Click Here
home features news forums classifieds faqs links search
6071 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

Support Amigaworld.net
Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
Donate

Menu
Main sections
» Home
» Features
» News
» Forums
» Classifieds
» Links
» Downloads
Extras
» OS4 Zone
» IRC Network
» AmigaWorld Radio
» Newsfeed
» Top Members
» Amiga Dealers
Information
» About Us
» FAQs
» Advertise
» Polls
» Terms of Service
» Search

IRC Channel
Server: irc.amigaworld.net
Ports: 1024,5555, 6665-6669
SSL port: 6697
Channel: #Amigaworld
Channel Policy and Guidelines

Who's Online
14 crawler(s) on-line.
 164 guest(s) on-line.
 1 member(s) on-line.


 Hypex

You are an anonymous user.
Register Now!
 Hypex:  2 mins ago
 Gunnar:  8 mins ago
 DiscreetFX:  20 mins ago
 saimo:  30 mins ago
 amigakit:  48 mins ago
 OldFart:  49 mins ago
 _ThEcRoW:  1 hr 8 mins ago
 NutsAboutAmiga:  1 hr 36 mins ago
 retrofaza:  1 hr 36 mins ago
 Rob:  1 hr 37 mins ago

/  Forum Index
   /  Amiga PPC Linux
      /  NEWS on amigaone-linux.sourceforge.net
Register To Post

Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 Next Page )
PosterThread
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 20-Jun-2006 13:31:33
#21 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Quote:
AFAIK it's the VIA that should do the ACPI. It's a bit of a mistery, like the onboard-sound issue shows.
Maybe there is a way to access the registers, but no one knows.


It's the same for the hardware monitor logic. Even if access to the registers (I/O port) is possible, the southbridge doesn't seem to monitor the fan speeds and supply voltages.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
AV 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 20-Jun-2006 15:42:27
#22 ]
Regular Member
Joined: 1-Aug-2003
Posts: 184
From: Unknown

@Geri

Quote:

It's the same for the hardware monitor logic. Even if access to the registers (I/O port) is possible, the southbridge doesn't seem to monitor the fan speeds and supply voltages.


As I hinted previously, it may be the same kind of problem with the onboard audio: it's possible that part of the VIA is not powered up.

Or the lines are not connected. I once tried looking into the VIA specs but couldn't understand how to do soft-off. Especially since you need an x86 BIOS to find the right pointers into the VIA address space (unless I misunderstood everything, that is also a possibility).
Do you have some ready-made code or textual example on how to do soft-off ? I might try that under OS4.


 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 20-Jun-2006 22:16:33
#23 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Quote:
Or the lines are not connected. I once tried looking into the VIA specs but couldn't understand how to do soft-off. Especially since you need an x86 BIOS to find the right pointers into the VIA address space (unless I misunderstood everything, that is also a possibility).
Do you have some ready-made code or textual example on how to do soft-off ? I might try that under OS4.


Well, I relocated the ACPI IO space to the addresses 0x4000 for power management/ACPI, 0x5000 for SMBus and 0x6000 for hardware monitoring, because one of my old PCs with a VIA82C686A southbridge uses the same IO space address locations (AFAIK the main difference between the A and the B version is just the IDE controller with UDMA100 support).
Naturally I enabled the IO bases by settings the bit 0x40[7] = 1 for ACPI, 0x74[0] = 1 for the hardware monitor and 0xD2[0] = 1 for the SMBus controller.

While I could write to the hardware monitoring registers (I think so at least), it was not possible to write to the ACPI registers. I tried to set the "Sleep Type" (IO space offset 0x5) in the ACPI IO space to "110b" which means "Power on Suspend with CPU/PCI reset". If I understood the VIA spec correctly a write to bit 0x05[5] (or 0x04[13] for 16 bit access) (SLP_EN) should put the system in soft-off state.

I also compared the ACPI PCI configuration space of the AmigaOne with those of my PC with the VIA82C686A southbridge and found out that the power supply is detected as of an AT type instead as of an ATX type (bit at 0x42[7]). Also the bits for "RTC enable gated with PSON" at 0x41[2], "STR Power Well Output Gating" at 0x54[6] and "SUSST1/GPO3 Select (Pin V10)" at 0x54[4] are not set on the A1. Maybe they influence the ACPI/soft-off control logic, but that doesn't explain the non writeable ACPI IO base.

