#288991 - 27/10/2006 03:31
Hijack v464: Required for drives larger than 132GB
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Hijack v464 is now available. This version fixes a rather nasty bug in the (previously untested) LBA48 drive support. As such, v464 is the first empeg kernel that functions correctly with drives larger than 132GB in capacity. Do not attempt to use such large drives with any kernel older than Hijack v464. And especially not with the stock player kernels -- doing so will corrupt data (messes up your tunes!). Along with this release, I have also updated my custom bigdisk builder image to include v464. Similarly, I have also customized some player software installers (the .upgrade files) to include Hijack v464, making it possible to install the software on large drives. The updated builder and software installers are temporarily available from here. Cheers
|
|
Top
|
|
|
|
#288992 - 27/10/2006 10:40
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
pooh-bah
Registered: 13/01/2002
Posts: 1649
Loc: Louisiana, USA
|
Awesome! You really know your stuff. I owe you one. I'll let you know how it works later today.
Thanks for the incredible support,
Stu
_________________________
If you want it to break, buy Sony!
|
|
Top
|
|
|
|
|
#288993 - 27/10/2006 12:05
Re: Hijack v464: Required for drives larger than 132GB
[Re: maczrool]
|
pooh-bah
Registered: 13/01/2002
Posts: 1649
Loc: Louisiana, USA
|
Okay, I tried it on an MK2 drive with the 160GB drive and still no joy.  It seems to be failing sooner now. Weird. I'll try it on an MK2A unit and see if the extra 4 megs makes a difference. I've attached a boot log (from Hyperterminal sorry). Any more ideas or should I just take a crack at manually building it? Stu
_________________________
If you want it to break, buy Sony!
|
|
Top
|
|
|
|
|
#288994 - 27/10/2006 13:06
Re: Hijack v464: Required for drives larger than 132GB
[Re: maczrool]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
A manual build should not give any different results here, because the error is at the device-driver level in the kernel. It is possible that Hijack's minimalist LBA48 implementation is insufficient for that drive -- dunno. If you can get a shell prompt, then send me the output of "cat /proc/ide/hda/identify".
Cheers
|
|
Top
|
|
|
|
|
#288995 - 27/10/2006 13:08
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Actually, a readable boot log would *really* help a lot here.. The one you posted is utter rubbish.
Cheers
|
|
Top
|
|
|
|
|
#288996 - 27/10/2006 13:43
Re: Hijack v464: Required for drives larger than 132GB
[Re: maczrool]
|
carpal tunnel
Registered: 25/12/2000
Posts: 16706
Loc: Raleigh, NC US
|
As posted in the FAQ, you can get around HyperTerm's copy-and-paste bug by writing a log of the session instead.
_________________________
Bitt Faulk
|
|
Top
|
|
|
|
|
#288997 - 27/10/2006 20:00
Re: Hijack v464: Required for drives larger than 132GB
[Re: wfaulk]
|
pooh-bah
Registered: 13/01/2002
Posts: 1649
Loc: Louisiana, USA
|
No need now.  I reran the new builder on the MK2A that the old builder failed on and there was much joy! It's either the extra four megs or the MK2 I was using is ailing and I don't know it yet. So anyway, all seems well. Many thanks, Stu
_________________________
If you want it to break, buy Sony!
|
|
Top
|
|
|
|
|
#288998 - 28/10/2006 18:17
Re: Hijack v464: Required for drives larger than 132GB
[Re: maczrool]
|
pooh-bah
Registered: 19/09/2002
Posts: 2494
Loc: East Coast, USA
|
So that's extra 4 megs running 2.0 and not 2.01, right? Or do you need 2.01 (which is broken?) to benefit?
_________________________
- FireFox31 110gig MKIIa (30+80), Eutronix lights, 32 meg stacked RAM, Filener orange gel lens, Greenlights Lit Buttons green set
|
|
Top
|
|
|
|
|
#288999 - 28/10/2006 18:38
Re: Hijack v464: Required for drives larger than 132GB
[Re: FireFox31]
|
pooh-bah
Registered: 13/01/2002
Posts: 1649
Loc: Louisiana, USA
|
I used Mark's builder and it wouldn't work on the MK2 and it would on the MK2A. Running the built drive with MK2 and the 2.01 with Hijack preinstalled, the MK2 works fine even with a 160 and an 80 gigabyte in there.
Stu
_________________________
If you want it to break, buy Sony!
|
|
Top
|
|
|
|
|
#289000 - 29/10/2006 00:46
Re: Hijack v464: Required for drives larger than 132GB
[Re: FireFox31]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
2.01 (which is broken?)
??????
In what way is 2.01 "broken" ??
|
|
Top
|
|
|
|
|
#289001 - 29/10/2006 12:48
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
pooh-bah
Registered: 19/09/2002
Posts: 2494
Loc: East Coast, USA
|
I'd read that 2.01 with 64 meg had too much memory for the cache so the player choked trying to keep it always full. Or does Hijack v464 not require the full 64 meg for the larger drives? Is there a certain memory requirement?
_________________________
- FireFox31 110gig MKIIa (30+80), Eutronix lights, 32 meg stacked RAM, Filener orange gel lens, Greenlights Lit Buttons green set
|
|
Top
|
|
|
|
|
#289002 - 29/10/2006 13:12
Re: Hijack v464: Required for drives larger than 132GB
[Re: FireFox31]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
I think the "broken" part there is more the 64MB of RAM, rather than the player software that tries to use it. If using v2.00, then there's no point in having 64MB of RAM in the first place. Sure, one or two small apps might benefit from a couple of extra MB, but not 48MB extra.
-ml
|
|
Top
|
|
|
|
|
#289003 - 29/10/2006 13:43
Re: Hijack v464: Required for drives larger than 132GB
[Re: FireFox31]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
I'd read that 2.01 with 64 meg had too much memory for the cache so the player choked trying to keep it always full. Or does Hijack v464 not require the full 64 meg for the larger drives? Is there a certain memory requirement?
If a player has the standard amount of RAM, then v2.01 should behave just about identically to v2.00. Not broken.
If a player has more than 16MB of RAM, v2.00 won't use it.
If a player has more than 16MB of RAM, v2.01 will try to use all of it, which can be bad.
The player uses the inefficient read(2) system call to fetch data from disk. The way that read(2) works is that first the kernel allocates RAM (in 4KB chunks) into which the data is read from disk, and then the kernel copies from there to the player's buffers. Thus, read(2) can require twice as much memory as the size of data being read. This could be avoided if the player used mmap(2) instead of read(2).
Since reading can use up to double the data size for memory, this will cause problems if the player grabs all of the extra memory for itself, leaving none for the kernel's 4KB buffers of the same data.
So, a rough solution would be to allot no more than 50% of extra memory to the player software, leaving the other 50% for the kernel and other applications to divy up.
There's nothing magic about the 50%, other than that it matches the theoretical worst case. In practice, the kernel does not actually need a full 1:1 ratio for internal 4KB pages versus player data.
Paul has done something like this, using his ReserveCache=180 flag on his 48MB player. Now, 48MB is 32MB of extra RAM. 50% of that would be 16MB. In practice, Paul is letting his player use somewhat more than 50% of it: Each ReserveCache increment is 64KB, so 180 * 64KB = about 11.5MB, or about 35% of the extra RAM is left to the kernel, with the player taking the rest.
So.. with a 32MB player example, one has 16MB of extra RAM. With the rule of Paul, this means a ReserveCache=90 might be a good starting point. Bump it up in increments of (say..) 8 until things work reliably.
For a 64MB player, there's 48MB of extra RAM. The same rule would suggest using ReserveCache=270 as a good starting value. Dunno.
Cheers
Edited by mlord (29/10/2006 13:51)
|
|
Top
|
|
|
|
|
#289004 - 29/10/2006 20:23
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 15/08/2000
Posts: 4859
Loc: New Jersey, USA
|
"The rule of Paul"??? This works in almost every single player I have seen. Except one... I currently run with 2.01 with 64MB and a ReserveCache of 180 without an issue.
_________________________
Paul Grzelak
200GB with 48MB RAM, Illuminated Buttons and Digital Outputs
|
|
Top
|
|
|
|
|
#289005 - 30/10/2006 13:55
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 18/06/2001
Posts: 2504
Loc: Roma, Italy
|
THANK YOU, Mark!
Will we need to run the Set_Empeg_Max_FID on the empeg, after rebuilding a disk with this?
_________________________
= Taym =
MK2a #040103216 * 100Gb *All/Colors* Radio * 3.0a11 * Hijack = taympeg
|
|
Top
|
|
|
|
|
#289006 - 30/10/2006 14:22
Re: Hijack v464: Required for drives larger than 132GB
[Re: Taym]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
THANK YOU, Mark!
Will we need to run the Set_Empeg_Max_FID on the empeg, after rebuilding a disk with this?
Oh.. Yes, one will.
I suppose I could have done it already to the player binary when making those images, but I did not. That's probably a wise accidental choice  because the set_max_fid thing does alter the on-disk format of some of the information stored by the player s/w, and some people might not be happy about it (okay, dubious but possible, I suppose).
What does everyone think about it? Should the install images include the max_fid changes by default? It would be rather convenient, I suppose..
???
|
|
Top
|
|
|
|
|
#289007 - 30/10/2006 15:31
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 18/06/2001
Posts: 2504
Loc: Roma, Italy
|
Of course the implicit meaning in my question was that I personally think it would be great if it was included, so I vote for it!  Thanks again, Mark. It is thanks to these great updates that an already excellent device like the empeg is still unbeated, after so many years.
