Re: 2.6.16-rc1-mm2 pata driver confusion

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

(sorry about breaking the threading)

On 1/24/06, Ed Sweetman <safemode@comcast.net> wrote:
> Randy.Dunlap wrote:
>
> >On Tue, 24 Jan 2006, Ed Sweetman wrote:
> >
> >
> >>Randy.Dunlap wrote:
> >>
> >>
> >>>On Tue, 24 Jan 2006, Ed Sweetman wrote:
> >>>
> >>>
> >>>>Alan Cox wrote:
> >>>>
> >>>>
> >>>>>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> >>>>>
> >>>>>
> >>>>>>problem.  The problem is that there appears to be two nvidia/amd ata
> >>>>>>drivers and I'm unsure which I should try using, if i compile both in,
> >>>>>>which get loaded first (i assume scsi is second to ide) and if i want my
> >>>>>>pata disks loaded under the new libata drivers, will my cdrom work under
> >>>>>>them too, or do i still need some sort of regular ide drivers loaded
> >>>>>>just for cdrom (to use native ata mode for recording access).
> >>>>>>
> >>>>>>
> >>>>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> >>>>>its entirity with code using the newer and cleaner libata logic. There
> >>>>>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> >>>>>are using libata and the additional patch patches still queued
> >>>>>
> >>>>>
> >>>>>>1.  Atapi is most definitely not supported by libata, right now.
> >>>>>>
> >>>>>>
> >>>>>It works in the -mm tree.
> >>>>>
> >>>>>
> >>>>Intriguing, when I had no ide chipset compiled in kernel, only libata
> >>>>drivers, I got no mention at all about my dvd writer.  I even had the
> >>>>scsi cd driver installed and generic devices, still nothing seemed to
> >>>>initialize the dvd drive.  It detected the second pata bus but no
> >>>>devices attached to it.
> >>>>
> >>>>this is using the kernel mentioned in the subject header.
> >>>>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
> >>>>
> >>>>Is there anything i can do to give more info to the list to figure out
> >>>>why my atapi writer is being ignored by pata even when there are no ide
> >>>>drivers loaded?
> >>>>
> >>>>
> >>>Currently you need to use libata.atapi_enabled=1
> >>>(assuming that libata is in the kernel image, not a loadable module).
> >>>
> >>>I just built/tested this also, working for me as well.
> >>>(hard drives, not ATAPI)
> >>>
> >>>
> >>I assume libata.atapi_enabled=1 is a boot arg, not some structure member
> >>in the source for the pata driver that i need to set to 1, correct?
> >>
> >
> >Yes, it's a kernel boot option if libata is in the kernel image.
> >If libata is a loadable module, just use something like
> >       modprobe libata atapi_enabled=1
> >
> >
> >>And you just built and tested it, how did you test if the atapi argument
> >>worked when you then say "not ATAPI" as something you tested?
> >>
> >
> >Sorry, I mean that I built and booted a kernel with libata/PATA
> >hard drive (vs. legacy drivers/ide/ PATA support).  I have not
> >tested ATAPI at all and didn't mean to imply that I had.
> >
> >I reported on libata.atapi_enabled=1 based on documentation
> >and other emails that I have read.
> >
> >
> >>In any case, i'll try out libata.atapi_enabled=1 and see if it detects
> >>the dvd drive.
> >>
> >
> >HTH.  Please continue to post any questions or problems.
> >
>
> I Rebooted several times, both setting the option in the kernel boot
> args and editing the source to have it set by default.  No atapi devices
> are found/mentioned or even described as not found in dmesg/bootup.   So
> apparently, on my chipset, the amd/nvidia pata driver is not detecting
> atapi devices.
>
> 0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
> 0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA
> Controller (rev f3)
> 0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA
> Controller (rev f3)
>
> 0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
> (prog-if 8a [Master SecP PriP])
>         Subsystem: Unknown device f043:815a
>         Flags: bus master, 66MHz, fast devsel, latency 0
>         I/O ports at f000 [size=16]
>         Capabilities: [44] Power Management version 2
>
>
> the atapi device in question is a plextor px-712A, it's the only device
> on the secondary channel.

And this is with using only ATA (libata) drivers in drivers/scsi/
and not ATA drivers in drivers/ide/, right?

Hm.  I guess we treat this as a bug report for NV ATA/ATAPI then.

I just tested my system with a Plextor PX-712SA drive plus
booting with libata.atapi_enabled=1 and the driver (not nv)
sees the ATAPI drive and can read it.

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Randy.Dunlap wrote:

>(sorry about breaking the threading)
>
>On 1/24/06, Ed Sweetman <safemode@comcast.net> wrote:
>  
>
>>Randy.Dunlap wrote:
>>
>>    
>>
>>>On Tue, 24 Jan 2006, Ed Sweetman wrote:
>>>
>>>
>>>      
>>>
>>>>Randy.Dunlap wrote:
>>>>
>>>>
>>>>        
>>>>
>>>>>On Tue, 24 Jan 2006, Ed Sweetman wrote:
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>Alan Cox wrote:
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>>>problem.  The problem is that there appears to be two nvidia/amd ata
>>>>>>>>drivers and I'm unsure which I should try using, if i compile both in,
>>>>>>>>which get loaded first (i assume scsi is second to ide) and if i want my
>>>>>>>>pata disks loaded under the new libata drivers, will my cdrom work under
>>>>>>>>them too, or do i still need some sort of regular ide drivers loaded
>>>>>>>>just for cdrom (to use native ata mode for recording access).
>>>>>>>>
>>>>>>>>
>>>>>>>>                
>>>>>>>>
>>>>>>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
>>>>>>>its entirity with code using the newer and cleaner libata logic. There
>>>>>>>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
>>>>>>>are using libata and the additional patch patches still queued
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>>>1.  Atapi is most definitely not supported by libata, right now.
>>>>>>>>
>>>>>>>>
>>>>>>>>                
>>>>>>>>
>>>>>>>It works in the -mm tree.
>>>>>>>
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>Intriguing, when I had no ide chipset compiled in kernel, only libata
>>>>>>drivers, I got no mention at all about my dvd writer.  I even had the
>>>>>>scsi cd driver installed and generic devices, still nothing seemed to
>>>>>>initialize the dvd drive.  It detected the second pata bus but no
>>>>>>devices attached to it.
>>>>>>
>>>>>>this is using the kernel mentioned in the subject header.
>>>>>>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
>>>>>>
>>>>>>Is there anything i can do to give more info to the list to figure out
>>>>>>why my atapi writer is being ignored by pata even when there are no ide
>>>>>>drivers loaded?
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>Currently you need to use libata.atapi_enabled=1
>>>>>(assuming that libata is in the kernel image, not a loadable module).
>>>>>
>>>>>I just built/tested this also, working for me as well.
>>>>>(hard drives, not ATAPI)
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>I assume libata.atapi_enabled=1 is a boot arg, not some structure member
>>>>in the source for the pata driver that i need to set to 1, correct?
>>>>
>>>>        
>>>>
>>>Yes, it's a kernel boot option if libata is in the kernel image.
>>>If libata is a loadable module, just use something like
>>>      modprobe libata atapi_enabled=1
>>>
>>>
>>>      
>>>
>>>>And you just built and tested it, how did you test if the atapi argument
>>>>worked when you then say "not ATAPI" as something you tested?
>>>>
>>>>        
>>>>
>>>Sorry, I mean that I built and booted a kernel with libata/PATA
>>>hard drive (vs. legacy drivers/ide/ PATA support).  I have not
>>>tested ATAPI at all and didn't mean to imply that I had.
>>>
>>>I reported on libata.atapi_enabled=1 based on documentation
>>>and other emails that I have read.
>>>
>>>
>>>      
>>>
>>>>In any case, i'll try out libata.atapi_enabled=1 and see if it detects
>>>>the dvd drive.
>>>>
>>>>        
>>>>
>>>HTH.  Please continue to post any questions or problems.
>>>
>>>      
>>>
>>I Rebooted several times, both setting the option in the kernel boot
>>args and editing the source to have it set by default.  No atapi devices
>>are found/mentioned or even described as not found in dmesg/bootup.   So
>>apparently, on my chipset, the amd/nvidia pata driver is not detecting
>>atapi devices.
>>
>>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
>>0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA
>>Controller (rev f3)
>>0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA
>>Controller (rev f3)
>>
>>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
>>(prog-if 8a [Master SecP PriP])
>>        Subsystem: Unknown device f043:815a
>>        Flags: bus master, 66MHz, fast devsel, latency 0
>>        I/O ports at f000 [size=16]
>>        Capabilities: [44] Power Management version 2
>>
>>
>>the atapi device in question is a plextor px-712A, it's the only device
>>on the secondary channel.
>>    
>>
>
>And this is with using only ATA (libata) drivers in drivers/scsi/
>and not ATA drivers in drivers/ide/, right?
>
>Hm.  I guess we treat this as a bug report for NV ATA/ATAPI then.
>
>I just tested my system with a Plextor PX-712SA drive plus
>booting with libata.atapi_enabled=1 and the driver (not nv)
>sees the ATAPI drive and can read it.
>
>  
>
Indeed, this is with only libata.  I had scsi disk driver and cdrom 
driver compiled in as well, because i assumed that the "low level" 
libata drivers required those scsi interfaces to access the disks and 
atapi devices that are found by libata.  ide isn't even compiled in.

Like i said, i booted with libata.atapi_enabled=1 and that produced 
nothing about cdroms/atapi devices and then I simply set the variable to 
1 in source and recompiled and booted and same problem.

my board is an Asus A8N-E and my plextor is on the PATA controller, not 
the SATA like yours.  Perhaps mine would work too if it was sata, but it 
appears that the pata driver has no provisions for atapi devices yet.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Alan Cox]

> I just tested my system with a Plextor PX-712SA drive plus
> booting with libata.atapi_enabled=1 and the driver (not nv)
> sees the ATAPI drive and can read it.

Parse error ;)

What driver is "the driver" ?

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Mon, 30 Jan 2006, Alan Cox wrote:

> > I just tested my system with a Plextor PX-712SA drive plus
> > booting with libata.atapi_enabled=1 and the driver (not nv)
> > sees the ATAPI drive and can read it.
>
> Parse error ;)
>
> What driver is "the driver" ?

Oh, it was generic in that sentence, just not nv.
I was using libata-core + libata-scsi + ahci + scsi infrastructure.
Maybe I should have said "the drivers."  8;)

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Still using mm kernels here, now on 2.6.16-rc1-mm5.   Still no atapi 
device detection - much less function - when using the libata pata amd 
driver for the nvidia Nforce4 chipset.   I tried, libata.atapi_enabled=1 
and just atapi_enabled=1 in the boot args and nothing was mentioned 
about atapi devices in dmesg.

Is it a known issue with the pata libata drivers that atapi isn't 
working yet? ...  all i've seen is people with sata atapi devices 
chiming in.





>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
>
>
>              
>
>>problem.  The problem is that there appears to be two nvidia/amd ata
>>drivers and I'm unsure which I should try using, if i compile both in,
>>which get loaded first (i assume scsi is second to ide) and if i want my
>>pata disks loaded under the new libata drivers, will my cdrom work under
>>them too, or do i still need some sort of regular ide drivers loaded
>>just for cdrom (to use native ata mode for recording access).

>>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
>its entirity with code using the newer and cleaner libata logic. There
>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
>are using libata and the additional patch patches still queued
>>>              
>>>1.  Atapi is most definitely not supported by libata, right now.

>>>It works in the -mm tree.
>>>              
>Intriguing, when I had no ide chipset compiled in kernel, only libata
>drivers, I got no mention at all about my dvd writer.  I even had the
>scsi cd driver installed and generic devices, still nothing seemed to
>initialize the dvd drive.  It detected the second pata bus but no
>devices attached to it.

>this is using the kernel mentioned in the subject header.
>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.

>Is there anything i can do to give more info to the list to figure out
>why my atapi writer is being ignored by pata even when there are no ide
>drivers loaded?




>>>>>Currently you need to use libata.atapi_enabled=1
>>>>>(assuming that libata is in the kernel image, not a loadable module).
>>>>I just built/tested this also, working for me as well.
>>>>>(hard drives, not ATAPI)
>>>          
>>>I assume libata.atapi_enabled=1 is a boot arg, not some structure member
>>>>in the source for the pata driver that i need to set to 1, correct?
>>        
>Yes, it's a kernel boot option if libata is in the kernel image.
>>>If libata is a loadable module, just use something like
>>>      modprobe libata atapi_enabled=1

>>>      
>And you just built and tested it, how did you test if the atapi argument
>>>>worked when you then say "not ATAPI" as something you tested?
>>        
>Sorry, I mean that I built and booted a kernel with libata/PATA
>>>hard drive (vs. legacy drivers/ide/ PATA support).  I have not
>>>tested ATAPI at all and didn't mean to imply that I had.
>I reported on libata.atapi_enabled=1 based on documentation
>>>and other emails that I have read.

>>>      
>In any case, i'll try out libata.atapi_enabled=1 and see if it detects
>>>>the dvd drive.
>>        
>HTH.  Please continue to post any questions or problems.