Another possibility would be to set bit 0x54[5] "SUSC# = 0 for STR Power Well Output Gating" (together with bit 0x54[6]) to put the system into soft-off state, but I didn't have the time to try this out.

Geri

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 21-Jun-2006 9:27:26
#24 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

Quote:
I tried to set the "Sleep Type" (IO space offset 0x5) in the ACPI IO space to "110b" which means "Power on Suspend with CPU/PCI reset".


Oops, SLP_TYP = "010b" "Suspend to Disk" should power down the system.

BTW: I took a look at the ACPI implementation in Linux and it seems that writing to the SLP_EN and SLP_TYP bits is the right way to force the system into another power mode.

Last edited by Geri on 21-Jun-2006 at 04:42 PM.
Last edited by Geri on 21-Jun-2006 at 09:27 AM.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
acefnq 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 21-Jun-2006 11:45:07
#25 ]
Cult Member
Joined: 11-Jan-2006
Posts: 617
From: Adelaide, South Australia

@lionstorm and Geri
Thanks guys, best of luck Geri.

ace

 Status: Offline
Profile     Report this post  
AV 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 21-Jun-2006 22:33:36
#26 ]
Regular Member
Joined: 1-Aug-2003
Posts: 184
From: Unknown

@Geri

Thanks for the info.
In case you succeed with this lastest method, post some notes here and I'll try to
do something on the OS4 side.
When you say that you've relocated the various address spaces, you mean you moved
the BARs ?

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 22-Jun-2006 7:31:18
#27 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Quote:
When you say that you've relocated the various address spaces, you mean you moved the BARs ?


Yes. AFAIK u-boot doesn't initialize the BARs, right?

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 22-Jun-2006 19:53:58
#28 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Hmm, soft-off doesn't seem to work just by setting the SLP_TYP and SLP_EN bits.

On the other side hardware monitoring is working (fan speeds, 3.3V and 12V only).

I will try to investigate this after I have released a new kernel.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
AV 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 22-Jun-2006 22:11:11
#29 ]
Regular Member
Joined: 1-Aug-2003
Posts: 184
From: Unknown

@Geri

Quote:

Yes. AFAIK u-boot doesn't initialize the BARs, right?


Don't know, I should check in the sources. IIRC there is some "PCI autoconfiguration" but I don't know how well it works, if it works at all. Rogue should know better.

In what state do you find the BARs under Linux ? Do they make sense ?

 Status: Offline
Profile     Report this post  
AV 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 22-Jun-2006 22:13:49
#30 ]
Regular Member
Joined: 1-Aug-2003
Posts: 184
From: Unknown

@Geri

Quote:


Hmm, soft-off doesn't seem to work just by setting the SLP_TYP and SLP_EN bits.


Blast.

Quote:

On the other side hardware monitoring is working (fan speeds, 3.3V and 12V only).

I will try to investigate this after I have released a new kernel.


Wow, just knowing that the HW monitoring is working is very good news. Good work!
Once you're done with the kernel, could you spare few minutes and post some code snippets that show how to get the monitoring values ? Thanks.

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 23-Jun-2006 7:22:42
#31 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Quote:
In what state do you find the BARs under Linux ? Do they make sense ?


The PCI IO BARs are cleared.

Quote:

Quote:
Hmm, soft-off doesn't seem to work just by setting the SLP_TYP and SLP_EN bits.


Blast.


IMHO this has todo something with this AT/ATX power supply bit in the ACPI config space, which is set to AT style power supply on the AmigaOne. Maybe the SUSC# signal is ANDed with this bit and therefore doesn't allow to powerdown the system.
Somebody with a better knowledge of the A1 hardware than me should check, if this bit can be forced to '0' in hardware, because the bit is read only.

Quote:
Wow, just knowing that the HW monitoring is working is very good news. Good work! Once you're done with the kernel, could you spare few minutes and post some code snippets that show how to get the monitoring values ? Thanks.


Thanks!

Just set the I/O base address (BAR) of the hardware monitoring block and enable it, after you have powered up the VIA. Then you have to start the hardware monitoring by setting a bit in the I/O port. I will send you a code snippet, when I have done the kernel. You should take a look at the VIA lm-sensors driver http://lxr.linux.no/source/drivers/i2c/chips/via686a.c to see how the monitoring values are read out, because I just tried to get this driver working.

I did a quick test of the hardware monitoring logic and I think the fan speed is to high. I will do some measurements next week.

Geri

Last edited by Geri on 23-Jun-2006 at 07:53 AM.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 23-Jun-2006 23:59:13
#32 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@AV

Hmm, seems OS4 fooled me a little bit.

OS4 powered up also the hardware monitoring logic, therefore it should be enough to set the I/O base address, enable the hardware I/O block (0x74[0] = 1) and set the start bit at I/O offset 0x40[0] (+ I/O base address) to start the hardware monitoring logic.

How about integrating the powerup code and setting the ACPI I/O bases in U-boot?
(I don't want to do a bad hack in Linux)

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
zErec 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 9:56:02
#33 ]
Cult Member
Joined: 17-Dec-2003
Posts: 514
From: Germany

@Geri

Hi Geri, how are you proceeding with the Kernel and MOL?

greetz zErec

_________________
- AmigaX1Ooo with OS4.1FE // CD32/SX32Pro/SX1/CD32Shuttle
- A600-Vampire600 FPGA // A4KT/CSPPC233,144MB,PCI, AOS4.1.6FE
- MACmini 1.5GHz // Pegasos 2 1GHz // Efika
- ATARI Falcon/060/SuperVidel // ATARI Falcon MK X

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 11:06:49
#34 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@zErec

Quote:
Hi Geri, how are you proceeding with the Kernel and MOL?


Sorry, no MOL modules yet. I was busy trying to fix a problem with the ALSA drivers (which unfortunately didn't work out) and compiling kernels.

I'll will see, if I can upload a new kernel later this day.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 13:50:55
#35 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

On amigaone-linux.sourceforge.net/download.php you can find a first testing version of a 2.6.8 Linux kernel for the A1.

The following changes were made in comparison to PegXlin's lastest kernel:

1. Generic non cache coherent DMA support:
This fix should avoid any data corruption with harddisks operating in UDMA mode or during ethernet transfers. The patch was already included in the kernel and has been made to work on the A1 now. It seems to work, even if Freescale's PowerPC Programming Environment Manual says something else.

Unfortunately not all drivers can make use of this patch, for example the ALSA sound drivers. The generic DMA API was never extended to support some features required by the ALSA core and drivers. The ALSA developers are working on this, but it may take a while until it will find its way into an official kernel. Therefore ALSA still uses the normal DMA allocation functions (not the non cache coherent ones), which may result in an unstable system!

2. Automatic DMA support for IDE drives:
The kernel now activates DMA automatically for IDE drives, if you add ide0=ata66 and ide1=ata66 to the kernels command line. The kernel will choose the right DMA mode for the drive. You can still change the DMA rate with 'hdparm'.
Don't expect high transfer rates for IDE drives and ethernet, because the non coherent DMA patch slows down the system somewhat. IIRC I get around 20 to 30MB/s with a harddisk running in UDMA5 mode (VIA southbridge IDE controller). If you experience DMA timeout errors or lost interrupts during UDMA operation, you should switch off DMA mode to avoid data corruption! (I already limited the request size for UDMA transfers, but that may not help in all cases.)

3. Highmem support:
The kernel should support more that 896MB RAM now. If you have more RAM than that installed in your A1 you can remove the memory limit in U-boot or the "mem=" option on the command line. I only have 512MB RAM, so I couldn't test this feature. Reports on this are welcome!

4. Removed PegXlin VIA patch.
PegXlin included a patch for the VIA southbridge in his A1 kernel. This patch was originally for the VIA (KT133?) northbridge, so it did nothing useful on the A1.

5. L2 cache setting:
The L2 cache is now activated automatically (0x80000000), if you don't specify the address of the L2 cache on the command line.

-- edit: revised --
6. MOL 0.9.70 kernel modules included (just not as Debian package). I didn't test them, so MOL may not work at all! I just did a quick compile! Install the Debian sarge MOL packages to use MOL with this kernel.
-- end --

7. Support for hardware monitoring is not yet included, because I think this should be integrated into U-boot.

The source code will be made available:

Some words about the future of the A1 Linux kernel:

The Linux PPC developers are moving to a new architecture for PowerPC Linux (ppc -> powerpc) since kernel 2.6.15. To be able to support the AmigaOne under the new powerpc architecture, I would need a new A1 U-boot release, which supports a flattended device tree (FDT). AFAIK support for the FDT is a requirement for the powerpc architecture.

