[Clfs-support] i586 target with cmov

[Craig Garner]

i586, why?  Yes using some of the really cheap chip on a whatever...

Anything that you could make at this point having any value is i686 and
x86_64.

On Sat, May 21, 2016 at 6:57 PM, Chris Fowler <linxdev at gmail.com> wrote:

> This is where I'm confused.  I guess I was expecting was that all packages
> created in chapter 5 to be truly targeted to i586.  Of course I'm still
> compiling in 6.  Add in that I'm doing this on x86_64 and targeting i586
> the host and target are too similar to get a good lesson in cross-compiling
> a system.  It may be helpful if I was to use a host
> of arm-cross-linux-gnueabihf to target i586.  Probably be a two day
> compile. :)
>
> Currently I'm at 10.11 and have chroot'ed into CLFS via the i586 machine
> by mounting it NFS, setting virtuals, and doing the chroot.  I'm going to
> finish up the chapter there.
>
> If your curious this is the target:
>
> root:~/ch10# cat /proc/cpuinfo
> processor : 0
> vendor_id : Genuine  RDC
> cpu family : 5
> model : 8
> model name : 05/08
> stepping : 6
> cpu MHz : 999.991
> fdiv_bug : no
> hlt_bug : no
> f00f_bug : no
> coma_bug : no
> fpu : yes
> fpu_exception : yes
> cpuid level : 1
> wp : yes
> flags : fpu tsc cx8 mmx up
> bogomips : 1999.98
> clflush size : 32
> cache_alignment : 32
> address sizes : 32 bits physical, 32 bits virtual
> power management:
>
>
> Linux cs-buford.k4fh.com  2.6.38 #135 SMP Fri May 13 19:16:23 UTC 2016
> i586 GNU/Linux
>
> Eventually I'd like to do the multi, but for all packages.  The same root
> would target the RDC (i586) and a J1900 BayTrail (i686 32).  My kernels
> will be same version, but different processor config.  I could use initrd
> to then create all proper symlinks in root before executing pivot_root.
> These systems are small and the root it actually loaded off flash into a
> ramdisk.  The RDC sis discontinued, but I still need to support it.
>
> Chris
>
> On Tue, May 17, 2016 at 4:35 PM, William Harrington <kb0iic at berzerkula.org
> > wrote:
>
>> On Tue, May 17, 2016 19:06, Chris Fowler wrote:
>> > Yea, I've compiled Linus' code and ran on the RDC3308 itself.  I've not
>> > compiled the code in this environment.  I'm not sure if I should see
>> cmov
>> > instructions in the binary or not.  I am using uname hack.  If I rebuilt
>> > gmp and coreutils would cmov be included?
>> >
>> Hello Chris,
>>
>> In gmp configure git master:
>>
>> i586 | pentium)
>>         gcc_cflags_cpu="-mtune=pentium -mcpu=pentium -m486"
>>         gcc_cflags_arch="-march=pentium"
>>         path="x86/pentium x86"
>>
>> You can do what is done in ch10 during ch6 GMP build:
>>
>> mv -v config{fsf,}.guess
>> mv -v config{fsf,}.sub
>>
>> Then it'll use generic targets not optimized with the host CPU.
>>
>> I don't know if cmov will be removed. GMP checks the assembler to see if
>> supports cmov, and is supposed to be enabled in gmp when i686/pentium pro
>> or newer is in the target triplet. It isn't supposed to be included with
>> i586/pentium or older.
>>
>> Sincerely,
>>
>> William Harrington
>> _______________________________________________
>> Clfs-support mailing list
>> Clfs-support at lists.clfs.org
>> http://lists.clfs.org/listinfo.cgi/clfs-support-clfs.org
>>
>
>
> _______________________________________________
> Clfs-support mailing list
> Clfs-support at lists.clfs.org
> http://lists.clfs.org/listinfo.cgi/clfs-support-clfs.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clfs.org/pipermail/clfs-support-clfs.org/attachments/20160521/474bf857/attachment-0002.htm>