>> > > > >   The cpuctl(4) device operates as follows. The pseudo-device node
>> > cpuctlX
>> > > > >   is created for each cpu present in the systems. The pseudo-device
>> > minor
>> > > > >   number corresponds to the cpu number in the system. The cpuctl(4)
>> > pseudo-
>> > > > >   device allows a number of ioctl to be preformed, namely
>> > RDMSR/WRMSR/CPUID
>> > > > >   and UPDATE. The first pair alows the caller to read/write
>> > machine-specific
>> > > > >   registers from the correspondent CPU. cpuid data could be
> retrieved
>> > using
>> > > > >   the CPUID call, and microcode updates are applied via UPDATE.
>> > > > >
>> > > > >   The permissions are inforced based on the pseudo-device file
>> > permissions.
>> > > > >   RDMSR/CPUID will be allowed when the caller has read access to the
>> > device
>> > > > >   node, while WRMSR/UPDATE will be granted only when the node is
> opened
>> > > > >   for writing. There're also a number of priv(9) checks.
>> > > > >
>> > > > >   The cpucontrol(8) utility is intened to provide userland access to
>> > > > >   the cpuctl(4) device features. The utility also allows one to
> apply
>> > > > >   cpu microcode updates.
>> > > > >
>> > > > >   Currently only Intel and AMD cpus are supported and were tested.
>> > > >
>> > > > Note that cpuid isn't a privileged instruction, so I'm not sure it's
>> > really
>> > > > worth having an ioctl for that particular case.
>> > >
>> > > It was discussed when patch was reviewed on current at . The ioctl allows
>> > > to get cpuid information for specific processor, as opposed to some
>> > > random core curthread happens to run ATM.
>> >
>> > You can achieve that now with cpuset. :)  (See my ping-pong test program
>> > recently which used cpuid to fetch the APIC ID to test for ping-ponging in
>> > the scheduler.)
>>
>> If this is a backout request (for cpuid functionality) then we will do it.
>>
>> But I considered it much easier and cleaner to do
>>       fd = open("/dev/cpuctlN", O_RDWR);
>>       ioctl(fd, CPUCTL_CPUID, &x);
>>       if (x.y)
>>               ioctl(fd, CPUCTL_WRMSR, ...);
>>       close(fd);
>> then
>>       fd = open("/dev/cpuctlN", O_RDWR);
>>       cpuset(...); /* bind to cpu */
>>       __asm__("cpuid" : =0 (x));
>>       if (x.y)
>>               ioctl(fd, CPUCTL_WRMSR, ...);
>>       cpuset(...); /* restore prev mask */
>>       close(fd);
>
> You can leave it.  It is useful to specify the CPU I suppose.

I just don't think it is particularly useful to add a restriction /
priv check for information that is available in an unprivileged
fashion by other means.  I think the priv check should go away since
it doesn't achieve anything.

-- 
Peter Wemm - peter at wemm.org; peter at FreeBSD.org; peter at yahoo-inc.com; KI6FJV
"All of this is for nothing if we don't go to the stars" - JMS/B5
"If Java had true garbage collection, most programs would delete
themselves upon execution." -- Robert Sewell