Bug 205331

Created attachment 98535 [details]
Log made under SuSE 10.2 alpha4

Created attachment 98536 [details]
Logs made under knoppoix 5.0.1

Detection of hard disk and CD/DVD drive works fine in SuSE Linux 9.1 !

Detection of hard disk and CD/DVD drive works fine in SuSE Linux 9.2 Prof !

Detection of hard disk and CD/DVD drive works fine in Fedora 5 !
Have also successfully installed Fedora, but:

[root@localhost ~]# hdparm -tT /dev/hda

/dev/hda:
 Timing cached reads:   2104 MB in  2.00 seconds = 1053.01 MB/sec
 Timing buffered disk reads:    6 MB in  4.06 seconds =   1.48 MB/sec
[root@localhost ~]# hdparm -i /dev/hda

/dev/hda:

 Model=SAMSUNG MP0804H, FwRev=YS200-06, SerialNo=S042J10L788149
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
 RawCHS=16383/16/63, TrkSize=34902, SectSize=554, ECCbytes=4
 BuffType=DualPortCache, BuffSize=8192kB, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=156368016
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4
 DMA modes:  mdma0 mdma1 mdma2
 UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
 AdvancedPM=yes: disabled (255) WriteCache=enabled
 Drive conforms to: (null):  ATA/ATAPI-1 ATA/ATAPI-2 ATA/ATAPI-3 ATA/ATAPI-4 ATA/ATAPI-5 ATA/ATAPI-6 ATA/ATAPI-7

Copying of 81 MB file takes about 2 min !

The kernel is unable to detect the IDE hardware...

See boot.msg file in comment #1 (osuse10.2a4):

<6>PCI: Ignoring BAR0-3 of IDE controller 0000:00:1f.2
[...]
<6>Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
<6>ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
<5>SCSI subsystem initialized
<6>st: Version 20050830, fixed bufsize 32768, s/g segs 256
<6>BIOS EDD facility v0.16 2004-Jun-25, 1 devices found
<6>Non-volatile memory driver v1.2
<6>NET: Registered protocol family 17
<6>loop: loaded (max 8 devices)
<6>powernow: This module only works with AMD K7 CPUs
<7>libata version 2.00 loaded.
<7>ahci 0000:00:1f.2: version 2.0
<4>ahci: probe of 0000:00:1f.2 failed with error -12
<7>ata_piix 0000:00:1f.2: version 2.00
<6>ata_piix 0000:00:1f.2: MAP [ XX XX XX XX ]
<3>ata_piix 0000:00:1f.2: invalid MAP value 1
<4>PCI: Enabling device 0000:00:1f.2 (0000 -> 0001)
<7>PCI: Setting latency timer of device 0000:00:1f.2 to 64
<6>ata1: SATA max UDMA/133 cmd 0x1F0 ctl 0x3F6 bmdma 0x18B0 irq 14
<6>scsi0 : ata_piix
<6>ata2: SATA max UDMA/133 cmd 0x170 ctl 0x376 bmdma 0x18B8 irq 15
<6>scsi1 : ata_piix
<4>ATA: abnormal status 0x7F on port 0x177

and compare with dmesg.txt from comment #2 (knopix):

PCI: Ignoring BAR0-3 of IDE controller 0000:00:1f.2
[...]
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
Probing IDE interface ide0...
hda: SAMSUNG MP0804H, ATA DISK drive
hdb: HL-DT-ST DVDRAM GMA-4082N, ATAPI CD/DVD-ROM drive
ide1: I/O resource 0x170-0x177 not free.
ide1: ports already in use, skipping probe
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: max request size: 512KiB
hda: 156368016 sectors (80060 MB) w/8192KiB Cache, CHS=16383/255/63
hda: cache flushes supported
 hda: hda1 hda2 < hda5 hda6 hda7 hda8 >
hdb: ATAPI 24X DVD-ROM DVD-R-RAM CD-R/RW drive, 2048kB Cache
Uniform CD-ROM driver Revision: 3.20
libata version 1.20 loaded.
ahci 0000:00:1f.2: version 1.2
ACPI: PCI Interrupt 0000:00:1f.2[B] -> GSI 17 (level, low) -> IRQ 16
ACPI: PCI interrupt for device 0000:00:1f.2 disabled
ahci: probe of 0000:00:1f.2 failed with error -12
ata_piix 0000:00:1f.2: version 1.05
ata_piix 0000:00:1f.2: MAP [ XX XX XX XX ]
ata_piix 0000:00:1f.2: invalid MAP value 1
ACPI: PCI Interrupt 0000:00:1f.2[B] -> GSI 17 (level, low) -> IRQ 16
ata: 0x1f0 IDE port busy
PCI: Setting latency timer of device 0000:00:1f.2 to 64
ata1: SATA max UDMA/133 cmd 0x170 ctl 0x376 bmdma 0x18B8 irq 15
ata1: SATA port has no device.
scsi0 : ata_piix

Resassign to kernel maintainers. see above comment.