>>>
>>I Rebooted several times, both setting the option in the kernel boot
>>args and editing the source to have it set by default.  No atapi devices
>>are found/mentioned or even described as not found in dmesg/bootup.   So
>>apparently, on my chipset, the amd/nvidia pata driver is not detecting
>>atapi devices.
>>
>>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
>>0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA
>>Controller (rev f3)
>>0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA
>>Controller (rev f3)
>>
>>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
>>(prog-if 8a [Master SecP PriP])
>>        Subsystem: Unknown device f043:815a
>>        Flags: bus master, 66MHz, fast devsel, latency 0
>>        I/O ports at f000 [size=16]
>>        Capabilities: [44] Power Management version 2
>the atapi device in question is a plextor px-712A, it's the only device
>>on the secondary channel.
>>    
>>
>
>And this is with using only ATA (libata) drivers in drivers/scsi/
>and not ATA drivers in drivers/ide/, right?
>
>Hm.  I guess we treat this as a bug report for NV ATA/ATAPI then.
>
>I just tested my system with a Plextor PX-712SA drive plus
>booting with libata.atapi_enabled=1 and the driver (not nv)
>sees the ATAPI drive and can read it.
>
>  
>
 >Indeed, this is with only libata.  I had scsi disk driver and cdrom
 >driver compiled in as well, because i assumed that the "low level"
 >libata drivers required those scsi interfaces to access the disks and
 >atapi devices that are found by libata.  ide isn't even compiled in.
 >
 >Like i said, i booted with libata.atapi_enabled=1 and that produced
 >nothing about cdroms/atapi devices and then I simply set the variable 
 >to 1 in source and recompiled and booted and same problem.

 >my board is an Asus A8N-E and my plextor is on the PATA controller, not
 >the SATA like yours.  Perhaps mine would work too if it was sata, but 
 >it appears that the pata driver has no provisions for atapi devices >yet.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Fri, 3 Feb 2006, Ed Sweetman wrote:

>
> Still using mm kernels here, now on 2.6.16-rc1-mm5.   Still no atapi
> device detection - much less function - when using the libata pata amd
> driver for the nvidia Nforce4 chipset.   I tried, libata.atapi_enabled=1
> and just atapi_enabled=1 in the boot args and nothing was mentioned
> about atapi devices in dmesg.
>
> Is it a known issue with the pata libata drivers that atapi isn't
> working yet? ...  all i've seen is people with sata atapi devices
> chiming in.

Agreeing with your paragraph above.  I tested my SATA ATAPI CD/DVD
drive with libata/PATA (ata_piix controller driver) and could not
see the CD/DVD drive.

> >On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> >
> >
> >
> >
> >>problem.  The problem is that there appears to be two nvidia/amd ata
> >>drivers and I'm unsure which I should try using, if i compile both in,
> >>which get loaded first (i assume scsi is second to ide) and if i want my
> >>pata disks loaded under the new libata drivers, will my cdrom work under
> >>them too, or do i still need some sort of regular ide drivers loaded
> >>just for cdrom (to use native ata mode for recording access).
>
> >>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> >its entirity with code using the newer and cleaner libata logic. There
> >is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> >are using libata and the additional patch patches still queued
> >>>
> >>>1.  Atapi is most definitely not supported by libata, right now.
>
> >>>It works in the -mm tree.
> >>>
> >Intriguing, when I had no ide chipset compiled in kernel, only libata
> >drivers, I got no mention at all about my dvd writer.  I even had the
> >scsi cd driver installed and generic devices, still nothing seemed to
> >initialize the dvd drive.  It detected the second pata bus but no
> >devices attached to it.
>
> >this is using the kernel mentioned in the subject header.
> >2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
>
> >Is there anything i can do to give more info to the list to figure out
> >why my atapi writer is being ignored by pata even when there are no ide
> >drivers loaded?
>
>
>
>
> >>>>>Currently you need to use libata.atapi_enabled=1
> >>>>>(assuming that libata is in the kernel image, not a loadable module).
> >>>>I just built/tested this also, working for me as well.
> >>>>>(hard drives, not ATAPI)
> >>>
> >>>I assume libata.atapi_enabled=1 is a boot arg, not some structure member
> >>>>in the source for the pata driver that i need to set to 1, correct?
> >>
> >Yes, it's a kernel boot option if libata is in the kernel image.
> >>>If libata is a loadable module, just use something like
> >>>      modprobe libata atapi_enabled=1
>
> >>>
> >And you just built and tested it, how did you test if the atapi argument
> >>>>worked when you then say "not ATAPI" as something you tested?
> >>
> >Sorry, I mean that I built and booted a kernel with libata/PATA
> >>>hard drive (vs. legacy drivers/ide/ PATA support).  I have not
> >>>tested ATAPI at all and didn't mean to imply that I had.
> >I reported on libata.atapi_enabled=1 based on documentation
> >>>and other emails that I have read.
>
> >>>
> >In any case, i'll try out libata.atapi_enabled=1 and see if it detects
> >>>>the dvd drive.
> >>
> >HTH.  Please continue to post any questions or problems.
>
> >>>
> >>I Rebooted several times, both setting the option in the kernel boot
> >>args and editing the source to have it set by default.  No atapi devices
> >>are found/mentioned or even described as not found in dmesg/bootup.   So
> >>apparently, on my chipset, the amd/nvidia pata driver is not detecting
> >>atapi devices.
> >>
> >>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
> >>0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA
> >>Controller (rev f3)
> >>0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA
> >>Controller (rev f3)
> >>
> >>0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
> >>(prog-if 8a [Master SecP PriP])
> >>        Subsystem: Unknown device f043:815a
> >>        Flags: bus master, 66MHz, fast devsel, latency 0
> >>        I/O ports at f000 [size=16]
> >>        Capabilities: [44] Power Management version 2
> >the atapi device in question is a plextor px-712A, it's the only device
> >>on the secondary channel.
> >>
> >>
> >
> >And this is with using only ATA (libata) drivers in drivers/scsi/
> >and not ATA drivers in drivers/ide/, right?
> >
> >Hm.  I guess we treat this as a bug report for NV ATA/ATAPI then.
> >
> >I just tested my system with a Plextor PX-712SA drive plus
> >booting with libata.atapi_enabled=1 and the driver (not nv)
> >sees the ATAPI drive and can read it.
> >
> >
> >
>  >Indeed, this is with only libata.  I had scsi disk driver and cdrom
>  >driver compiled in as well, because i assumed that the "low level"
>  >libata drivers required those scsi interfaces to access the disks and
>  >atapi devices that are found by libata.  ide isn't even compiled in.
>  >
>  >Like i said, i booted with libata.atapi_enabled=1 and that produced
>  >nothing about cdroms/atapi devices and then I simply set the variable
>  >to 1 in source and recompiled and booted and same problem.
>
>  >my board is an Asus A8N-E and my plextor is on the PATA controller, not
>  >the SATA like yours.  Perhaps mine would work too if it was sata, but
>  >it appears that the pata driver has no provisions for atapi devices >yet.
>
>
>

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Harald Dunkel]

[-- Attachment #1: Type: text/plain, Size: 3112 bytes --]

Hi all,

Randy.Dunlap wrote:
> On Fri, 3 Feb 2006, Ed Sweetman wrote:
> 
> 
> 
> Agreeing with your paragraph above.  I tested my SATA ATAPI CD/DVD
> drive with libata/PATA (ata_piix controller driver) and could not
> see the CD/DVD drive.
> 

Maybe it would be of general interest if somebody could post
how it is _supposed_ to work? Is there a conflict between
ata_piix and piix/mpiix? A short summary could be very helpfull
to identify problems, and to reduce confusion.

