Re: sata_sil data corruption
Jeff Garzik wrote: Been there done that, all on this list actually. Bios upgrade, bios reset, new cable, different drive, correct functionality in windows, and now the drive and cable ahave found their home in my other PC, a kt600 board using the sata_via driver, where it works flawlessly. I'm not even interested in using the sata interface on this machine anymore, but I still try to make an effort whenever possible to contribute byt reporting bugs. So many people with so many different hardware combinations, sometimes it takes the right combination of hardware/software to reveal a bug. That being said, if you want to investigate further, I can provide any information you want. If you want to ignore it, thats fine too. Evidence strongly leans towards a driver bug though. I also noticed that I get oops now sometimes even with no drive attatched at all What is a blacklist entry? A combination of drive/controller that is determined to be incompatable together? I would assume for a drive to make it to a blacklists it would have to be incompatable regardless of OS right? The drive in question is a WD2000JD Doug In this case, the bug _reports_ are hard to find. Each case with sata_sil is either solved with a BIOS update, a blacklist entry, or new cables. Just read through bugzilla.kernel.org. Jeff - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html -- http://nostar.net/ - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
Jeff Garzik schrieb: On Mon, Feb 21, 2005 at 11:25:55AM +, Doug McLain wrote: Jeff Garzik wrote: Doug McLain wrote: The sata_sil driver is without a doubt, totally hosed. I, along with "without a doubt" being defined, of course, as "it works for a lot of people." Jeff Thats like saying "turn up the radio" when your car makes a funny noise, or "if a tree falls in the woods and nobody is there to hear it, does it make a sound?" It's tempting and comforting to pick the good ones as an example, and some bugs are hard enough to find, let alone fix. In the end though, if one is broke, it's still broke, isn't it? In this case, the bug _reports_ are hard to find. Each case with sata_sil is either solved with a BIOS update, a blacklist entry, or new cables. Just read through bugzilla.kernel.org. I personally think the driver is OK (works for me like a charm with my Samsung drive), but as I reported a few times, SiI will corrupt data if ext-p2p discard time setting in bios is set to too low values. So, Jeff, don't you think adding a quirk to the kernel would prevent such reports. (I am damn sure, this reports are cause of the setting I am talking about.) Perhaps you could ask SiI how to do this in the kernel... -- Prakash Punnoor formerly known as Prakash K. Cheemplavam signature.asc Description: OpenPGP digital signature
Re: sata_sil data corruption
On Mon, Feb 21, 2005 at 11:25:55AM +, Doug McLain wrote: > Jeff Garzik wrote: > >Doug McLain wrote: > > > >>The sata_sil driver is without a doubt, totally hosed. I, along with > > > > > >"without a doubt" being defined, of course, as "it works for a lot of > >people." > > > >Jeff > > > > > > > > > Thats like saying "turn up the radio" when your car makes a funny noise, > or "if a tree falls in the woods and nobody is there to hear it, does it > make a sound?" > > It's tempting and comforting to pick the good ones as an example, and > some bugs are hard enough to find, let alone fix. In the end though, if > one is broke, it's still broke, isn't it? In this case, the bug _reports_ are hard to find. Each case with sata_sil is either solved with a BIOS update, a blacklist entry, or new cables. Just read through bugzilla.kernel.org. Jeff - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
Jeff Garzik wrote: Doug McLain wrote: The sata_sil driver is without a doubt, totally hosed. I, along with "without a doubt" being defined, of course, as "it works for a lot of people." Jeff Thats like saying "turn up the radio" when your car makes a funny noise, or "if a tree falls in the woods and nobody is there to hear it, does it make a sound?" It's tempting and comforting to pick the good ones as an example, and some bugs are hard enough to find, let alone fix. In the end though, if one is broke, it's still broke, isn't it? -- http://nostar.net/ - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
Doug McLain wrote: The sata_sil driver is without a doubt, totally hosed. I, along with "without a doubt" being defined, of course, as "it works for a lot of people." Jeff - To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
The sata_sil driver is without a doubt, totally hosed. I, along with
some others I've found on this list and google, are also getting kernel
oops and hard freezes while loading this driver, with confirmed good
hardware.
