#348929 - 06/11/2011 05:05
Re: Current HD->CompactFlash status?
[Re: mlord]
|
carpal tunnel
Registered: 08/06/1999
Posts: 7868
|
One expensive route I did find for a DIY SSD is this unit: http://www.jpcentury.com/pro_con.aspx?id=P_00000025It takes 3 CF cards and presents them as one drive. The company selling them appears to be moving away from IDE to newer SATA ones, so if this interests anyone else, you may want to track them down soon. Though from reports by others here, SATA to PATA adaptors seem to work well in the empeg. Price is a bit high at ~$230 or so. But this may be worth it if someone is sitting on several higher capacity CF cards.
|
Top
|
|
|
|
#348992 - 07/11/2011 19:35
Re: Current HD->CompactFlash status?
[Re: drakino]
|
carpal tunnel
Registered: 08/06/1999
Posts: 7868
|
Hrm, something somewhere has gone wrong, and was exposed by my upgrade from 2.0 to 3.0 alpha 11. empeg:/empeg# ls -al
total 4
drwxr-xr-x 4 0 0 1024 Jul 25 2005 .
drwxr-xr-x 15 521 220 1024 Oct 6 01:45 ..
drwxr-xr-x 2 0 0 1024 Jul 25 2005 bin
lrwxrwxrwx 1 0 220 14 Jul 25 2005 fids0 -> ../drive0/fids
lrwxrwxrwx 1 0 220 14 Jul 25 2005 fids1 -> ../drive1/fids
drwxr-xr-x 6 0 0 1024 Jul 25 2005 lib
lrwxrwxrwx 1 0 220 13 Jul 25 2005 var -> ../drive0/var
empeg:/empeg# ls -al /drive0
total 2
drwxr-xr-x 2 0 0 1024 Jul 6 1999 .
drwxr-xr-x 15 521 220 1024 Oct 6 01:45 ..
empeg:/empeg# ls -al /drive1
total 2
drwxr-xr-x 2 0 0 1024 Jun 2 1999 .
drwxr-xr-x 15 521 220 1024 Oct 6 01:45 ..
empeg:/empeg# mount
/dev/root on / type ext2 (rw)
none on /proc type proc (rw)
empeg:/empeg# fdisk -l /dev/hda
Disk /dev/hda: 255 heads, 63 sectors, 3899 cylinders
Units = cylinders of 16065 * 512 bytes
Device Boot Start End Blocks Id System
/dev/hda1 1 5 40131 5 Extended
/dev/hda2 6 10 40162+ 83 Linux
/dev/hda3 11 13 24097+ 10 OPUS
/dev/hda4 14 3899 31214295 83 Linux
/dev/hda5 1 3 24034+ 83 Linux
/dev/hda6 4 5 16033+ 82 Linux swap
empeg:/empeg# fdisk -l /dev/hdc
Disk /dev/hdc: 255 heads, 63 sectors, 3899 cylinders
Units = cylinders of 16065 * 512 bytes
Device Boot Start End Blocks Id System
/dev/hdc1 1 5 33232+ 5 Extended
Partition 1 does not end on cylinder boundary:
phys=(65, 15, 63) should be (65, 254, 63)
/dev/hdc2 5 9 33264 83 Linux
Partition 2 does not end on cylinder boundary:
phys=(131, 15, 63) should be (131, 254, 63)
/dev/hdc3 9 11 16632 10 OPUS
Partition 3 does not end on cylinder boundary:
phys=(164, 15, 63) should be (164, 254, 63)
/dev/hdc4 11 3900 31242960 83 Linux
Partition 4 does not end on cylinder boundary:
phys=(1023, 15, 63) should be (1023, 254, 63)
/dev/hdc5 1 3 16569 83 Linux
/dev/hdc6 3 5 16600+ 82 Linux swap For the drive expansion, I ended up using gparted with the CF cards attached to a USB to IDE adaptor. /dev/hda is the CF card I cloned in the empeg first. /dev/hdc was cloned on my Mac using dd. I think my attempt to hang onto the nostalgia of all my old music loadout, along with whatever crazy hacks* I was doing in the past, and the HDD->CF conversion has introduced so many variables. Time I think to just nuke it from orbit and move forward with the new plans to integrate it into my newer music strategy. * I was finding remnants of the GPS app, empire, emphatic, various web interfaces, my old distributed.net setup, strongbad TTS clock, random bits of debian linux, and some other things still on the drives.
|
Top
|
|
|
|
#348996 - 07/11/2011 20:33
Re: Current HD->CompactFlash status?
[Re: drakino]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14493
Loc: Canada
|
What is it that you've found that "has gone wrong?"
Edit: oh, okay.. funny looking sector/cylinder ranges. Still, it all looks valid. What's it supposed to look like?
Edited by mlord (07/11/2011 20:36)
|
Top
|
|
|
|
#349001 - 07/11/2011 22:22
Re: Current HD->CompactFlash status?