I got the impression that the current implementation seems to
be very fragile wrt. the kernel configuration, the module load
sequence, and the attributes to libata.


Hardware (Aopen MZ915-M):

0000:00:00.0 Host bridge: Intel Corporation 915G/P/GV/GL/PL/910GL Processor to I/O Controller (rev 0e)
0000:00:02.0 VGA compatible controller: Intel Corporation 82915G/GV/910GL Express Chipset Family Graphics Controller (rev 0e)
0000:00:02.1 Display controller: Intel Corporation 82915G Express Chipset Family Graphics Controller (rev 0e)
0000:00:1b.0 0403: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) High Definition Audio Controller (rev 04)
0000:00:1c.0 PCI bridge: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) PCI Express Port 1 (rev 04)
0000:00:1c.3 PCI bridge: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) PCI Express Port 4 (rev 04)
0000:00:1d.0 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #1 (rev 04)
0000:00:1d.1 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #2 (rev 04)
0000:00:1d.2 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #3 (rev 04)
0000:00:1d.3 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #4 (rev 04)
0000:00:1d.7 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB2 EHCI Controller (rev 04)
0000:00:1e.0 PCI bridge: Intel Corporation 82801 PCI Bridge (rev d4)
0000:00:1f.0 ISA bridge: Intel Corporation 82801FB/FR (ICH6/ICH6R) LPC Interface Bridge (rev 04)
0000:00:1f.2 IDE interface: Intel Corporation 82801FB/FW (ICH6/ICH6W) SATA Controller (rev 04)
0000:00:1f.3 SMBus: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) SMBus Controller (rev 04)
0000:02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 Gigabit Ethernet Controller (rev 19)
0000:03:00.0 Multimedia video controller: Internext Compression Inc iTVC16 (CX23416) MPEG-2 Encoder (rev 01)

0000:00:00.0 0600: 8086:2580 (rev 0e)
0000:00:02.0 0300: 8086:2582 (rev 0e)
0000:00:02.1 0380: 8086:2782 (rev 0e)
0000:00:1b.0 0403: 8086:2668 (rev 04)
0000:00:1c.0 0604: 8086:2660 (rev 04)
0000:00:1c.3 0604: 8086:2666 (rev 04)
0000:00:1d.0 0c03: 8086:2658 (rev 04)
0000:00:1d.1 0c03: 8086:2659 (rev 04)
0000:00:1d.2 0c03: 8086:265a (rev 04)
0000:00:1d.3 0c03: 8086:265b (rev 04)
0000:00:1d.7 0c03: 8086:265c (rev 04)
0000:00:1e.0 0604: 8086:244e (rev d4)
0000:00:1f.0 0601: 8086:2640 (rev 04)
0000:00:1f.2 0101: 8086:2651 (rev 04)
0000:00:1f.3 0c05: 8086:266a (rev 04)
0000:02:00.0 0200: 11ab:4362 (rev 19)
0000:03:00.0 0400: 4444:0016 (rev 01)


Many thanx

Harri

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: 2.6.16-rc1-mm2 pata driver confusion

[Alan Cox]

On Llu, 2006-02-06 at 20:34 +0100, Harald Dunkel wrote:
> how it is _supposed_ to work? Is there a conflict between
> ata_piix and piix/mpiix? A short summary could be very helpfull
> to identify problems, and to reduce confusion.

MPIIX is totally different PCI identifiers so a different driver. It is
unrelated to any goings on here.

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Ed Sweetman]

2.6.16-rc2 with the libata patch alan cox provided resulted in a system 
which correctly found my atapi drive on the pata bus.  

Also, moving from the mm kernels to 2.6.16-rc2 resulted in my pm timer 
working again. 

So, it may have been all along that libata had supported my atapi 
device, but either the patch when it got merged with mm became broken or 
something else in mm broke it, and something else in mm caused my pm 
timer to stop being used. 

Hopefully more people chime in with issues like this so the bug can be 
tracked down in mm before the offending code makes it's way to a regular 
release. 


Also just to note, the patch to 2.6.16-rc2 alan cox made switches the 
load order of pata and sata from mm, resulting in device name changes.  
I guess now is a good time to start making label'd mounting a priority 
in getting into actual use on my machine.


Ed Sweetman wrote:

> Alan Cox wrote:
>
>> On Llu, 2006-02-06 at 20:34 +0100, Harald Dunkel wrote:
>>  
>>
>>> how it is _supposed_ to work? Is there a conflict between
>>> ata_piix and piix/mpiix? A short summary could be very helpfull
>>> to identify problems, and to reduce confusion.
>>>   
>>
>>
>> MPIIX is totally different PCI identifiers so a different driver. It is
>> unrelated to any goings on here.
>>
>>  
>>
> Ok, but my nforce board works just fine with pata hdd's but doesn't 
> with atapi devices on the pata bus, which is what his message was in 
> reference to.   So forget about piix and such, how are you _supposed_ 
> to get atapi devices seen on the pata drivers (assuming they're 
> functioning at all on the pata drivers) so those that are testing 
> these drivers can make bug reports about them not working, instead of 
> just possibly getting the configuration wrong.
>
>
> No mention of atapi was made in the last patch / status report so i 
> take it You're under the assumption that atapi is _supposed_ to be 
> working in pata (at least the drives should be initialized and brought 
> up). 
>
> So if any info is needed, just mention what it is...  the drivers have 
> been rock solid so far for my sata and pata hdds. I'd like to continue 
> testing with the atapi drive detected because I can't have both ide 
> and pata drivers loaded at the same time, and switching between the 
> two changes device names around. -
> To unsubscribe from this list: send the line "unsubscribe 
> linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Andrew Morton]

Ed Sweetman <safemode@comcast.net> wrote:
>

(eek, top-posting!)

> 2.6.16-rc2 with the libata patch alan cox provided resulted in a system 
> which correctly found my atapi drive on the pata bus.  
> 
> Also, moving from the mm kernels to 2.6.16-rc2 resulted in my pm timer 
> working again. 

I don't recall that.  Even 2.6.16-rc1-mm5?  What about rc1-mm4?

> So, it may have been all along that libata had supported my atapi 
> device, but either the patch when it got merged with mm became broken or 
> something else in mm broke it, and something else in mm caused my pm 
> timer to stop being used. 
> 
> Hopefully more people chime in with issues like this so the bug can be 
> tracked down in mm before the offending code makes it's way to a regular 
> release. 
> 
> 
> Also just to note, the patch to 2.6.16-rc2 alan cox made switches the 
> load order of pata and sata from mm, resulting in device name changes.  

