support of arm64 architecture

support of arm64 architecture

[Raphaël Poggi]

Hi all,

I am working on porting barebox on arm64 architecture.

So I have some questions about it:

- Is there any interest for barebox to support this architecture ?

- My port is running at EL1 for the moment, does the bootloader have
to run at EL3 ? If yes, do you have hint to achieve this ?
      I think UEFI is running at EL3, but I am confused about that,
what is the difference between UEFI and barebox ? Are they both
booloader ? (I guess it is a "noob" question :/)

- I have some issues with malloc, when I use tlsf all malloc failed,
but not with dlmalloc implementation, do you have an idea about this ?


At the moment the current limitations of my port are :
     - MMU not implemented
     - barebox running at EL1
     - barebox env not working
     - only test on qemu

If you want to test it:
https://github.com/raphui/barebox/tree/dev/armv8a_cleanup

I can also send a patch with [RFC] tag is needed.

If you have any suggestions or questions, tell me.

Thank you,

Raphaël Poggi

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: support of arm64 architecture

[Sascha Hauer]

Hi Raphaël,

On Fri, Apr 08, 2016 at 02:01:21PM +0200, Raphaël Poggi wrote:
> Hi all,
> 
> I am working on porting barebox on arm64 architecture.
> 
> So I have some questions about it:
> 
> - Is there any interest for barebox to support this architecture ?

Definitely, yes.

> 
> - My port is running at EL1 for the moment, does the bootloader have
> to run at EL3 ? If yes, do you have hint to achieve this ?
>       I think UEFI is running at EL3, but I am confused about that,
> what is the difference between UEFI and barebox ? Are they both
> booloader ? (I guess it is a "noob" question :/)

UEFI is an interface to the firmware whereas barebox is a firmware. We
could implement a UEFI interface for barebox (the U-Boot guys already
did this). Also barebox can run on top of firmware with a (U)Efi
interface.

> 
> - I have some issues with malloc, when I use tlsf all malloc failed,
> but not with dlmalloc implementation, do you have an idea about this ?

Do you have TLSF_64BIT defined? You may have defined it automatically
when you have CONFIG_64BIT defined, but if not tlsf malloc will not
work.

> 
> 
> At the moment the current limitations of my port are :
>      - MMU not implemented
>      - barebox running at EL1
>      - barebox env not working
>      - only test on qemu
> 
> If you want to test it:
> https://github.com/raphui/barebox/tree/dev/armv8a_cleanup

From a first look it the arm64 code looks very familiar. Do you think we
could merge it into the arm architecture rather than creating a new
architecture?

I don't know in which exception level barebox should finally run in, but
I don't think that's a showstopper.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: support of arm64 architecture

[Lucas Stach]

Am Freitag, den 08.04.2016, 14:01 +0200 schrieb Raphaël Poggi:
> Hi all,
> 
> I am working on porting barebox on arm64 architecture.
> 
> So I have some questions about it:
> 
> - Is there any interest for barebox to support this architecture ?
> 
Now that I have an actually interesting ARMv8 system myself I would
like to see some architecture support land in barebox.

> - My port is running at EL1 for the moment, does the bootloader have
> to run at EL3 ? If yes, do you have hint to achieve this ?

Depending on your system setup both is a valid thing. If you have a
full blown firmware hypervisor below barebox EL1 is okay.

Though normally barebox is the system firmware itself, in which case it
must be able to run at EL3 to provide all system services including
duties of the secure monitor. If there is only a secure monitor below,
barebox may be started in EL2.

Regards,
Lucas

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: support of arm64 architecture

[Raphaël Poggi]

Hi,

Thank you for your answer.

At the beginning, I have tried to add arm64 specific code on the arm
architecture folder, but it becomes a mess...

Nevertheless, I can give an another try with this solution and then we
can decide.

2016-04-08 19:30 GMT+02:00 Sascha Hauer <s.hauer@pengutronix.de>:
> Hi Raphaël,
>
> On Fri, Apr 08, 2016 at 02:01:21PM +0200, Raphaël Poggi wrote:
>> Hi all,
>>
>> I am working on porting barebox on arm64 architecture.
>>
>> So I have some questions about it:
>>
>> - Is there any interest for barebox to support this architecture ?
>
> Definitely, yes.
>
>>
>> - My port is running at EL1 for the moment, does the bootloader have
>> to run at EL3 ? If yes, do you have hint to achieve this ?
>>       I think UEFI is running at EL3, but I am confused about that,
>> what is the difference between UEFI and barebox ? Are they both
>> booloader ? (I guess it is a "noob" question :/)
>
> UEFI is an interface to the firmware whereas barebox is a firmware. We
> could implement a UEFI interface for barebox (the U-Boot guys already
> did this). Also barebox can run on top of firmware with a (U)Efi
> interface.
>
>>
>> - I have some issues with malloc, when I use tlsf all malloc failed,
>> but not with dlmalloc implementation, do you have an idea about this ?
>
> Do you have TLSF_64BIT defined? You may have defined it automatically
> when you have CONFIG_64BIT defined, but if not tlsf malloc will not
> work.
>
>>
>>
>> At the moment the current limitations of my port are :
>>      - MMU not implemented
>>      - barebox running at EL1
>>      - barebox env not working
>>      - only test on qemu
>>
>> If you want to test it:
>> https://github.com/raphui/barebox/tree/dev/armv8a_cleanup
>
> From a first look it the arm64 code looks very familiar. Do you think we
> could merge it into the arm architecture rather than creating a new
> architecture?
>
> I don't know in which exception level barebox should finally run in, but
> I don't think that's a showstopper.
>
> Sascha
>
> --
> Pengutronix e.K.                           |                             |
> Industrial Linux Solutions                 | http://www.pengutronix.de/  |
> Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
>
> _______________________________________________
> barebox mailing list
> barebox@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/barebox

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: support of arm64 architecture

[Sascha Hauer]

On Mon, Apr 11, 2016 at 08:49:00AM +0200, Raphaël Poggi wrote:
> Hi,
> 
> Thank you for your answer.
> 
> At the beginning, I have tried to add arm64 specific code on the arm
> architecture folder, but it becomes a mess...
> 
> Nevertheless, I can give an another try with this solution and then we
> can decide.

I think the end result should be that both architectures are merged, but
maybe to get there it is easier to work with separated architectures in
the mean time.

Sascha


-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox