linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
@ 2004-08-27 11:34 Andrey Volkov
  2004-08-27 14:15 ` Kumar Gala
  2004-08-27 15:00 ` Dan Malek
  0 siblings, 2 replies; 6+ messages in thread
From: Andrey Volkov @ 2004-08-27 11:34 UTC (permalink / raw)
  To: linuxppc-embedded


Hello all,

Since G2_LE is NOT a "classic" core ( in particular - true little-endian),
may be now is time to remove it from CLASSIC_PPC part of
arch/ppc/kernel/cputable.c, and move it to itself part.
I bear in mind next:

from arch/ppc/kernel/cputable.c:

#if CLASSIC_PPC
.....
    {   /* All G2_LE (603e core, plus some) have the same pvr */
        0x7fff0000, 0x00820000, "G2_LE",
        CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
        CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
        COMMON_PPC,
        32, 32,
        __setup_cpu_603
    },
....
#endif /* CLASSIC_PPC */

May be better like this:

#if CONFIG_G2_LE
    {   /* All G2_LE (603e core, plus some) have the same pvr */
        0x7fff0000, 0x00820000, "G2_LE",
        CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
        CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
        COMMON_PPC,
        32, 32,
        __setup_cpu_603
    },
#endif

With that move we'll conserve some memory of kernel,
but need some explanations which PPCs based on
it, for modify Kconfig ( I know MPC52xx and MPC82xx based on it).

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
  2004-08-27 11:34 [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c Andrey Volkov
@ 2004-08-27 14:15 ` Kumar Gala
  2004-08-27 15:00   ` Re[2]: " Andrey Volkov
  2004-08-27 15:00 ` Dan Malek
  1 sibling, 1 reply; 6+ messages in thread
From: Kumar Gala @ 2004-08-27 14:15 UTC (permalink / raw)
  To: Andrey Volkov; +Cc: linuxppc-embedded


What purpose does this change really serve.  Linux does not use the
true little-endian feature of the G2_LE core, beyond that it is very
much in the class of classic ppc.  Is the memory savings really that
much?

- kumar

On Aug 27, 2004, at 6:34 AM, Andrey Volkov wrote:

>
> Hello all,
>
> Since G2_LE is NOT a "classic" core ( in particular - true
> little-endian),
> may be now is time to remove it from CLASSIC_PPC part of
> arch/ppc/kernel/cputable.c, and move it to itself part.
> I bear in mind next:
>
> from arch/ppc/kernel/cputable.c:
>
> #if CLASSIC_PPC
> .....
>     {   /* All G2_LE (603e core, plus some) have the same pvr */
>         0x7fff0000, 0x00820000, "G2_LE",
>         CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
>         CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
>         COMMON_PPC,
>         32, 32,
>         __setup_cpu_603
>     },
> ....
> #endif /* CLASSIC_PPC */
>
> May be better like this:
>
> #if CONFIG_G2_LE
>     {   /* All G2_LE (603e core, plus some) have the same pvr */
>         0x7fff0000, 0x00820000, "G2_LE",
>         CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
>         CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
>         COMMON_PPC,
>         32, 32,
>         __setup_cpu_603
>     },
> #endif
>
> With that move we'll conserve some memory of kernel,
> but need some explanations which PPCs based on
> it, for modify Kconfig ( I know MPC52xx and MPC82xx based on it).
>
> --
> Best regards,
>  Andrey Volkov
>
>


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re[2]: [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
  2004-08-27 14:15 ` Kumar Gala
@ 2004-08-27 15:00   ` Andrey Volkov
  0 siblings, 0 replies; 6+ messages in thread
From: Andrey Volkov @ 2004-08-27 15:00 UTC (permalink / raw)
  To: Kumar Gala; +Cc: linuxppc-embedded


Hello Kumar,

Friday, August 27, 2004, 6:15:20 PM, you wrote:

> What purpose does this change really serve.  Linux does not use the
> true little-endian feature of the G2_LE core, beyond that it is very
> much in the class of classic ppc.
> Is the memory savings really that much?

sizeof(*cpu_table) ~ 1300 bytes
+ code in cpu_setup_6xx.S ~ 1000 bytes
+ more simple to filter out some features in Kconfig (ex. SMP,
ALTIVEC...)

And optionally (?):
+ LE (if it will be needed sometime for somebody)
+ "An additional input interrupt signal, core_cint,
   implements a critical interrupt function."

>>
>> Since G2_LE is NOT a "classic" core ( in particular - true
>> little-endian),
>> may be now is time to remove it from CLASSIC_PPC part of
>> arch/ppc/kernel/cputable.c, and move it to itself part.
>> I bear in mind next:
>>
>> from arch/ppc/kernel/cputable.c:
>>
>> #if CLASSIC_PPC
>> .....
>>     {   /* All G2_LE (603e core, plus some) have the same pvr */
>>         0x7fff0000, 0x00820000, "G2_LE",
>>         CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
>>         CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
>>         COMMON_PPC,
>>         32, 32,
>>         __setup_cpu_603
>>     },
>> ....
>> #endif /* CLASSIC_PPC */
>>
>> May be better like this:
>>
>> #if CONFIG_G2_LE
>>     {   /* All G2_LE (603e core, plus some) have the same pvr */
>>         0x7fff0000, 0x00820000, "G2_LE",
>>         CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_CAN_DOZE | CPU_FTR_USE_TB |
>>         CPU_FTR_CAN_NAP | CPU_FTR_HAS_HIGH_BATS,
>>         COMMON_PPC,
>>         32, 32,
>>         __setup_cpu_603
>>     },
>> #endif
>>
>> With that move we'll conserve some memory of kernel,
>> but need some explanations which PPCs based on
>> it, for modify Kconfig ( I know MPC52xx and MPC82xx based on it).

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
  2004-08-27 11:34 [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c Andrey Volkov
  2004-08-27 14:15 ` Kumar Gala
@ 2004-08-27 15:00 ` Dan Malek
  2004-08-27 15:16   ` Re[2]: " Andrey Volkov
  1 sibling, 1 reply; 6+ messages in thread
From: Dan Malek @ 2004-08-27 15:00 UTC (permalink / raw)
  To: Andrey Volkov; +Cc: linuxppc-embedded


On Aug 27, 2004, at 7:34 AM, Andrey Volkov wrote:

> Since G2_LE is NOT a "classic" core ( in particular - true
> little-endian),

How is it not a classic core?  It certainly isn't bookE, and it isn't
8xx.
No one should care about little-endian anyway.  I'm voting for the
first candidate that makes that illegal. :-)

> With that move we'll conserve some memory of kernel,

How does this single change save anything but a few bytes of
cpu table?

> but need some explanations which PPCs based on
> it, for modify Kconfig ( I know MPC52xx and MPC82xx based on it).

What other kernel changes would you make based on
this configuration?


	-- Dan


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re[2]: [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
  2004-08-27 15:00 ` Dan Malek
@ 2004-08-27 15:16   ` Andrey Volkov
  2004-08-27 17:49     ` Dan Malek
  0 siblings, 1 reply; 6+ messages in thread
From: Andrey Volkov @ 2004-08-27 15:16 UTC (permalink / raw)
  To: Dan Malek; +Cc: linuxppc-embedded


Hello Dan,

Friday, August 27, 2004, 7:00:59 PM, you wrote:

>> Since G2_LE is NOT a "classic" core ( in particular - true
>> little-endian),

> How is it not a classic core?  It certainly isn't bookE, and it isn't
> 8xx.
> No one should care about little-endian anyway.  I'm voting for the
> first candidate that makes that illegal. :-)
Agree.

>> With that move we'll conserve some memory of kernel,

> How does this single change save anything but a few bytes of
> cpu table?

Maybe ~2K of rubbish for somebody "few" - for me not :) (don't forget
its embedded platform).

>> but need some explanations which PPCs based on
>> it, for modify Kconfig ( I know MPC52xx and MPC82xx based on it).

> What other kernel changes would you make based on
> this configuration?
Only Kconfig, nothing more.

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Re[2]: [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c
  2004-08-27 15:16   ` Re[2]: " Andrey Volkov
@ 2004-08-27 17:49     ` Dan Malek
  0 siblings, 0 replies; 6+ messages in thread
From: Dan Malek @ 2004-08-27 17:49 UTC (permalink / raw)
  To: Andrey Volkov; +Cc: linuxppc-embedded


On Aug 27, 2004, at 11:16 AM, Andrey Volkov wrote:

> Maybe ~2K of rubbish for somebody "few" - for me not :) (don't forget
> its embedded platform).

If you have a system of this magnitude where a couple of Kbytes is
going to be the difference between success or failure, I'd suggest
looking other places than adding an superfluous #ifdef around
the cpu table.

There are lots of places in the kernel where thorough analysis
and careful configuration will save you lots more than this and
probably make a difference.


	-- Dan


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2004-08-27 17:49 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-08-27 11:34 [RFC] Remove G2_LE core decl from CLASSIC_PPC part of arch/ppc/kernel/cputable.c Andrey Volkov
2004-08-27 14:15 ` Kumar Gala
2004-08-27 15:00   ` Re[2]: " Andrey Volkov
2004-08-27 15:00 ` Dan Malek
2004-08-27 15:16   ` Re[2]: " Andrey Volkov
2004-08-27 17:49     ` Dan Malek

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).