Re: Linux v2.6.16-rc6

[Paul Blazejowski <paulb@blazebox.homeip.net>]

[-- Attachment #1: Type: text/plain, Size: 4120 bytes --]

On Sun, 2006-03-12 at 16:46 -0500, Lee Revell wrote:
> On Sun, 2006-03-12 at 13:45 -0500, Paul Blazejowski wrote:
> > On recent kernel 2.6.15.6 (or any 2.6.15.x) and latest testing
> > 2.6.16-rc6 libata detects and sets wrong UDMA modes for one of the
> > SATA-1 drives. This seems to be a bug.
> > 
> > My setup is as follows:
> > 
> > ASUS A8N-SLI-Premium Nforce4 mainboard
> > AMD Athlon X2 CPU running SMP
> > GCC 3.3.6
> > Slackware 10.2 Linux
> > 
> > The drives are used in RAID1 array (dmraid), they are WDC-WD2000JD
> > series purchased few months apart. Sata is compiled in the kernel as
> > module sata_nv and functions properly, no errors or any other anomalies
> > were noticed but the UDMA mode detection seem wrong on the second drive.
> > 
> > Drive one reports ata3: dev 0 configured for UDMA/100 while drive two
> > ata4: dev 0 configured for UDMA/133
> 
> This bug report is still somewhat unclear.
> 
> What are the correct modes you expect to see?
> 
> Lee
> 
> 

I belive the modes should say DMA100 because UDMA133 would be mode ATA-7
and DMA100 ATA-6 mode. This is the info i get from hdparm -I on the ata3
drive: DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5

while on the ata4 one:  DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3
udma4 udma5 *udma6

The thing is that the older drive is the one with the mode being set at
133 while the newer with mode 100. I belive they come from the same
factory but do carry different firmware revisons.

I also tried the drives on sata_sil (sil3114) controller and they show
the same modes being detected:

sata_sil 0000:05:0a.0: Applying R_ERR on DMA activate FIS errata fix
ata5: SATA max UDMA/100 cmd 0xF9402080 ctl 0xF940208A bmdma 0xF9402000
irq 23
ata6: SATA max UDMA/100 cmd 0xF94020C0 ctl 0xF94020CA bmdma 0xF9402008
irq 23
ata7: SATA max UDMA/100 cmd 0xF9402280 ctl 0xF940228A bmdma 0xF9402200
irq 23
ata8: SATA max UDMA/100 cmd 0xF94022C0 ctl 0xF94022CA bmdma 0xF9402208
irq 23
ata5: SATA link up 1.5 Gbps (SStatus 113)
ata5: dev 0 cfg 49:2f00 82:306b 83:7e01 84:4003 85:3068 86:3c01 87:4003
88:203f
ata5: dev 0 ATA-6, max UDMA/100, 390721968 sectors: LBA48
ata5: dev 0 configured for UDMA/100
scsi5 : sata_sil
ata6: SATA link up 1.5 Gbps (SStatus 113)
ata6: dev 0 cfg 49:2f00 82:346b 83:7f61 84:4003 85:3468 86:3c41 87:4003
88:207f
ata6: dev 0 ATA-6, max UDMA/133, 390721968 sectors: LBA48
ata6: dev 0 configured for UDMA/100
scsi6 : sata_sil
ata7: SATA link down (SStatus 0)
scsi7 : sata_sil
ata8: SATA link down (SStatus 0)
scsi8 : sata_sil
  Vendor: ATA       Model: WDC WD2000JD-60K  Rev: 08.0
  Type:   Direct-Access                      ANSI SCSI revision: 05
SCSI device sdb: 390721968 512-byte hdwr sectors (200050 MB)
sdb: Write Protect is off
sdb: Mode Sense: 00 3a 00 00
SCSI device sdb: drive cache: write back
SCSI device sdb: 390721968 512-byte hdwr sectors (200050 MB)
sdb: Write Protect is off
sdb: Mode Sense: 00 3a 00 00
SCSI device sdb: drive cache: write back
 sdb: sdb1 sdb2 sdb3 sdb4
sd 5:0:0:0: Attached scsi disk sdb
  Vendor: ATA       Model: WDC WD2000JD-00H  Rev: 08.0
  Type:   Direct-Access                      ANSI SCSI revision: 05
SCSI device sdc: 390721968 512-byte hdwr sectors (200050 MB)
sdc: Write Protect is off
sdc: Mode Sense: 00 3a 00 00
SCSI device sdc: drive cache: write back
SCSI device sdc: 390721968 512-byte hdwr sectors (200050 MB)
sdc: Write Protect is off
sdc: Mode Sense: 00 3a 00 00
SCSI device sdc: drive cache: write back
 sdc: sdc1 sdc2 sdc3 sdc4
sd 6:0:0:0: Attached scsi disk sdc

I also see a difference with the transfer rates from hdparm -Tt:

ata3 drive (mode UDMA100) shows:

Timing buffered disk reads:  172 MB in  3.00 seconds =  57.30 MB/sec

while ata4 drive (mode UDMA133) shows:

Timing buffered disk reads:  118 MB in  3.03 seconds =  38.96 MB/sec

At this point is this due to drive capabilites in regards to modes
supported, broken drive? or libata code bug? I am trying to be as clear
as possible, anything else i should provide?

Thanks,

Paul B.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 191 bytes --]