That's bad.

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Matthew Garrett]

Andrew Morton <akpm@osdl.org> wrote:

> That's bad.

Given that libata goes through the scsi layer at the moment, shifting
from the traditional PATA drivers to the libata ones is going to result
in a shift from hdfoo to sdbar. We're not really looking forward to this
from the distribution point of view, though I think the same thing
happened in the past when shifting from the ancient SATA drivers to the
libata ones.

-- 
Matthew Garrett | mjg59-chiark.mail.linux-rutgers.kernel@srcf.ucam.org

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Alan Cox wrote:

>On Llu, 2006-02-06 at 20:34 +0100, Harald Dunkel wrote:
>  
>
>>how it is _supposed_ to work? Is there a conflict between
>>ata_piix and piix/mpiix? A short summary could be very helpfull
>>to identify problems, and to reduce confusion.
>>    
>>
>
>MPIIX is totally different PCI identifiers so a different driver. It is
>unrelated to any goings on here.
>
>  
>
Ok, but my nforce board works just fine with pata hdd's but doesn't with 
atapi devices on the pata bus, which is what his message was in 
reference to.   So forget about piix and such, how are you _supposed_ to 
get atapi devices seen on the pata drivers (assuming they're functioning 
at all on the pata drivers) so those that are testing these drivers can 
make bug reports about them not working, instead of just possibly 
getting the configuration wrong.


No mention of atapi was made in the last patch / status report so i take 
it You're under the assumption that atapi is _supposed_ to be working in 
pata (at least the drives should be initialized and brought up).  


So if any info is needed, just mention what it is...  the drivers have 
been rock solid so far for my sata and pata hdds. I'd like to continue 
testing with the atapi drive detected because I can't have both ide and 
pata drivers loaded at the same time, and switching between the two 
changes device names around. 

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Ed Sweetman]

Matthew Garrett wrote:

>Andrew Morton <akpm@osdl.org> wrote:
>
>  
>
>>That's bad.
>>    
>>
>
>Given that libata goes through the scsi layer at the moment, shifting
>from the traditional PATA drivers to the libata ones is going to result
>in a shift from hdfoo to sdbar. We're not really looking forward to this
>from the distribution point of view, though I think the same thing
>happened in the past when shifting from the ancient SATA drivers to the
>libata ones.
>
>  
>
Well, the badness mentioned above is the swapping of what gets loaded 
first from the testing branch to the upstream patches.   In mm, sata 
gets loaded before pata in libata land.  In alan cox's patches it's the 
reverse.  This results in different device names for the same config 
when switching between mm and release, which is bad, but is a problem 
that can be overcome with the use of labels instead of device names.

Perhaps from a distribution standpoint, moving to a label method of 
describing what gets mounted where would be best, rather than worrying 
about scsi naming schemes or ide ones.  Just think of the fun of a 
system with multiple usb storage devices and such.  
I'm just not sure if grub and the kernel "root=" parameter can handle it.

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Ed Sweetman]

Andrew Morton wrote:

>Ed Sweetman <safemode@comcast.net> wrote:
>  
>
>
>(eek, top-posting!)
>
>  
>
>>2.6.16-rc2 with the libata patch alan cox provided resulted in a system 
>>which correctly found my atapi drive on the pata bus.  
>>
>>Also, moving from the mm kernels to 2.6.16-rc2 resulted in my pm timer 
>>working again. 
>>    
>>
>
>I don't recall that.  Even 2.6.16-rc1-mm5?  What about rc1-mm4?
>
>  
>


yes, 2.6.16-rc1-mm5 was the last kernel I was using before getting out 
of mm. 
I never tried mm4, but mm2 was also affected. 

Re: 2.6.16-rc1-mm2 pata driver confusion + tsc sync issues

[Alan Cox]

On Llu, 2006-02-06 at 22:33 -0500, Ed Sweetman wrote:
> first from the testing branch to the upstream patches.   In mm, sata 
> gets loaded before pata in libata land.  In alan cox's patches it's the 
> reverse.  This results in different device names for the same config 

Thats just down the shape of the Makefile. It'll get resolved in the
merge of the code upstream over time according to the order Jeff puts
them in

> Perhaps from a distribution standpoint, moving to a label method of 
> describing what gets mounted where would be best, rather than worrying

One reason I've not worried about this is I use Fedora so it "just
works"

> about scsi naming schemes or ide ones.  Just think of the fun of a 
> system with multiple usb storage devices and such.  
> I'm just not sure if grub and the kernel "root=" parameter can handle it.

They can't but they don't need too. See the Fedora/Red Hat mkinitrd
script and tools. The 'root' is the initrd and the tools it contains
find the real root by label. No kernel hackery needed.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Randy.Dunlap wrote:

>On Tue, 24 Jan 2006, Ed Sweetman wrote:
>
>  
>
>>Randy.Dunlap wrote:
>>
>>    
>>
>>>On Tue, 24 Jan 2006, Ed Sweetman wrote:
>>>
>>>      
>>>
>>>>Alan Cox wrote:
>>>>
>>>>        
>>>>
>>>>>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>problem.  The problem is that there appears to be two nvidia/amd ata
>>>>>>drivers and I'm unsure which I should try using, if i compile both in,
>>>>>>which get loaded first (i assume scsi is second to ide) and if i want my
>>>>>>pata disks loaded under the new libata drivers, will my cdrom work under
>>>>>>them too, or do i still need some sort of regular ide drivers loaded
>>>>>>just for cdrom (to use native ata mode for recording access).
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
>>>>>its entirity with code using the newer and cleaner libata logic. There
>>>>>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
>>>>>are using libata and the additional patch patches still queued
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>1.  Atapi is most definitely not supported by libata, right now.
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>It works in the -mm tree.
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>Intriguing, when I had no ide chipset compiled in kernel, only libata
>>>>drivers, I got no mention at all about my dvd writer.  I even had the
>>>>scsi cd driver installed and generic devices, still nothing seemed to
>>>>initialize the dvd drive.  It detected the second pata bus but no
>>>>devices attached to it.
>>>>
>>>>this is using the kernel mentioned in the subject header.
>>>>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
>>>>
>>>>Is there anything i can do to give more info to the list to figure out
>>>>why my atapi writer is being ignored by pata even when there are no ide
>>>>drivers loaded?
>>>>
>>>>        
>>>>
>>>Currently you need to use libata.atapi_enabled=1
>>>(assuming that libata is in the kernel image, not a loadable module).
>>>
>>>I just built/tested this also, working for me as well.
>>>(hard drives, not ATAPI)
>>>
>>>
>>>      
>>>
>>I assume libata.atapi_enabled=1 is a boot arg, not some structure member
>>in the source for the pata driver that i need to set to 1, correct?
>>    
>>
>
>Yes, it's a kernel boot option if libata is in the kernel image.
>If libata is a loadable module, just use something like
>	modprobe libata atapi_enabled=1
>
>  
>
>>And you just built and tested it, how did you test if the atapi argument
>>worked when you then say "not ATAPI" as something you tested?
>>    
>>
>
>Sorry, I mean that I built and booted a kernel with libata/PATA
>hard drive (vs. legacy drivers/ide/ PATA support).  I have not
>tested ATAPI at all and didn't mean to imply that I had.
>
>I reported on libata.atapi_enabled=1 based on documentation
>and other emails that I have read.
>
>  
>
>>In any case, i'll try out libata.atapi_enabled=1 and see if it detects
>>the dvd drive.
>>    
>>
>
>HTH.  Please continue to post any questions or problems.
>  
>