We'll probably upgrade the libata stack to the latest version, which comes with an improved error recovery. That should fix this issue, too.

Created attachment 99878 [details]
Messages from kernel 2.6.18-1.2189 (Fedora 5)

Have now installed the last kernel (2.6.18-1.2189) under Fedora 5
but the disk performance is not better :-( Even after tuning with
"hdparm -c3 -a64 -u1 /dev/hda" i get:

[root@localhost tmp]# hdparm -Tt /dev/hda

/dev/hda:
 Timing cached reads:   2040 MB in  2.00 seconds = 1020.39 MB/sec
 Timing buffered disk reads:    8 MB in  3.39 seconds =   2.36 MB/sec

Have tryed 10.2b1 today. Still the same error message :-(

Karsten Merker (from BoLUG mailinglist), sent his experiences with the
Notebook to me (here my partialy translation):

"The hardware setup of the notebook is some uncommon. Both the hard disk and
the CD/DVD-ROM drive are normal (P-ATA) IDE/ATAPI devices and not, SATA
models. On the PCI bus appears however only SATA CONTROLLER:

00:1f.2 IDE interface: Intel Corporation 82801FBM (ICH6M)
  SATA Controller (rev 03)

The first suspicion was, there is a IDE-to-SATA bridge which connects
the devices with SATA controller. The installation procedure load
the drivers for SATA controller, but the driver cannot find any
devices connected.

According to data sheet for the ICH6M, the chip has also a "classical" P-ATA/IDE-channel, that should be available as PCI DEVICE 00:1 f.1.
but is not visible on the PCI bus. The drives (hard disk and CD/DVD)
are physical attached to IDE-chanel, although the associated PCI DEVICE
is faded out and only the SATA part of the chip is visible as PCI DEVICE.
[...]
The piix driver, normally responsible for IDE devices, feels itself
due to the faded out PCI DEVICE not responsible. This problem could be
however be solved with one small Kernel Patch, that admits the piix driver
those appropriate PCI IDs. The Patch was published on the Linux IDE mailing
list:

http://marc.theaimsgroup.com/?l=linux-ide&m=114317385718285&w=2

With kernel 2.6.17 and these path works the piix driver in dma
mode acceptable"

I try to test it the next days.
And here is the original german text from Karsten (maybe someone
could translate it better then me :-| ):

"Der Versuch, mittels der aktuellen Debian-Installer-Beta
Debian Etch (derzeit noch Testing) zu installieren, scheiterte
daran, dass der Installer keine Festplatte fand, obwohl die Platte
mit der Knoppix ansprechbar war. Wie sich dann herausgestellt
hat, ist das Hardware-Setup in diesem Notebook wohl etwas
ungewöhnlich. Sowohl die Festplatte als auch das CDROM-Laufwerk
sind normale (P-ATA) IDE/ATAPI-Laufwerke, keine SATA-Modelle.
Auf dem PCI-Bus erscheint aber ausschliesslich ein
SATA-Controller:

00:1f.2 IDE interface: Intel Corporation 82801FBM (ICH6M) SATA
Controller (rev 03)

Der erste Verdacht war, dass hier vielleicht IDE-Laufwerke mit
einer IDE-zu-SATA-Bridge an einem SATA-Controller angeschlossen
sind. Der Debian-Installer lädt ordnungsgemäß die zu diesem
SATA-Controller gehörenden Treibermodule, findet aber keine daran
angeschlossenen Platten. 