[Re: mlord]
|
carpal tunnel
Registered: 08/06/1999
Posts: 7868
|
The other issue is that the music partition wasn't mounting properly, hence the missing fids folders /empeg/fids0 tries to point to. *edit* Also, hda and hdc should have been identical in fdisk. The block boundaries were weirdly out of place, probably a gparted failure that it didn't report well.
I could have dug deeper, but I just didn't trust the state of things any more. I can't remember every little thing I did to this unit, some of it dating back to my hacks I carried over from the Mark 1 in 99.
If I do want to revisit the "classic" empeg I used in the past, the data is still saved off, both on the drives I replaced, and as a backup I took.
Edited by drakino (07/11/2011 22:57)
|
Top
|
|
|
|
#349010 - 08/11/2011 01:22
Re: Current HD->CompactFlash status?
[Re: drakino]
|
enthusiast
Registered: 21/02/2006
Posts: 325
|
Hi,
I noticed this statement in your post....
"Though from reports by others here, SATA to PATA adaptors seem to work well in the empeg."
I am still trying to get these to work. I could not get Emplode to play well with it.
I don't think I have seen a post of success yet. Did I miss something?
Thanks,
Ross
_________________________
In SI, a little termination and attention to layout goes a long way. In EMC, without SI, you'll spend 80% of the effort on the last 3dB.
|
Top
|
|
|
|
#349011 - 08/11/2011 01:46
Re: Current HD->CompactFlash status?
[Re: Ross Wellington]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14493
Loc: Canada
|
Oh, so there's supposed to be tunes/playlists there already. That's what I blindly missed. And yeah, I'm with Ross: I don't recall seeing many success reports with adaptors yet. That said, they should work, though I do know that a lot of them are quirky. Cheers
|
Top
|
|
|
|
#349057 - 09/11/2011 19:55
Re: Current HD->CompactFlash status?
[Re: mlord]
|
carpal tunnel
Registered: 19/05/1999
Posts: 3457
Loc: Palo Alto, CA
|
cat (treating it as a char device) The device is a block device, regardless of the program opening/reading from it. cat uses 32KB buffers by default in such situations. The kernel reads a maximum of about 64KB at a time from IDE drives. I understand that, but isn't it going to be going through at least one more set of buffers using cat vs dd? Memory bandwidth isn't amazing on the empeg either
|
Top
|
|
|
|
#349059 - 09/11/2011 20:29
Re: Current HD->CompactFlash status?
[Re: altman]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14493
Loc: Canada
|
No, dd and cat both do pretty much the same system calls normally: read() followed by write().
Newer versions of dd have a "DIRECT_IO" flag, which gets rid of one memcpy (or possibly both) from the loop. That would be quicker on an empeg, but probably neutral on a PC.
With dd, one can also easily specify a very large buffer size, which will reduce the number of read/write calls, saving a tiny amount of overhead.
Cheers
|
Top
|
|
|
|
#350569 - 06/03/2012 04:56
Re: Current HD->CompactFlash status?
[Re: mlord]
|
member
Registered: 16/12/1999
Posts: 188
Loc: Melbourne, Australia
|
Some more datapoints relating to SATA -> PATA...
My empeg hard disk has been making funny noises, and is occasionally (although with increasing frequency) refusing to start up. I figured I'd try out the SATA situation.
I ordered two separate adapters, and have tried them with both a 500G WD drive, and a 120G Intel SSD, just with one drive at a time.
With the JMicron based one, the empeg didn't seem able to accurately identify the drives, for either drive.
With the Addonics adapter, the empeg was able to identify each drive properly.
I was able to partition and format the 500G drive with no problems. However, as Ross has experienced, the ethernet was broken with this adapter. Also, the USB was broken. Both work fine when I use the old PATA hard drive instead.
I've attached a boot log from this. There seems to be an issue detecting the ethernet adapter, and I can't help but wonder if I was able to force the kernel to look for it at a particular address, maybe it would work?
With the Intel SSD, things seem to go wrong when I start writing significant amounts of data to the drive. Trying to format a partition or install the bigdisk builder on it fails. I haven't catalogued the errors yet, but the most common one I see on the empeg display is hda: err: write_intr (which is obviously missing all the interesting stuff at the end). (The SSD works fine plugged into a SATA->USB converter on my PC.)
Is it possible that the SSD is just returning from writes too quickly? This seems unlikely: given that HDDs have caches I would expect many writes to return very quickly.
Anyway, my plan is to see if I can get some more details about the errors with the SSD, and maybe that will be helpful.
If anybody has ideas about what sort of logging / debugging might be useful for this, I'm all ears.
Richard.
Attachments
bad-boot-log.txt (412 downloads)
|
Top
|
|
|
|
#350695 - 09/03/2012 01:42
Re: Current HD->CompactFlash status?