I Rebooted several times, both setting the option in the kernel boot 
args and editing the source to have it set by default.  No atapi devices 
are found/mentioned or even described as not found in dmesg/bootup.   So 
apparently, on my chipset, the amd/nvidia pata driver is not detecting 
atapi devices.

0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2)
0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA 
Controller (rev f3)
0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA 
Controller (rev f3)

0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2) 
(prog-if 8a [Master SecP PriP])
        Subsystem: Unknown device f043:815a
        Flags: bus master, 66MHz, fast devsel, latency 0
        I/O ports at f000 [size=16]
        Capabilities: [44] Power Management version 2



the atapi device in question is a plextor px-712A, it's the only device 
on the secondary channel.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Tue, 24 Jan 2006, Ed Sweetman wrote:

> Randy.Dunlap wrote:
>
> >On Tue, 24 Jan 2006, Ed Sweetman wrote:
> >
> >>Alan Cox wrote:
> >>
> >>>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> >>>
> >>>
> >>>>problem.  The problem is that there appears to be two nvidia/amd ata
> >>>>drivers and I'm unsure which I should try using, if i compile both in,
> >>>>which get loaded first (i assume scsi is second to ide) and if i want my
> >>>>pata disks loaded under the new libata drivers, will my cdrom work under
> >>>>them too, or do i still need some sort of regular ide drivers loaded
> >>>>just for cdrom (to use native ata mode for recording access).
> >>>>
> >>>>
> >>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> >>>its entirity with code using the newer and cleaner libata logic. There
> >>>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> >>>are using libata and the additional patch patches still queued
> >>>
> >>>
> >>>>1.  Atapi is most definitely not supported by libata, right now.
> >>>>
> >>>>
> >>>It works in the -mm tree.
> >>>
> >>>
> >>Intriguing, when I had no ide chipset compiled in kernel, only libata
> >>drivers, I got no mention at all about my dvd writer.  I even had the
> >>scsi cd driver installed and generic devices, still nothing seemed to
> >>initialize the dvd drive.  It detected the second pata bus but no
> >>devices attached to it.
> >>
> >>this is using the kernel mentioned in the subject header.
> >>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
> >>
> >>Is there anything i can do to give more info to the list to figure out
> >>why my atapi writer is being ignored by pata even when there are no ide
> >>drivers loaded?
> >>
> >
> >Currently you need to use libata.atapi_enabled=1
> >(assuming that libata is in the kernel image, not a loadable module).
> >
> >I just built/tested this also, working for me as well.
> >(hard drives, not ATAPI)
> >
> >
> I assume libata.atapi_enabled=1 is a boot arg, not some structure member
> in the source for the pata driver that i need to set to 1, correct?

Yes, it's a kernel boot option if libata is in the kernel image.
If libata is a loadable module, just use something like
	modprobe libata atapi_enabled=1

> And you just built and tested it, how did you test if the atapi argument
> worked when you then say "not ATAPI" as something you tested?

Sorry, I mean that I built and booted a kernel with libata/PATA
hard drive (vs. legacy drivers/ide/ PATA support).  I have not
tested ATAPI at all and didn't mean to imply that I had.

I reported on libata.atapi_enabled=1 based on documentation
and other emails that I have read.

> In any case, i'll try out libata.atapi_enabled=1 and see if it detects
> the dvd drive.

HTH.  Please continue to post any questions or problems.
-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Randy.Dunlap wrote:

>On Tue, 24 Jan 2006, Ed Sweetman wrote:
>
>  
>
>>Alan Cox wrote:
>>
>>    
>>
>>>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
>>>
>>>
>>>      
>>>
>>>>problem.  The problem is that there appears to be two nvidia/amd ata
>>>>drivers and I'm unsure which I should try using, if i compile both in,
>>>>which get loaded first (i assume scsi is second to ide) and if i want my
>>>>pata disks loaded under the new libata drivers, will my cdrom work under
>>>>them too, or do i still need some sort of regular ide drivers loaded
>>>>just for cdrom (to use native ata mode for recording access).
>>>>
>>>>
>>>>        
>>>>
>>>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
>>>its entirity with code using the newer and cleaner libata logic. There
>>>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
>>>are using libata and the additional patch patches still queued
>>>
>>>
>>>      
>>>
>>>>1.  Atapi is most definitely not supported by libata, right now.
>>>>
>>>>
>>>>        
>>>>
>>>It works in the -mm tree.
>>>
>>>
>>>      
>>>
>>Intriguing, when I had no ide chipset compiled in kernel, only libata
>>drivers, I got no mention at all about my dvd writer.  I even had the
>>scsi cd driver installed and generic devices, still nothing seemed to
>>initialize the dvd drive.  It detected the second pata bus but no
>>devices attached to it.
>>
>>this is using the kernel mentioned in the subject header.
>>2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
>>
>>Is there anything i can do to give more info to the list to figure out
>>why my atapi writer is being ignored by pata even when there are no ide
>>drivers loaded?
>>    
>>
>
>Currently you need to use libata.atapi_enabled=1
>(assuming that libata is in the kernel image, not a loadable module).
>
>I just built/tested this also, working for me as well.
>(hard drives, not ATAPI)
>  
>
I assume libata.atapi_enabled=1 is a boot arg, not some structure member 
in the source for the pata driver that i need to set to 1, correct? 

And you just built and tested it, how did you test if the atapi argument 
worked when you then say "not ATAPI" as something you tested?