Laut Datenblatt zum ICH6M hat der Chip auch einen "klassischen"
P-ATA/IDE-Kanal, der eigentlich als PCI-Device 00:1f.1 vorhanden
sein müsste, auf dem Bus hier aber nicht sichtbar ist. Wie sich
herausgestellt hat, sind die Laufwerke physikalisch an diesem
IDE-Kanal angeschlossen, obwohl das zugehörige PCI-Device
ausgeblendet und nur der SATA-Teil des Chips als PCI-Device
sichtbar ist. Bei der Knoppix erfolgte der Zugriff auf die Platte
anscheinend mit dem alten Standard-IDE-Treiber über die
Legacy-ISA-Register-Emulation und nicht über das PCI-Device,
weshalb die Laufwerke dort erreichbar waren.

Letztendlich habe ich das Problem so gelöst, dass ich zunächst
ein Debian Sarge (Stable) installiert und später ein dist-upgrade
auf Etch ausgeführt habe, da der Sarge-Installer den
Legacy-ISA-IDE-Treiber verwendet. Leider ist mit diesem kein
DMA-Betrieb möglich, was die Zugriffe auf Platte und CDROM extrem
zäh macht. Der piix-Treiber, der normalerweise bei
Intel-Chipsätzen für den IDE-Controller zum Einsatz kommt, fühlt
sich aufgrund des ausgeblendeten PCI-Devices für den IDE-Betrieb
nicht "zuständig". Dieses Problem läßt sich jedoch mit einem
kleinen Kernel-Patch lösen, der dem piix-Treiber die
entsprechenden PCI-IDs bekannt macht. Ein entsprechender Patch
wurde auf der Linux-IDE-Mailingliste publiziert:

http://marc.theaimsgroup.com/?l=linux-ide&m=114317385718285&w=2

Mit Kernel 2.6.17 und diesem Patch funktioniert der DMA-Betrieb
mittels des piix-Treibers einwandfrei."

The Bug apply also to Beta 1, so i change the Field "Found in version"
to "Beta 1" and set the priority higher

Hello,

It seems your ich6m is reporting 11b for MAP value, which, if I'm not mistaken, is invalid according to the h/w doc.  Can you test the attached patch and report dmesg?

Created attachment 103922 [details]
MAP 11b for ich6m

Created attachment 104035 [details]
dmesg output from patched kernel

I have currently kernel 2.6.18-1.2200 (patched for
fedora core 5) installed. I have made the changes
im ata_piix.c manualy because the file is on
different place (drivers/scsi/ata_piix.c).

Should i try it with vanila kernel ?

Created attachment 104040 [details]
dmesg output after patch metioned from Karsten

I have now applied a patch mentioned earlier form Karsten and the hard disk performance is better.
I cannot say, if detecting of hard disk and DVD-ROM
works properly but the notebook is now usable:

[root@localhost ~]# /sbin/hdparm -tT /dev/hda

/dev/hda:
 Timing cached reads:   2332 MB in  2.00 seconds = 1165.56 MB/sec
 Timing buffered disk reads:   96 MB in  3.02 seconds =  31.77 MB/sec

Nope, sorry, I was looking at the wrong kernel (libata devel tree).  The bug you've filed is already fixed in #linus by the following commit.

commit c6446a4cdadaf411bafe1565e9fa7666f3c2fe95
Author: Tejun Heo <htejun@gmail.com>
Date:   Mon Oct 9 13:23:58 2006 +0900

    [PATCH] ata_piix: allow 01b MAP for both ICH6M and ICH7M
    
    ICH7M was separated from ICH6M to allow undocumented MAP value 01b
    which was spotted on an ASUS notebook.  However, there is also
    notebooks with MAP value 01b on ICH6M.  This patch re-merges ICH6M and
    ICH7M entries and allows MAP value 01b for both.
    
    This problem has been reported and initial patch provided by Jonathan
    Dieter.
    
    Signed-off-by: Tejun Heo <htejun@gmail.com>
    Cc: Jonathan Dieter <jdieter@gmail.com>
    Cc: Tom Deblauwe <tom.deblauwe@telenet.be>
    Signed-off-by: Jeff Garzik <jeff@garzik.org>

I'm attaching a patch rebased against v2.6.18.  Please report the result.

Created attachment 104056 [details]
ich6m map 01b fix

@ Tejun Heo

The patch you mentioned was commited on Oct 9. If the kernel
on 10.2b1 (boot+installation procedure) contain the patch,
then it will not work. I have tested 10.2b1 on Oct 27 (see above).

Created attachment 104207 [details]
dmesg after patch "ich6m map 01b fix"

No better hard disk performance:

[root@localhost]# /sbin/hdparm -tT /dev/hda 
/dev/hda:
 Timing cached reads:   2216 MB in  2.00 seconds = 1107.71 MB/sec
 Timing buffered disk reads:    4 MB in  3.01 seconds =   1.33 MB/sec

:-(

Okay, two things.