PALFFY Daniel wrote:
On Sat, 19 Feb 2005, Jeff Garzik wrote:
PALFFY Daniel wrote:
Hi,
I'm trying to set up a machine with a si3112a controller (lspci: 1095:3112
(rev 01) Subsystem: 1095:6112, cheap PCI card) and a ST3200822AS Rev:
3.01 disk and I see continuous silent data corruption while reading the
disk. Writing seems to be ok. I have 2.6.10-ac1 built with conservative
options (UP, no PREEMPT, 8k stack, no regparm). After seeing problems I
tried to blacklist my drive to do MOD15, but it didn't help.
Finally I did
unsigned long long i = 0;
while (write(1, &i, sizeof(i)) != -1) i++;
to the only primary partition while running sata_sil. Reading it back with
unsigned long long i=0, j;
while (read(0, &j, sizeof(i)) == sizeof(i)) {
if (j != i) fprintf(stderr, "diff at %llx: read: %llx\n", i, j);
i++;
}
gives similar results, but always different:
diff at 5efff: read: 5ef24
diff at 7: read: 7ff08
diff at 8: read: 8ff51
diff at 97fff: read: 97f00
diff at a: read: aff00
diff at b: read: bffac
diff at d: read: dff00
diff at efffe: read: eff00
diff at e: read: eff00
diff at fffbf: read: f
and so on.
Reading back the same data with the ide siimage driver worked for at least
500MB without corrupted data, but dma doesn't work with that driver, this
is logged on about the first read attempt:
hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
hde: DMA disabled
ide2: reset phy, status=0x0113, siimage_reset
ide2: reset: success
The machine is an old Compaq Prosignia 200, with a p2 300 and
:00:00.0 Host bridge: Intel Corp. 440FX - 82441FX PMC [Natoma] (rev 02)
chipset. Relevant parts from dmesg:
sata_sil:
libata version 1.10 loaded.
sata_sil version 0.8
ata1: SATA max UDMA/100 cmd 0xC886A080 ctl 0xC886A08A bmdma 0xC886A000 irq 10
ata2: SATA max UDMA/100 cmd 0xC886A0C0 ctl 0xC886A0CA bmdma 0xC886A008 irq 10
ata1: dev 0 cfg 49:2f00 82:346b 83:7d01 84:4003 85:3469 86:3c01 87:4003 88:207f
ata1: dev 0 ATA, max UDMA/133, 390721968 sectors: lba48
ata1(0): applying Seagate errata fix
ata1: dev 0 configured for UDMA/100
scsi1 : sata_sil
ata2: no device found (phy stat )
scsi2 : sata_sil
Vendor: ATA Model: ST3200822AS Rev: 3.01
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sdb: 390721968 512-byte hdwr sectors (200050 MB)
SCSI device sdb: drive cache: write back
siimage:
SiI3112 Serial ATA: IDE controller at PCI slot :00:0a.0
SiI3112 Serial ATA: chipset revision 1
SiI3112 Serial ATA: 100% native mode on irq 10
ide2: MMIO-DMA , BIOS settings: hde:DMA, hdf:DMA
ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio
Probing IDE interface ide2...
hde: ST3200822AS, ATA DISK drive
hde: applying pessimistic Seagate errata fix
ide2 at 0xc886a080-0xc886a087,0xc886a08a on irq 10
hde: max request size: 7KiB
hde: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63
hde: cache flushes supported
hde:<3>hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
hde1
Probing IDE interface ide3...
hdg: no response (status = 0xfe)
hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
Hi,
Don't use --two-- drivers for the same hardware.
Can you re-test with siimage disabled?
Of course it was disabled while testing sata_sil and vice-versa. I've just
tried that driver to test if the hardware was faulty. Since then, I've
tested the same disk/controller combo in a BX based PentiumII system, and
it worked perfectly fine, but in the Compaq machine it still failed with
2.6.10-ac10. So I think it might be some low-level hardware
incompatibility...
--
Daniel
...and Linux for all.