In any case, i'll try out libata.atapi_enabled=1 and see if it detects 
the dvd drive.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Tue, 24 Jan 2006, Ed Sweetman wrote:

> Alan Cox wrote:
>
> >On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> >
> >
> >>problem.  The problem is that there appears to be two nvidia/amd ata
> >>drivers and I'm unsure which I should try using, if i compile both in,
> >>which get loaded first (i assume scsi is second to ide) and if i want my
> >>pata disks loaded under the new libata drivers, will my cdrom work under
> >>them too, or do i still need some sort of regular ide drivers loaded
> >>just for cdrom (to use native ata mode for recording access).
> >>
> >>
> >
> >The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> >its entirity with code using the newer and cleaner libata logic. There
> >is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> >are using libata and the additional patch patches still queued
> >
> >
>
> >>1.  Atapi is most definitely not supported by libata, right now.
> >>
> >>
> >
> >It works in the -mm tree.
> >
> >
> Intriguing, when I had no ide chipset compiled in kernel, only libata
> drivers, I got no mention at all about my dvd writer.  I even had the
> scsi cd driver installed and generic devices, still nothing seemed to
> initialize the dvd drive.  It detected the second pata bus but no
> devices attached to it.
>
> this is using the kernel mentioned in the subject header.
> 2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.
>
> Is there anything i can do to give more info to the list to figure out
> why my atapi writer is being ignored by pata even when there are no ide
> drivers loaded?

Currently you need to use libata.atapi_enabled=1
(assuming that libata is in the kernel image, not a loadable module).

I just built/tested this also, working for me as well.
(hard drives, not ATAPI)

> >>4.  moving to pata libata drivers _will_ change the enumeration of your
> >>sata devices, it seems that pata is initialized first, so when setting
> >>up your fstab entries and grub, you'll have to take into account how
> >>many pata devices you have and offset your current sata device names by
> >>that amount.
> >>
> >>
> >
> >Or use labels. As you move into the world of hot pluggable hardware it
> >becomes more and more impractical to guarantee drive ordering by name.
> >
> >You can mix and match the drivers providing you don't try and load both
> >libata and old ide drives for the same chip. Even then it should fail
> >correctly with one of them reporting resources unavailable.
> >
> >In fact I do this all the time when debugging so I've got a stable disk
> >for debug work and a devel disk.
> >
> >Alan
> >
> >
>
> I'm not familiar with labeling ...will have to look into it, since
> borking a kernel after changing the drivers can result in a non-bootable
> system depending on how the partitions are setup across the devices.
> (since you'd have to change them in fstab and such before a reboot).  A
> way around that would be very useful.

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

Alan Cox wrote:

>On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
>  
>
>>problem.  The problem is that there appears to be two nvidia/amd ata 
>>drivers and I'm unsure which I should try using, if i compile both in, 
>>which get loaded first (i assume scsi is second to ide) and if i want my 
>>pata disks loaded under the new libata drivers, will my cdrom work under 
>>them too, or do i still need some sort of regular ide drivers loaded 
>>just for cdrom (to use native ata mode for recording access).  
>>    
>>
>
>The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
>its entirity with code using the newer and cleaner libata logic. There
>is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
>are using libata and the additional patch patches still queued
>  
>

>>1.  Atapi is most definitely not supported by libata, right now.
>>    
>>
>
>It works in the -mm tree.
>  
>
Intriguing, when I had no ide chipset compiled in kernel, only libata 
drivers, I got no mention at all about my dvd writer.  I even had the 
scsi cd driver installed and generic devices, still nothing seemed to 
initialize the dvd drive.  It detected the second pata bus but no 
devices attached to it. 

this is using the kernel mentioned in the subject header.  
2.6.16-rc1-mm2.  using the amd/nvidia drivers for pata and sata.

Is there anything i can do to give more info to the list to figure out 
why my atapi writer is being ignored by pata even when there are no ide 
drivers loaded?

>>4.  moving to pata libata drivers _will_ change the enumeration of your 
>>sata devices, it seems that pata is initialized first, so when setting 
>>up your fstab entries and grub, you'll have to take into account how 
>>many pata devices you have and offset your current sata device names by 
>>that amount.
>>    
>>
>
>Or use labels. As you move into the world of hot pluggable hardware it
>becomes more and more impractical to guarantee drive ordering by name.
>
>You can mix and match the drivers providing you don't try and load both
>libata and old ide drives for the same chip. Even then it should fail
>correctly with one of them reporting resources unavailable.
>
>In fact I do this all the time when debugging so I've got a stable disk
>for debug work and a devel disk.
>
>Alan
>  
>

I'm not familiar with labeling ...will have to look into it, since 
borking a kernel after changing the drivers can result in a non-bootable 
system depending on how the partitions are setup across the devices. 
(since you'd have to change them in fstab and such before a reboot).  A 
way around that would be very useful.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Alan Cox]

On Maw, 2006-01-24 at 09:08 -0800, Randy.Dunlap wrote:
> > and while I'm looking at the config menu, why do both
> > Compaq Triflex and Intel PATA MPIIX say (Raving Lunatic)?
> 
> Lots of them say Raving Lunatic.  Are all of these Alan's libata
> patches?

They are a subset of them. The full patch now covers all the PCI devices
that are not platform specific except CMD640B if I counted right. It
also supports generic ISA, ISAPnP, PCMCIA and some VLB devices. 

Needless to say not all of this works and less of it has been heavily
tested yet.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Alan Cox]

On Maw, 2006-01-24 at 09:05 -0800, Randy.Dunlap wrote:
> What is "MPIIX" anyway?

MPIIX was an early intel mobile chip. Its a PCI bridge, glue and the
like for pentium laptops with a built in PIO ATA controller.

> and while I'm looking at the config menu, why do both
> Compaq Triflex and Intel PATA MPIIX say (Raving Lunatic)?

So people don't casually select them. The current MPIIX driver will
remove the raving lunatic once it gets upstream.

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Tue, 24 Jan 2006, Randy.Dunlap wrote:

> On Tue, 24 Jan 2006, Alan Cox wrote:
>
> > On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> > > problem.  The problem is that there appears to be two nvidia/amd ata
> > > drivers and I'm unsure which I should try using, if i compile both in,
> > > which get loaded first (i assume scsi is second to ide) and if i want my
> > > pata disks loaded under the new libata drivers, will my cdrom work under
> > > them too, or do i still need some sort of regular ide drivers loaded
> > > just for cdrom (to use native ata mode for recording access).
> >
> > The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> > its entirity with code using the newer and cleaner libata logic. There
> > is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> > are using libata and the additional patch patches still queued.
>
> What is "MPIIX" anyway?
>
> and while I'm looking at the config menu, why do both
> Compaq Triflex and Intel PATA MPIIX say (Raving Lunatic)?

Lots of them say Raving Lunatic.  Are all of these Alan's libata
patches?

