Harddrives renaming themselves?

[Eadwine Rose]

Only if I get this working for a week without issues will I call this a success but so far so good. Booted a few times to this config due to fried printers, and no strangeness yet.

[Gaer Boy]

Definitely running AHCI here:

Code: Select all

Feb 17 18:27:53 eadwinemx14beta2 kernel: [    0.360692] pci 0000:00:11.0: set SATA to AHCI mode

Code: Select all

Feb 17 18:27:53 eadwinemx14beta2 kernel: [    2.548969] ata3: SATA link up 3.0 Gbps (SStatus 123 SControl 300)

Just rechecked, IDE on both *shrug*

So the kernel is overwriting the BIOS settings - I didn't notice that. There's nothing wrong with that in principle, although it may cause user confusion (not you, Eadwine). My SATA settings are different, and those lines don't appear in my syslog. In the interests of research, I will change my BIOS settings and see if I can reproduce the problem.

[Eadwine Rose]

OK.. remember the change I made to cable select? Nothing else was changed, just changed both Plextor and Sony jumpers to CS.

Didn't do the trick unfortunately.


Going to set the boot order back to HDD first, but here is the output of various commands again, attached.

[Gaer Boy]

Changing the SATA mode to IDE in the BIOS gave me some fun. Booted OK and my syslog looks about the same as Eadwine's - "set SATA to AHCI mode" is included. The I tried booting with a usb data stick plugged in. First 5 attempts hung, with the final one locking the machine with a blank screen. There was no video signal because the monitor was cycling through it's 3 modes.

I reset SATA to AHCI mode in the BIOS, and tried again (usb stick still in) to get "missing operating system". A lunch break seemed to sort that and I'm back to normal.

Anyway, I couldn't find an way of reproducing Eadwine's problem.

[qtech]

Code: Select all

 *-disk                  
       description: SCSI Disk
       product: DVDR   PX-820A
       vendor: PLEXTOR
       physical id: 0.0.0
       bus info: scsi@0:0.0.0
       logical name: /dev/sda
       version: 1.00
       serial: [
       capabilities: removable
       configuration: ansiversion=5 sectorsize=512
     *-medium
          physical id: 0
          logical name: /dev/sda

logical name: /dev/sda

Ok, work with me here as I try to explain this.

Something believes the Plextor is a SCSI Hard Disk. When this happens the Plextor is mounted by Linux as a hard drive,

Code: Select all

/dev/sda

instead of as a DVD drive,

Code: Select all

sr0

Those intermittent times that the Plextor mounts is what is changing your drive letters (Windows speak) from /sda to /sdb or whatever. Likewise, the problem is resolved by you inserting a DVD in the Plextor and booting the computer. With a Disc in the drive, the computer knows not to mount the Plextor as a SCSI disc drive but as DVD instead. Not mounting the confused Plextor means the drive letters don't change.

Technically, it is correct that the Plextor theoretically could be legitimately recognized a SCSI disk drive but it's obviously not what we desire in this situation. Now, if you recall, the guys were talking about "timing" issues earlier. You have IDE, AHCI and USB all trying to mount drives at roughly the same time. Part of the issue here is the timing in which these drives detect and mount (and thus why it is an intermittent problem, the timings are not always the same). The solution is going to be to manipulate that timing. This might be as simple as moving the Lacie to a different USB port (USB timing issues are the more common sort). Or it might require changing the boot order (or other drive related settings) in the BIOS. Worst case scenario, you reverse DVD drives and make the Plextor the slave drive. Of course, then the Sony would try to mount as sda...

Maybe my reasoning here is flawed or the IDE ribbon cable has gone bad but I think we're getting closer to resolving this issue.

[Eadwine Rose]

Then explain to me why this was never an issue in earlier Mepis versions, and ONLY started creeping up in MX14? THAT is what I don't get.

'Sloppy reading' so to speak by the kernel or whatever that thing is called seems more likely to me to be the cause, rather than my hardware that has never given issues.

