[9fans] Plan 9 on mmuless systems.

[9fans] Plan 9 on mmuless systems.

[Noah Evans]

Has there been any work on porting Plan 9 to systems without an mmu?  
I seem to recall Sape saying something about running Plan 9 embedded  
at the labs. Has any of that work made it back to the main tree?

Best Regards,

Noah

Re: [9fans] Plan 9 on mmuless systems.

[Russ Cox]

> Has there been any work on porting Plan 9 to systems without an mmu?
> I seem to recall Sape saying something about running Plan 9 embedded
> at the labs. Has any of that work made it back to the main tree?

All the embedded systems that run Plan 9 have MMUs.

Russ

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

That's too bad. I'm stuck with inferno I guess. :(

Noah

On Jan 15, 2006, at 9:15 PM, Russ Cox wrote:

>> Has there been any work on porting Plan 9 to systems without an mmu?
>> I seem to recall Sape saying something about running Plan 9 embedded
>> at the labs. Has any of that work made it back to the main tree?
>
> All the embedded systems that run Plan 9 have MMUs.
>
> Russ

Re: [9fans] Plan 9 on mmuless systems.

[Bruce Ellis]

i found that for various projects (e.g. ps2) no mmu was great.
no tlb misses - it just screams.  in a situation where you don't
need a mmu inferno rocks.

brucee

On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
> That's too bad. I'm stuck with inferno I guess. :(
>
> Noah
>
> On Jan 15, 2006, at 9:15 PM, Russ Cox wrote:
>
> >> Has there been any work on porting Plan 9 to systems without an mmu?
> >> I seem to recall Sape saying something about running Plan 9 embedded
> >> at the labs. Has any of that work made it back to the main tree?
> >
> > All the embedded systems that run Plan 9 have MMUs.
> >
> > Russ
>
>

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

On the case :) Last night before I went to bed I was trying to cross  
compile for an ARM using Inferno hosted on OSX. The endianness of my  
binaries seems to be reversed using 5a. Any idea what would cause  
this? Also, if you try to put in ROM headers in 5a using WORD how do  
you avoid 5a optimizing away your data(it's unreachable). Put in  
another TEXT?

Noah

On Jan 16, 2006, at 3:08 AM, Bruce Ellis wrote:

> i found that for various projects (e.g. ps2) no mmu was great.
> no tlb misses - it just screams.  in a situation where you don't
> need a mmu inferno rocks.
>
> brucee
>
> On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
>> That's too bad. I'm stuck with inferno I guess. :(
>>
>> Noah
>>
>> On Jan 15, 2006, at 9:15 PM, Russ Cox wrote:
>>
>>>> Has there been any work on porting Plan 9 to systems without an  
>>>> mmu?
>>>> I seem to recall Sape saying something about running Plan 9  
>>>> embedded
>>>> at the labs. Has any of that work made it back to the main tree?
>>>
>>> All the embedded systems that run Plan 9 have MMUs.
>>>
>>> Russ
>>
>>

Re: [9fans] Plan 9 on mmuless systems.

[Charles Forsyth]

5a isn't usually involved; i'd expect it to be 5l if anything.  it and 5c assume
little-endian (but the assumptions are isolated).
you've got a big-endian arm?  what is it?

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

Hey,

No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo DS). I  
dumped both the l.5 and the 5.out and both seem to be reversed. FWIW  
when I get the linker to produce assembly the values are correct.

Noah

On Jan 16, 2006, at 9:07 AM, Charles Forsyth wrote:

> 5a isn't usually involved; i'd expect it to be 5l if anything.  it  
> and 5c assume
> little-endian (but the assumptions are isolated).
> you've got a big-endian arm?  what is it?
>

Re: [9fans] Plan 9 on mmuless systems.

[Charles Forsyth]

> No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo DS). I  
> dumped both the l.5 and the 5.out and both seem to be reversed. FWIW  

dumped them how?

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

OS X od. My hex values are getting reversed, i.e. 3031    9600  
becomes 0096    3130 when I dump it. When I use 5l -a though, the  
values are correct. What am I doing wrong?

Noah

On Jan 16, 2006, at 10:25 AM, Charles Forsyth wrote:

>> No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo DS). I
>> dumped both the l.5 and the 5.out and both seem to be reversed. FWIW
>
> dumped them how?
>

Re: [9fans] Plan 9 on mmuless systems.