I would like to thank the following people:

- Benjamin Herrenschmidt - for his constant and patient support!
- Ross Vumbaca - for giving me access to the amigaone-linux sourceforge project.
- all the other Debian developes that commented my questions and last but not least
- all the A1 users that still use Linux on their A1s.

Last edited by Geri on 20-Jul-2006 at 07:37 AM.
Last edited by Geri on 05-Jul-2006 at 05:39 PM.
Last edited by Geri on 02-Jul-2006 at 01:58 PM.
Last edited by Geri on 02-Jul-2006 at 01:54 PM.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
Kicko 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 14:02:05
#36 ]
Elite Member
Joined: 19-Jun-2004
Posts: 5009
From: Sweden

Cool with an update. Will test as soon as i have time.

 Status: Offline
Profile     Report this post  
deakmann 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 16:03:44
#37 ]
Regular Member
Joined: 20-Mar-2003
Posts: 360
From: Unknown

@Geri

This is great news , it`s been a while since i`ve updated my Linux setup still running Ross`s 2.4.26 with KDE 3.1.5.

I learn`t a lot getting this system , sadly forgot most of it now but I still make a fare bit of use of LInux, especially since my OS4 drive has been playing up a bit lately.

I take it I can use the same instructions to update the kernel? Will I have all the same functinonality with the new kernel?

On the question of ALSA i`m currently using the BT 878 audio DMA option for sound , is this ALSA?

Is it safe to use the VIA now? I`m currently using a SIL card, as I understand it Ross`s kernel had a patch for the cache coherency problem but the SIL was still required because of a problem with the VIA.

Do you have any plans to do a new installation CD?

Cheers

Andy

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 17:23:23
#38 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@deakmann

Quote:
I take it I can use the same instructions to update the kernel? Will I have all the same functinonality with the new kernel?


Yes, the instructions from amigaone-linux.sourceforge.net/docs.php still apply.

Quote:
On the question of ALSA i`m currently using the BT 878 audio DMA option for sound , is this ALSA?


Yes, ALSA contains a BT878 sound driver.

Quote:
Is it safe to use the VIA now? I`m currently using a SIL card, as I understand it Ross`s kernel had a patch for the cache coherency problem but the SIL was still required because of a problem with the VIA.


Ross's kernel didn't have a coherency patch. It is safe to enable UDMA for IDE devices, if the VIA hardware patch was applied to your A1 board.

Quote:
Do you have any plans to do a new installation CD?


No, not yet. There's still a lot to do for the kernel.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  
deakmann 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 17:48:06
#39 ]
Regular Member
Joined: 20-Mar-2003
Posts: 360
From: Unknown

@Geri

Quote:
Yes, ALSA contains a BT878 sound driver.


So there could still be issues with sound ? or this there a reccommended alternative.

Quote:
Ross's kernel didn't have a coherency patch. It is safe to enable UDMA for IDE devices, if the VIA hardware patch was applied to your A1 board.


I though it had what Ross called a nasty hack written by Bill Mueller? Otherwise how would I be using DMA reliably on the Sil Card?

Quote:
No, not yet. There's still a lot to do for the kernel.


Look forward to seeing the results :)

 Status: Offline
Profile     Report this post  
Geri 
Re: NEWS on amigaone-linux.sourceforge.net
Posted on 2-Jul-2006 18:05:27
#40 ]
Elite Member
Joined: 7-Oct-2003
Posts: 2038
From: ST/AT

@deakmann

Quote:
So there could still be issues with sound ? or this there a reccommended alternative.


Right! This is a problem that affects all non cache coherent platforms (for example ARM, PPC 4xx/8xx), so we have to wait until the ALSA developers have finished rewriting their code.

Quote:
I though it had what Ross called a nasty hack written by Bill Mueller? Otherwise how would I be using DMA reliably on the Sil Card?


I wouldn't call this hack reliable, even with a SiI controller! This kernel uses uncached memory for all DMA buffers and therefore fixes the coherency problem of the ArticiaS northbridge. I did about 500 MD5 checksum tests of big ISO files and couldn't trigger any checksum error, as it was the case with older kernels.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

 Status: Offline
Profile     Report this post  

[ home ][ about us ][ privacy ] [ forums ][ classifieds ] [ links ][ news archive ] [ link to us ][ user account ]
Copyright (C) 2000 - 2019 Amigaworld.net.
Amigaworld.net was originally founded by David Doyle