[qtech]

Then explain to me why this was never an issue in earlier Mepis versions, and ONLY started creeping up in MX14? THAT is what I don't get.

'Sloppy reading' so to speak by the kernel or whatever that thing is called seems more likely to me to be the cause, rather than my hardware that has never given issues.

Eadwine, how old are those DVD players? How old is the IDE ribbon cable?

[BitJam]

Something believes the Plextor is a SCSI Hard Disk.

That "something" is the kernel, which was determined from ER's system log a while ago. Here is an excerpt from the most recent system log:

Code: Select all

Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.973162] ata1.01: ATAPI: SONY DVD-ROM DDU1612, DYS1, max UDMA/33
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.989093] ata1.00: configured for UDMA/66
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.989748] ata1.01: configured for UDMA/33
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.991649] scsi scan: INQUIRY result too short (5), using 36
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.991656] scsi 0:0:0:0: Direct-Access                                    PQ: 0 ANSI: 5
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    2.992101] scsi 0:0:1:0: CD-ROM            SONY     DVD-ROM DDU1612  DYS1 PQ: 0 ANSI: 5
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.016708] ata4: SATA link down (SStatus 0 SControl 300)
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.016773] ata5: SATA link down (SStatus 0 SControl 300)
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.016808] ata6: SATA link down (SStatus 0 SControl 300)
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.048697] usb 6-6: new high-speed USB device number 4 using ehci-pci
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.188644] ata3: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.189769] ata3.00: ATA-8: WDC WD5000AAKS-00WWPA0, 01.03B01, max UDMA/133
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.189773] ata3.00: 976773168 sectors, multi 16: LBA48 NCQ (depth 31/32), AA
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.190958] ata3.00: configured for UDMA/133
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.191152] scsi 2:0:0:0: Direct-Access     ATA      WDC WD5000AAKS-0 01.0 PQ: 0 ANSI: 5
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.196115] sr0: scsi3-mmc drive: 40x/40x cd/rw xa/form2 cdda tray
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.196123] cdrom: Uniform CD-ROM driver Revision: 3.20
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.196309] sr 0:0:1:0: Attached scsi CD-ROM sr0
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198163] sd 2:0:0:0: [sdb] 976773168 512-byte logical blocks: (500 GB/465 GiB)
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198245] sd 0:0:0:0: Attached scsi generic sg0 type 0
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198279] sd 2:0:0:0: [sdb] Write Protect is off
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198282] sd 2:0:0:0: [sdb] Mode Sense: 00 3a 00 00
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198338] sd 2:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198342] sr 0:0:1:0: Attached scsi generic sg1 type 5
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.198501] sd 2:0:0:0: Attached scsi generic sg2 type 0
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.199311] sd 0:0:0:0: [sda] READ CAPACITY(16) failed
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.199315] sd 0:0:0:0: [sda]  
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.199317] Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.199319] sd 0:0:0:0: [sda] Sense not available.
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.204706] sd 0:0:0:0: [sda] Write Protect is off
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.204712] sd 0:0:0:0: [sda] Mode Sense: 00 da 70 00
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.206883] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.217923] sd 0:0:0:0: [sda] READ CAPACITY(16) failed
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.217928] sd 0:0:0:0: [sda]  
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.217930] Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.217931] sd 0:0:0:0: [sda] Sense not available.
Mar  5 13:15:42 eadwinemx14beta2 kernel: [    3.224647] sd 0:0:0:0: [sda] Attached SCSI disk

It is the kernel that assigns the name sda to the dvd drive. The first time sda appears in the log is in a message that READ CAPACITY failed. I asked for the blkid output for extra confirmation the name sda is coming from the kernel (and not from udev) but when the error condition occurs, the sda device does not show up in the blkid output or in /proc/partitions or /proc/sys/dev/cdrom/info. It is not just a naming problem, the kernel is confused and cannot access the device.

