Linux on sparc v7

Linux on sparc v7

[Daniel Hellstrom]

Hello Dave,

I am trying to catch up on what has been happening lately with the sparc port. Have the latest patches removed support for SPARC V7 CPUs? The LEON VHDL model supports both V7 and V8 CPUs, FPU and 
non-FPU in any combination.

Daniel

Re: Linux on sparc v7

[Sam Ravnborg]

On Mon, May 21, 2012 at 04:14:20PM +0200, Daniel Hellstrom wrote:
> Hello Dave,
>
> I am trying to catch up on what has been happening lately with the sparc 
> port. Have the latest patches removed support for SPARC V7 CPUs?

Yes. The sun4c were the only users of v7 so today we tell gcc to generate v8
code. This imply support for muldiv & friends.

From the sparc Makefile:

    KBUILD_CFLAGS  += -mcpu=v8


As a side-note - we require support for the "flush" instruction now.
I assume LEON support this.

> The LEON VHDL model supports both V7 and V8 CPUs,
> FPU and non-FPU in any combination.

I do not have any overview on the FPU side of things.

	Sam

Re: Linux on sparc v7

[Daniel Hellstrom]

On 05/21/2012 06:16 PM, Sam Ravnborg wrote:
> On Mon, May 21, 2012 at 04:14:20PM +0200, Daniel Hellstrom wrote:
>> Hello Dave,
>>
>> I am trying to catch up on what has been happening lately with the sparc
>> port. Have the latest patches removed support for SPARC V7 CPUs?
> Yes. The sun4c were the only users of v7 so today we tell gcc to generate v8
> code. This imply support for muldiv&  friends.
>
> > From the sparc Makefile:
>
>      KBUILD_CFLAGS  += -mcpu=v8

Does this mean that if we change it to mcpu=v7 or mcpu=cypress (or removing mcpu=v8) the kernel can still be run on a V7 CPU? Or have software routines for making V8 mul/div operations on a V7 CPU 
been removed so that V7 CPUs can't run Linux regardless of compiler option?

> As a side-note - we require support for the "flush" instruction now.
> I assume LEON support this.
Yes, flush invalidates all of the instruction cache.

>> The LEON VHDL model supports both V7 and V8 CPUs,
>> FPU and non-FPU in any combination.
> I do not have any overview on the FPU side of things.
Thanks,
Daniel

Re: Linux on sparc v7

[Julian Calaby]

Hi,

On Tue, May 22, 2012 at 4:28 PM, Daniel Hellstrom <daniel@gaisler.com> wrote:
> On 05/21/2012 06:16 PM, Sam Ravnborg wrote:
>>
>> On Mon, May 21, 2012 at 04:14:20PM +0200, Daniel Hellstrom wrote:
>>>
>>> Hello Dave,
>>>
>>> I am trying to catch up on what has been happening lately with the sparc
>>> port. Have the latest patches removed support for SPARC V7 CPUs?
>>
>> Yes. The sun4c were the only users of v7 so today we tell gcc to generate
>> v8
>> code. This imply support for muldiv&  friends.
>>
>>
>> > From the sparc Makefile:
>>
>>     KBUILD_CFLAGS  += -mcpu=v8
>
>
> Does this mean that if we change it to mcpu=v7 or mcpu=cypress (or removing
> mcpu=v8) the kernel can still be run on a V7 CPU? Or have software routines
> for making V8 mul/div operations on a V7 CPU been removed so that V7 CPUs
> can't run Linux regardless of compiler option?

As I understand it, all support for v7 CPUs, including the mul/div
routines has been annihilated.

So from 3.5 onwards, Linux required a v8 CPU or newer.

Thanks,

-- 
Julian Calaby

Email: julian.calaby@gmail.com
Profile: http://www.google.com/profiles/julian.calaby/
.Plan: http://sites.google.com/site/juliancalaby/

Re: Linux on sparc v7

[David Miller]

From: Daniel Hellstrom <daniel@gaisler.com>
Date: Tue, 22 May 2012 08:28:24 +0200

> On 05/21/2012 06:16 PM, Sam Ravnborg wrote:
>> On Mon, May 21, 2012 at 04:14:20PM +0200, Daniel Hellstrom wrote:
>>> Hello Dave,
>>>
>>> I am trying to catch up on what has been happening lately with the
>>> sparc
>>> port. Have the latest patches removed support for SPARC V7 CPUs?
>> Yes. The sun4c were the only users of v7 so today we tell gcc to
>> generate v8
>> code. This imply support for muldiv&  friends.
>>
>> > From the sparc Makefile:
>>
>>      KBUILD_CFLAGS  += -mcpu=v8
> 
> Does this mean that if we change it to mcpu=v7 or mcpu=cypress (or
> removing mcpu=v8) the kernel can still be run on a V7 CPU? Or have
> software routines for making V8 mul/div operations on a V7 CPU been
> removed so that V7 CPUs can't run Linux regardless of compiler option?

The kernel will never run on a v7 cpu ever again, as we've removed
support for the sun4c architecture completely.

Re: Linux on sparc v7

[Daniel Hellstrom]

On 05/22/2012 08:28 AM, Daniel Hellstrom wrote:
> On 05/21/2012 06:16 PM, Sam Ravnborg wrote:
>> On Mon, May 21, 2012 at 04:14:20PM +0200, Daniel Hellstrom wrote:
>>> Hello Dave,
>>>
>>> I am trying to catch up on what has been happening lately with the sparc
>>> port. Have the latest patches removed support for SPARC V7 CPUs?
>> Yes. The sun4c were the only users of v7 so today we tell gcc to generate v8
>> code. This imply support for muldiv&  friends.
>>
>> > From the sparc Makefile:
>>
>>      KBUILD_CFLAGS  += -mcpu=v8
>
> Does this mean that if we change it to mcpu=v7 or mcpu=cypress (or removing mcpu=v8) the kernel can still be run on a V7 CPU? Or have software routines for making V8 mul/div operations on a V7 CPU 
> been removed so that V7 CPUs can't run Linux regardless of compiler option?
>
>> As a side-note - we require support for the "flush" instruction now.
>> I assume LEON support this.
> Yes, flush invalidates all of the instruction cache.

Note that it will also flush the LEON data cache. Flushing on a LEON is normally done:
  * I-cache, write anywhere using ASI=0x15
  * D-cache, write anywhere using ASI=0x16

Daniel

>
>>> The LEON VHDL model supports both V7 and V8 CPUs,
>>> FPU and non-FPU in any combination.
>> I do not have any overview on the FPU side of things.
> Thanks,
> Daniel
> -- 
> To unsubscribe from this list: send the line "unsubscribe sparclinux" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>