[Re: rjlov]
|
member
Registered: 16/12/1999
Posts: 188
Loc: Melbourne, Australia
|
I've got some more logs, and I'm not really sure what they mean. Some random, possibly interesting facts (some reiterated from before): - if the SATA->PATA adapter is a slave, then the ethernet is recognised (although writing this I realise that I didn't test if the ethernet actually works).
- SATA->PATA works fine (albeit stomping on USB and network) with a spinning disk, was able to build disk and also copy about 60G of music to it, with no errors that I noticed.
- SSD with SATA->PATA has issues with writing. Errors generally look like:
hdb: lost interrupt
hdb: stat_err: status=0x58
hdb: drive not ready for command
hdb: stat_timo: status=0xd0
ide0: reset timed-out, status=0x80
(See attached log of mke2fs for details) - I did once see errors like this, but haven't been able to reproduce reliably:
hdb: stat_timo: status=0x80
end_request: I/O error, dev 03:44 (hdb), sector 262232
hdb: drive not ready for command
hdb: stat_err: status=0xff
hdb: no DRQ after issuing WRITE
- SSD advertises itself as having a 0kb cache, (see attached output from hdparm -i/-I) I wonder if that might be part of the problem.
In the absence of suggestions I'll just keep pottering away. I'll post here in the event of a breakthrough. Cheers, Richard.
Attachments
hdparm-log.txt (481 downloads)mke2fs-log.txt (387 downloads)
|
Top
|
|
|
|
#350698 - 09/03/2012 05:31
Re: Current HD->CompactFlash status?
[Re: rjlov]
|
enthusiast
Registered: 21/02/2006
Posts: 325
|
Hi,
I ended up giving up on this for a while.
After I got the hardware somewhat working (much the same problems as you are experiencing), I ran into problems with Emplode and Jemplode not enjoying the SATA experience.
I hope you find something I missed.
In the meantime, the WD 250GB PATA drives went up in price.
Thanks,
Ross
_________________________
In SI, a little termination and attention to layout goes a long way. In EMC, without SI, you'll spend 80% of the effort on the last 3dB.
|
Top
|
|
|
|
#350702 - 09/03/2012 11:47
Re: Current HD->CompactFlash status?
[Re: Ross Wellington]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14493
Loc: Canada
|
Newer drives, especially SSDs, are a lot faster with ATA protocol state transitions than older drives. This exposes flaws in the original (Intel) ATA protocol spec, as well as races in the Linux IDE code that I wrote nearly 20 years ago.
I believe those races have been discovered and fixed in upstream kernels, and I could probably find/fix them here with a few hours of fussing about.
There's also the possibility that the protocol bridges themselves may require slightly special handling, though I doubt it. Those things are not well tested on ancient, slow, PIO-only hosts (like the empeg) though.
Cheers
|
Top
|
|
|
|
#359836 - 26/09/2013 16:25
Re: Current HD->CompactFlash status?
[Re: drakino]
|
old hand
Registered: 17/01/2003
Posts: 998
|
Well it looks like HD number three, or maybe four, I kind of lost track, is on the way out. I have to use the old wack and spin procedure to get the HD to spin up.
I'd really like to use a non-mechanical drive this time. So I'm looking for advice. Is this the path I should go down or has anything newer and better come along?
I only need about 20G of space. I just want “whatever” to last more than a couple years. I’m not hard on the empeg???
|
Top
|
|
|
|
#359838 - 26/09/2013 16:47
Re: Current HD->CompactFlash status?
[Re: Redrum]
|
carpal tunnel
Registered: 18/01/2000
Posts: 5683
Loc: London, UK
|
Is this the path I should go down or has anything newer and better come along? You can get PATA SSD drives which just work, you know...
_________________________
-- roger
|
Top
|
|
|
|
#359839 - 26/09/2013 17:00
Re: Current HD->CompactFlash status?
[Re: Roger]
|
old hand
Registered: 17/01/2003
Posts: 998
|
Wow, thanks. I was not looking forward to "making something fit." Just drop in it you say... great. I sure have been out of the loop lately on things like this. I really should spend more time inside reading and not out in the sun
|
Top
|
|
|
|
#359890 - 07/10/2013 11:41
Re: Current HD->CompactFlash status?
[Re: Roger]
|
old hand
Registered: 17/01/2003
Posts: 998
|
OK, now I’m totally transistorized. I think these transistorized devices will really catch on. No issues at all on the install. Just trying to recall Linux commands was the biggest problem. I hope this lasts longer than the other drives. I plan on having my Jeep and empeg til I'm dead.
|
Top
|
|
|
|
#359893 - 07/10/2013 19:40
Re: Current HD->CompactFlash status?
[Re: Redrum]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14493
Loc: Canada
|
Speaking of which. I recently replaced the 128GB mSATA drive in my notebook with a 256GB mSATA drive. Which left me with the original 128GB mSATA. Dx.com and various eBay sellers have mSATA-to-IDE44 adapters, which would allow it to work inside an empeg.
So if you find yourself with a spare mSATA, that's one option.
Cheers
|
Top
|
|
|
|
|
|