> > > 1.  Atapi is most definitely not supported by libata, right now.
> >
> > It works in the -mm tree.
> >
> > > 4.  moving to pata libata drivers _will_ change the enumeration of your
> > > sata devices, it seems that pata is initialized first, so when setting
> > > up your fstab entries and grub, you'll have to take into account how
> > > many pata devices you have and offset your current sata device names by
> > > that amount.
> >
> > Or use labels. As you move into the world of hot pluggable hardware it
> > becomes more and more impractical to guarantee drive ordering by name.
> >
> > You can mix and match the drivers providing you don't try and load both
> > libata and old ide drives for the same chip. Even then it should fail
> > correctly with one of them reporting resources unavailable.
> >
> > In fact I do this all the time when debugging so I've got a stable disk
> > for debug work and a devel disk.
>
>

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Randy.Dunlap]

On Tue, 24 Jan 2006, Alan Cox wrote:

> On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> > problem.  The problem is that there appears to be two nvidia/amd ata
> > drivers and I'm unsure which I should try using, if i compile both in,
> > which get loaded first (i assume scsi is second to ide) and if i want my
> > pata disks loaded under the new libata drivers, will my cdrom work under
> > them too, or do i still need some sort of regular ide drivers loaded
> > just for cdrom (to use native ata mode for recording access).
>
> The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
> its entirity with code using the newer and cleaner libata logic. There
> is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
> are using libata and the additional patch patches still queued.

What is "MPIIX" anyway?

and while I'm looking at the config menu, why do both
Compaq Triflex and Intel PATA MPIIX say (Raving Lunatic)?

> > 1.  Atapi is most definitely not supported by libata, right now.
>
> It works in the -mm tree.
>
> > 4.  moving to pata libata drivers _will_ change the enumeration of your
> > sata devices, it seems that pata is initialized first, so when setting
> > up your fstab entries and grub, you'll have to take into account how
> > many pata devices you have and offset your current sata device names by
> > that amount.
>
> Or use labels. As you move into the world of hot pluggable hardware it
> becomes more and more impractical to guarantee drive ordering by name.
>
> You can mix and match the drivers providing you don't try and load both
> libata and old ide drives for the same chip. Even then it should fail
> correctly with one of them reporting resources unavailable.
>
> In fact I do this all the time when debugging so I've got a stable disk
> for debug work and a devel disk.

-- 
~Randy

Re: 2.6.16-rc1-mm2 pata driver confusion

[Alan Cox]

On Maw, 2006-01-24 at 01:43 -0500, Ed Sweetman wrote:
> problem.  The problem is that there appears to be two nvidia/amd ata 
> drivers and I'm unsure which I should try using, if i compile both in, 
> which get loaded first (i assume scsi is second to ide) and if i want my 
> pata disks loaded under the new libata drivers, will my cdrom work under 
> them too, or do i still need some sort of regular ide drivers loaded 
> just for cdrom (to use native ata mode for recording access).  

The goal of the drivers/scsi/pata_* drivers is to replace drivers/ide in
its entirity with code using the newer and cleaner libata logic. There
is still much to do but my SIL680, SiS, Intel MPIIX, AMD and VIA boxes
are using libata and the additional patch patches still queued.

> 1.  Atapi is most definitely not supported by libata, right now.

It works in the -mm tree.

> 4.  moving to pata libata drivers _will_ change the enumeration of your 
> sata devices, it seems that pata is initialized first, so when setting 
> up your fstab entries and grub, you'll have to take into account how 
> many pata devices you have and offset your current sata device names by 
> that amount.

Or use labels. As you move into the world of hot pluggable hardware it
becomes more and more impractical to guarantee drive ordering by name.

You can mix and match the drivers providing you don't try and load both
libata and old ide drives for the same chip. Even then it should fail
correctly with one of them reporting resources unavailable.

In fact I do this all the time when debugging so I've got a stable disk
for debug work and a devel disk.

Alan

Re: 2.6.16-rc1-mm2 pata driver confusion

[Jeff Garzik]

On Tue, Jan 24, 2006 at 01:43:20AM -0500, Ed Sweetman wrote:
> I have an nforce4 based motherboard.  Currently i'm using the amd/nvidia 
> driver under the normal ide,ata driver section (2.6.14). 
> 
> It appears that the new ata code is hiding under scsi/sata drivers, 
> including apparently pata code.  This alone reads confusing, pata 

libata PATA support is under development.  Use only if you feel lucky.
Really lucky.  I mean, really really lucky.


> 1.  Atapi is most definitely not supported by libata, right now.

Not true.


> 2. whether libata sets the controller up better or not, ide cdroms MUST 
> be loaded before libata is or the ide controller will be detected as 
> "already in use" and the cdrom drivers wont have any device to attach 
> to, since unlike scsi drivers, ide drivers dont probe the hardware on 
> controllers to see if any driver has claimed them.

Either use drivers/ide or libata for PATA, not both.


> 3. For hdd's alone, the pata libata + sata drivers are a "complete" 
> replacement for the ide drivers and thus, if you dont have atapi 
> devices, you dont need to compile in ide support.

Again, ATAPI works just fine.


> 4.  moving to pata libata drivers _will_ change the enumeration of your 
> sata devices, it seems that pata is initialized first, so when setting 
> up your fstab entries and grub, you'll have to take into account how 
> many pata devices you have and offset your current sata device names by 
> that amount.

Enumeration of devices depends on which driver is loaded first.
Check your /etc/modprobe.conf.

	Jeff

2.6.16-rc1-mm2 pata driver confusion

[Ed Sweetman]

I have an nforce4 based motherboard.  Currently i'm using the amd/nvidia 
driver under the normal ide,ata driver section (2.6.14). 

It appears that the new ata code is hiding under scsi/sata drivers, 
including apparently pata code.  This alone reads confusing, pata 
drivers under the sata driver section, but that's not really the 
problem.  The problem is that there appears to be two nvidia/amd ata 
drivers and I'm unsure which I should try using, if i compile both in, 
which get loaded first (i assume scsi is second to ide) and if i want my 
pata disks loaded under the new libata drivers, will my cdrom work under 
them too, or do i still need some sort of regular ide drivers loaded 
just for cdrom (to use native ata mode for recording access).  

I couldn't find anything to explain how to deal with the new libata when 
it apparently overlaps functionality with the old ata/ide code in the 
Documentation directory, nor is it apparent if atapi is supported under 
the new libata code, or if either should be loaded first or not for best 
performance (if it matters). Or if they can both be loaded at the same 
time at all.


I have booted numerous configurations and have found the following to be 
true.

1.  Atapi is most definitely not supported by libata, right now.
2. whether libata sets the controller up better or not, ide cdroms MUST 
be loaded before libata is or the ide controller will be detected as 
"already in use" and the cdrom drivers wont have any device to attach 
to, since unlike scsi drivers, ide drivers dont probe the hardware on 
controllers to see if any driver has claimed them.
3. For hdd's alone, the pata libata + sata drivers are a "complete" 
replacement for the ide drivers and thus, if you dont have atapi 
devices, you dont need to compile in ide support.
4.  moving to pata libata drivers _will_ change the enumeration of your 
sata devices, it seems that pata is initialized first, so when setting 
up your fstab entries and grub, you'll have to take into account how 
many pata devices you have and offset your current sata device names by 
that amount.