[Charles Forsyth]

> OS X od. My hex values are getting reversed, i.e. 3031    9600  
> becomes 0096    3130 when I dump it. When I use 5l -a though, the  
> values are correct. What am I doing wrong?

assuming od is displaying the same order on a big endian machine as you'd see on a little-endian machine.
i see that od on mac is `obsolete', replaced by hexdump, which sounds fine,
but that dumps in octal too, half the time!  do they have brainstorming sessions where they invent these things?

try hexdump -C ...
and see if that makes any difference.

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

I don't think that's it. I have a reference ROM header that I'm  
comparing the dumps to and the ROM header dumps fine. I tried it in  
Inferno's xd just to be sure and the values were still reversed.  
Hmmmm... this is beginning to spam the list, should we take this  
conversation to IRC or IM?

Noah

On Jan 16, 2006, at 10:42 AM, Charles Forsyth wrote:

>> OS X od. My hex values are getting reversed, i.e. 3031    9600
>> becomes 0096    3130 when I dump it. When I use 5l -a though, the
>> values are correct. What am I doing wrong?
>
> assuming od is displaying the same order on a big endian machine as  
> you'd see on a little-endian machine.
> i see that od on mac is `obsolete', replaced by hexdump, which  
> sounds fine,
> but that dumps in octal too, half the time!  do they have  
> brainstorming sessions where they invent these things?
>
> try hexdump -C ...
> and see if that makes any difference.
>

Re: [9fans] Plan 9 on mmuless systems.

[Bruce Ellis]

using OS X?

brucee

On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
> OS X od. My hex values are getting reversed, i.e. 3031    9600
> becomes 0096    3130 when I dump it. When I use 5l -a though, the
> values are correct. What am I doing wrong?
>
> Noah
>
> On Jan 16, 2006, at 10:25 AM, Charles Forsyth wrote:
>
> >> No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo DS). I
> >> dumped both the l.5 and the 5.out and both seem to be reversed. FWIW
> >
> > dumped them how?
> >
>
>

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

Yes, although the same thing happens if I dump in Inferno :)

On Jan 16, 2006, at 10:45 AM, Bruce Ellis wrote:

> using OS X?
>
> brucee
>
> On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
>> OS X od. My hex values are getting reversed, i.e. 3031    9600
>> becomes 0096    3130 when I dump it. When I use 5l -a though, the
>> values are correct. What am I doing wrong?
>>
>> Noah
>>
>> On Jan 16, 2006, at 10:25 AM, Charles Forsyth wrote:
>>
>>>> No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo  
>>>> DS). I
>>>> dumped both the l.5 and the 5.out and both seem to be reversed.  
>>>> FWIW
>>>
>>> dumped them how?
>>>
>>
>>

Re: [9fans] Plan 9 on mmuless systems.

[Bruce Ellis]

just the usual response to "fix my mac".  make up
your own suggestion of what brucee would say.

brucee

On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
> Yes, although the same thing happens if I dump in Inferno :)
>
> On Jan 16, 2006, at 10:45 AM, Bruce Ellis wrote:
>
> > using OS X?
> >
> > brucee
> >
> > On 1/16/06, Noah Evans <noah.evans@gmail.com> wrote:
> >> OS X od. My hex values are getting reversed, i.e. 3031    9600
> >> becomes 0096    3130 when I dump it. When I use 5l -a though, the
> >> values are correct. What am I doing wrong?
> >>
> >> Noah
> >>
> >> On Jan 16, 2006, at 10:25 AM, Charles Forsyth wrote:
> >>
> >>>> No, I'm cross compiling (power)big->little(ARM7/ARM9 Nintendo
> >>>> DS). I
> >>>> dumped both the l.5 and the 5.out and both seem to be reversed.
> >>>> FWIW
> >>>
> >>> dumped them how?
> >>>
> >>
> >>
>
>

Re: [9fans] Plan 9 on mmuless systems.

[Noah Evans]

I lied, I meant just after linking, not after 5l.

Noah

On Jan 16, 2006, at 9:07 AM, Charles Forsyth wrote:

> 5a isn't usually involved; i'd expect it to be 5l if anything.  it  
> and 5c assume
> little-endian (but the assumptions are isolated).
> you've got a big-endian arm?  what is it?
>

Re: [9fans] Plan 9 on mmuless systems.

[Brantley Coile]

> That's too bad. I'm stuck with inferno I guess. :(

With no mmu, doing your coding in Limbo will be a huge win.
Limbo is type safe.  Something that has great advantages
when working without the `net' of an mmu.

 Brantley