-
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
http://nostar.net/
-
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
On Sat, 19 Feb 2005, Jeff Garzik wrote:
> PALFFY Daniel wrote:
> > Hi,
> >
> > I'm trying to set up a machine with a si3112a controller (lspci: 1095:3112
> > (rev 01) Subsystem: 1095:6112, cheap PCI card) and a ST3200822AS Rev:
> > 3.01 disk and I see continuous silent data corruption while reading the
> > disk. Writing seems to be ok. I have 2.6.10-ac1 built with conservative
> > options (UP, no PREEMPT, 8k stack, no regparm). After seeing problems I
> > tried to blacklist my drive to do MOD15, but it didn't help.
> >
> > Finally I did
> >
> > unsigned long long i = 0;
> > while (write(1, &i, sizeof(i)) != -1) i++;
> >
> > to the only primary partition while running sata_sil. Reading it back with
> >
> > unsigned long long i=0, j;
> > while (read(0, &j, sizeof(i)) == sizeof(i)) {
> > if (j != i) fprintf(stderr, "diff at %llx: read: %llx\n", i, j);
> > i++;
> > }
> >
> > gives similar results, but always different:
> > diff at 5efff: read: 5ef24
> > diff at 7: read: 7ff08
> > diff at 8: read: 8ff51
> > diff at 97fff: read: 97f00
> > diff at a: read: aff00
> > diff at b: read: bffac
> > diff at d: read: dff00
> > diff at efffe: read: eff00
> > diff at e: read: eff00
> > diff at fffbf: read: f
> > and so on.
> >
> > Reading back the same data with the ide siimage driver worked for at least
> > 500MB without corrupted data, but dma doesn't work with that driver, this
> > is logged on about the first read attempt:
> >
> > hde: dma_intr: bad DMA status (dma_stat=76)
> > hde: dma_intr: status=0x50 { DriveReady SeekComplete }
> >
> > ide: failed opcode was: unknown
> > hde: DMA disabled
> > ide2: reset phy, status=0x0113, siimage_reset
> > ide2: reset: success
> >
> > The machine is an old Compaq Prosignia 200, with a p2 300 and
> > :00:00.0 Host bridge: Intel Corp. 440FX - 82441FX PMC [Natoma] (rev 02)
> > chipset. Relevant parts from dmesg:
> >
> > sata_sil:
> >
> > libata version 1.10 loaded.
> > sata_sil version 0.8
> > ata1: SATA max UDMA/100 cmd 0xC886A080 ctl 0xC886A08A bmdma 0xC886A000 irq
> > 10
> > ata2: SATA max UDMA/100 cmd 0xC886A0C0 ctl 0xC886A0CA bmdma 0xC886A008 irq
> > 10
> > ata1: dev 0 cfg 49:2f00 82:346b 83:7d01 84:4003 85:3469 86:3c01 87:4003
> > 88:207f
> > ata1: dev 0 ATA, max UDMA/133, 390721968 sectors: lba48
> > ata1(0): applying Seagate errata fix
> > ata1: dev 0 configured for UDMA/100
> > scsi1 : sata_sil
> > ata2: no device found (phy stat )
> > scsi2 : sata_sil
> > Vendor: ATA Model: ST3200822AS Rev: 3.01
> > Type: Direct-Access ANSI SCSI revision: 05
> > SCSI device sdb: 390721968 512-byte hdwr sectors (200050 MB)
> > SCSI device sdb: drive cache: write back
> >
> > siimage:
> >
> > SiI3112 Serial ATA: IDE controller at PCI slot :00:0a.0
> > SiI3112 Serial ATA: chipset revision 1
> > SiI3112 Serial ATA: 100% native mode on irq 10
> > ide2: MMIO-DMA , BIOS settings: hde:DMA, hdf:DMA
> > ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio
> > Probing IDE interface ide2...
> > hde: ST3200822AS, ATA DISK drive
> > hde: applying pessimistic Seagate errata fix
> > ide2 at 0xc886a080-0xc886a087,0xc886a08a on irq 10
> > hde: max request size: 7KiB
> > hde: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63
> > hde: cache flushes supported
> > hde:<3>hde: dma_intr: bad DMA status (dma_stat=76)
> > hde: dma_intr: status=0x50 { DriveReady SeekComplete }
> >
> > ide: failed opcode was: unknown
> > hde1
> > Probing IDE interface ide3...
> > hdg: no response (status = 0xfe)
> > hde: dma_intr: bad DMA status (dma_stat=76)
> > hde: dma_intr: status=0x50 { DriveReady SeekComplete }
> >
> > ide: failed opcode was: unknown
Hi,
> Don't use --two-- drivers for the same hardware.