* The patch I posted was against vanialla 2.6.18 and doesn't apply to suse 10.2 kernel (sorry, still setting up devel env).  Have you adjusted the patch yourself?  If so, please post the modified patch or patched ata_piix.c.  Kernel is still reporting invalid MAP 1 which is just impossible with the patch applied.

* Please post the result of `hwinfo'.  It seems like generic IDE driver is grabbing ich6m controller which it shouldn't.  The correct driver for your controller is ata_piix and your harddrive should appear as /dev/sda and your cdrom as /dev/sr0.

Thanks.

Created attachment 104388 [details]
s. comment

* I have applied the pahch with the "patch" utility.
In attached are the original and patched ata_piix.c and the diff after original ata_piix.c was patched.

* Sorry, i'm here on fedora. hwinfo is not available :-(
But in /proc you can see that legacy ide drivers are 
used:

[root@localhost scsi]# cat /proc/ide/drivers
ide-disk version 1.18
ide-floppy version 0.99.newide
ide-cdrom version 4.61

[root@localhost scsi]# cat /proc/ide/hda/driver
ide-disk version 1.18
[root@localhost scsi]# cat /proc/ide/hdb/driver
ide-cdrom version 4.61

[root@localhost scsi]# cat /proc/scsi/scsi
Attached devices:

The patch and source files look correct, but it just can't work that way.  Can you please add the following the the patched driver and post dmesg?

In ata_piix.c::piix_init_sata_map(), right above dev_printk(KERN_INFO, &pdev->dev, "MAP [");

    printk("XXX map_value=0x%x map_db->mask=0x%x ich6m_map_db=%p map_db=%p\n",
            map_value, map_db->mask, ich6m_map_db, map_db);

Thanks.

Created attachment 104802 [details]
dmesg with printk

I have switched to Fedora 6 but the problem is still there. The dmesg output was generate from the old
ferdora 5 kernel (2.6.18)

Big step towards: 102b2 detect my hard disk (correctly ?) as /dev/sda but DVD-ROM drive is still not recognized. I have downloded all iso images
tonight and try to install current beta via smb. 

Created attachment 105250 [details]
output of hwinfo and dmesg

After some trouble i have finally installed 102b2 :-) The ATA hard disk appears as scsi device /dev/sda. But CD/DVD-ROM is still not
detected. Some ideas ?

Sorry but debugging info upto this point seems a bot contradicting.  Can you please test the next beta when it comes out and report if it's still broken?  It will contain all the upstream fixes on the area and AFAICS should fix your detection problem.  Thanks.

Hannes, both MAP 01b addition and polling detection are currently in libata development tree.  My suggestion is to import libata-dev#upstream.  There have been quite a few detection related bug fixes lately.

Created attachment 106953 [details]
hwinfo + boot.msg

The attached logs are from 10.2rc1. Hard disk
was (properly) detected as sda device but CD/DV-ROM
is still not recognized during installation.
So no changes since beta1 :(

Bug is not resolved with RC1

Please test the following kernel.

http://htj.dyndns.org/kernel-default-2.6.18.2-33tj.i586.rpm

Created attachment 107752 [details]
dmesg+hwinfo

Big step forward ! I can now access CD/DVD-ROM with accepable performance:

smyk@tolik:~/Downloads> date +%s; cp /media/SU1010.001/suse/i586/tetex-3.0-37.i586.rpm .; date +%s
1164915523
1164915531
smyk@tolik:~/Downloads> ll
insgesamt 78983
-r--r--r-- 1 smyk users 61255609 30. Nov 20:38 tetex-3.0-37.i586.rpm

(8 sec. for 60 MB file)

CD/DVD is detected as scsi device (/dev/sr0).
I hope the detection will work also durring installation wiht final 10.2 :-) 

Tell me, if you need additional logs/info.

Great and that's enough.

Hannes, this is patch 33 of 'libata: assorted updates against 2.6.18-25' patchset.  Please include that in the kernel cvs.

Patch added to our kernel CVS. Not sure whether it makes it into 10.2 GA, but the next kernel maintenance update will have it definitely.

can you post a patch file for the vanilla kernel too?
I'am not sure which is the right patch now...
or is it allright included in vanilla kernel 2.6.18?
thanks