The first error is:

Code: Select all

scan: INQUIRY result too short (5), using 36

Google(INQUIRY result too short) yielded a lot of hits. One [url=http://comments.gmane.org/gmane.linux.scsi/74353]succinct explanation was:

Well, it means what it says: the device wrongly returned only 5 bytes
for the INQUIRY command. Why it did this is a hardware issue.

Technically, it is correct that the Plextor theoretically could be legitimately recognized a SCSI disk drive

This is simply not true. It is either a problem with the hardware/harware-configuration or it is a bug in the kernel. If the hardware is working correctly then dvd drives should always be recognized as dvd drives, otherwise it is a bug in the kernel.

You have IDE, AHCI and USB all trying to mount drives at roughly the same time.

As I said before, I don't think physically different buses are directly affecting the IDE bus. That would be rather magical. Although it is possible they are affecting each other indirectly via the power supply. It takes a lot of power to spin up a hard drive (or even an optical drive). If they are all spinning up together then maybe this causes a small dip in the power that affects the Plextor drive. As I've said before, it is now legitimate for the device names to change each boot due to race conditions between the various devices becoming available. But it is not legitimate to classify a dvd drive as a hard drive.

Part of the issue here is the timing in which these drives detect and mount (and thus why it is an intermittent problem, the timings are not always the same.

I'm glad we agree.

The solution is going to be to manipulate that timing.

This may fix the problem but if so, it is a work-around, not a solution. IMO, once we have ruled out a hardware/hardware-config problem then the next step is to report a possible kernel bug.

I like your idea of swapping the jumper config on the Sony and Plextor drives. It will be interesting to see if the problem really does affect the Sony as well. This would help rule out that something in the Plextor drive is contributing to the problem. Another thing to try is to disconnect the Sony and see if the problem persists. I also like your idea of swapping out the ribbon cable. Just unplugging and replugging the connectors might help. Physical connectors have a relatively high failure rate, although it is curious that the problem always manifests the same way. Faulty connectors should create random errors whenever you access the drives.

[Eadwine Rose]

How old are the drives, well they are the old kind, so they are not new. Can't remember how old exactly but the Plextor was relatively new when this system was build so I wager about 4 - 4.5 years old. The Sony is likely a bit older than that.

The cables and mobo and stuff were all bought in Sept 2010.


The drives were unplugged taken out of the case and replugged Tuesday, in order to get the jumpers switched. I don't remember this correctly, but the cable does not look to be long enough to switch the cable order around.

I will have my friend read and see what he says.

[BitJam]

Then explain to me why this was never an issue in earlier Mepis versions, and ONLY started creeping up in MX14? THAT is what I don't get.

Have you tested it to see if the problem now occurs with older kernels? I don't think it will but if it does then that means it is a hardware problem the developed around the time you started using MX-14.

'Sloppy reading' so to speak by the kernel or whatever that thing is called seems more likely to me to be the cause, rather than my hardware that has never given issues.

If you are willing to change "sloppy reading" to "more aggressive reading" then I would agree with you, although this is only a guess. It is not unusual for newer kernels to have problems with older hardware. Sometimes a new boot parameter is needed that wasn't needed before. For example, on both my desktop computers I need to add nomodeset. If I don't then the nouveau driver makes them unusable.

Here is an official (but incomplete) list of kernel boot parameters. The boot parameter acpi=off often helps to boot older laptops. Older versions of the kernel have code that links ide to acpi so maybe that boot parameter will help you.

If it is not onerous then it would be interesting to see if the problem persists when you keep the Sony drive unplugged (IIUC, that is the only other drive on your ide bus). If this fixes things then it would add credence to the idea that the problem was caused by more aggressive scanning by the kernel, which caused a conflict on the ide bus. If it doesn't fix things then the problem is most likely in the Plextor drive and it might also occur with older kernels. Either answer could bring us closer to a solution.