>
> Can you re-test with siimage disabled?
Of course it was disabled while testing sata_sil and vice-versa. I've just
tried that driver to test if the hardware was faulty. Since then, I've
tested the same disk/controller combo in a BX based PentiumII system, and
it worked perfectly fine, but in the Compaq machine it still failed with
2.6.10-ac10. So I think it might be some low-level hardware
incompatibility...
--
Daniel
...and Linux for all.
-
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: sata_sil data corruption
PALFFY Daniel wrote:
Hi,
I'm trying to set up a machine with a si3112a controller (lspci: 1095:3112
(rev 01) Subsystem: 1095:6112, cheap PCI card) and a ST3200822AS Rev:
3.01 disk and I see continuous silent data corruption while reading the
disk. Writing seems to be ok. I have 2.6.10-ac1 built with conservative
options (UP, no PREEMPT, 8k stack, no regparm). After seeing problems I
tried to blacklist my drive to do MOD15, but it didn't help.
Finally I did
unsigned long long i = 0;
while (write(1, &i, sizeof(i)) != -1) i++;
to the only primary partition while running sata_sil. Reading it back with
unsigned long long i=0, j;
while (read(0, &j, sizeof(i)) == sizeof(i)) {
if (j != i) fprintf(stderr, "diff at %llx: read: %llx\n", i, j);
i++;
}
gives similar results, but always different:
diff at 5efff: read: 5ef24
diff at 7: read: 7ff08
diff at 8: read: 8ff51
diff at 97fff: read: 97f00
diff at a: read: aff00
diff at b: read: bffac
diff at d: read: dff00
diff at efffe: read: eff00
diff at e: read: eff00
diff at fffbf: read: f
and so on.
Reading back the same data with the ide siimage driver worked for at least
500MB without corrupted data, but dma doesn't work with that driver, this
is logged on about the first read attempt:
hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
hde: DMA disabled
ide2: reset phy, status=0x0113, siimage_reset
ide2: reset: success
The machine is an old Compaq Prosignia 200, with a p2 300 and
:00:00.0 Host bridge: Intel Corp. 440FX - 82441FX PMC [Natoma] (rev 02)
chipset. Relevant parts from dmesg:
sata_sil:
libata version 1.10 loaded.
sata_sil version 0.8
ata1: SATA max UDMA/100 cmd 0xC886A080 ctl 0xC886A08A bmdma 0xC886A000 irq 10
ata2: SATA max UDMA/100 cmd 0xC886A0C0 ctl 0xC886A0CA bmdma 0xC886A008 irq 10
ata1: dev 0 cfg 49:2f00 82:346b 83:7d01 84:4003 85:3469 86:3c01 87:4003 88:207f
ata1: dev 0 ATA, max UDMA/133, 390721968 sectors: lba48
ata1(0): applying Seagate errata fix
ata1: dev 0 configured for UDMA/100
scsi1 : sata_sil
ata2: no device found (phy stat )
scsi2 : sata_sil
Vendor: ATA Model: ST3200822AS Rev: 3.01
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sdb: 390721968 512-byte hdwr sectors (200050 MB)
SCSI device sdb: drive cache: write back
siimage:
SiI3112 Serial ATA: IDE controller at PCI slot :00:0a.0
SiI3112 Serial ATA: chipset revision 1
SiI3112 Serial ATA: 100% native mode on irq 10
ide2: MMIO-DMA , BIOS settings: hde:DMA, hdf:DMA
ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio
Probing IDE interface ide2...
hde: ST3200822AS, ATA DISK drive
hde: applying pessimistic Seagate errata fix
ide2 at 0xc886a080-0xc886a087,0xc886a08a on irq 10
hde: max request size: 7KiB
hde: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63
hde: cache flushes supported
hde:<3>hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
hde1
Probing IDE interface ide3...
hdg: no response (status = 0xfe)
hde: dma_intr: bad DMA status (dma_stat=76)
hde: dma_intr: status=0x50 { DriveReady SeekComplete }
ide: failed opcode was: unknown
Don't use --two-- drivers for the same hardware.
Can you re-test with siimage disabled?
Jeff
-
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