_________________________
= Taym =
MK2a #040103216 * 100Gb *All/Colors* Radio * 3.0a11 * Hijack = taympeg
|
|
Top
|
|
|
|
|
#289008 - 30/10/2006 15:39
Re: Hijack v464: Required for drives larger than 132GB
[Re: Taym]
|
carpal tunnel
Registered: 19/01/2002
Posts: 3583
Loc: Columbus, OH
|
unbeaten too!
What are the negative effects, if any, to having the maxfid patch running on a 240 gig player with only 5 gig of mp3s or something silly like that?
_________________________
~ John
|
|
Top
|
|
|
|
|
#289009 - 30/10/2006 16:23
Re: Hijack v464: Required for drives larger than 132GB
[Re: JBjorgen]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
unbeaten too!
What are the negative effects, if any, to having the maxfid patch running on a 240 gig player with only 5 gig of mp3s or something silly like that?
There aren't any, really -- well, I think one of the bookmark lists gets shorter or something.
The only real maybe-issue, is that the on-disk locations of things like the "seek profiles" for each track will change. So if somebody has carefully built-up profiles (for use with the SeekTool INFO screen), then those get lost and will have to be reacquired over time.
-ml
|
|
Top
|
|
|
|
|
#289010 - 30/10/2006 16:31
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 18/01/2000
Posts: 5682
Loc: London, UK
|
Quote:
those get lost and will have to be reacquired over time.
As long as the EQ and tuner settings are preserved, I'll vote +1.
_________________________
-- roger
|
|
Top
|
|
|
|
|
#289011 - 30/10/2006 16:34
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 19/01/2002
Posts: 3583
Loc: Columbus, OH
|
Hmm...since I don't use (and never have used) seek profiles, I'll add another vote toward including the patch.
The likelihood of me ever needing it is slim, though.
_________________________
~ John
|
|
Top
|
|
|
|
|
#289012 - 30/10/2006 16:36
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
pooh-bah
Registered: 12/02/2002
Posts: 2298
Loc: Berkeley, California
|
We're talking about integrating it into the disk builder image, right? So the player is going to lose all the cached seek info anyways, right? Even if we're talking about hijack, I'd much rather lose all the seek info once and never have to worry about the maxfid patch.
Matthew
|
|
Top
|
|
|
|
|
#289013 - 30/10/2006 17:15
Re: Hijack v464: Required for drives larger than 132GB
[Re: matthew_k]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
We're talking about integrating it into the disk builder image, right?
No, but rather into the .upgrade files for v3a11, v2.01, and v2b13.
Mind you, those are really most likely to only be used on new drives nowadays anyway, so no loss whatsoever I suppose.
And what about our fidsift.sh buddy.. ?
Cheers
|
|
Top
|
|
|
|
|
#289014 - 30/10/2006 17:26
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
addict
Registered: 27/10/2002
Posts: 568
|
But aren't the number from Set_Empeg_Max_FID depending on the disk, so it could change from one player to the next?
Stig
|
|
Top
|
|
|
|
|
#289015 - 30/10/2006 17:28
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
pooh-bah
Registered: 12/02/2002
Posts: 2298
Loc: Berkeley, California
|
Throw it in too, definitely. Does it still need to be rerun after adding music with 2.0 software? If so, any chance of having it be available from the hijack menu?
Matthew
|
|
Top
|
|
|
|
|
#289016 - 30/10/2006 17:30
Re: Hijack v464: Required for drives larger than 132GB
[Re: matthew_k]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
Throw it in too, definitely. Does it still need to be rerun after adding music with 2.0 software? If so, any chance of having it be available from the hijack menu?
I think it is still needed in some cases with v.20[1] software. And yes, It can be made available from the Hijack menu. In fact, you can do it yourself, I believe, with a config.ini MENUEXEC line and some Bitt wizardry to stop/restart the player.
Bitt ?
Cheers
|
|
Top
|
|
|
|
|
#289017 - 30/10/2006 17:39
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
pooh-bah
Registered: 12/02/2002
Posts: 2298
Loc: Berkeley, California
|
Any chance the bit wizardry could be included in the standard builder created config.ini? I suspect emplode creates the config.ini, so perhaps that wouldn't work.
|
|
Top
|
|
|
|
|
#289018 - 30/10/2006 19:45
Re: Hijack v464: Required for drives larger than 132GB
[Re: mlord]
|
carpal tunnel
Registered: 13/07/2000
Posts: 4174
Loc: Cambridge, England
|
Quote:
The only real maybe-issue, is that the on-disk locations of things like the "seek profiles" for each track will change.
Is that really true? I thought all the patch did was move the last two bookmarks into an unused area, extend the first two bookmarks into the space freed-up by that, and extend the per-track data at the end? None of that would invalidate existing seek profiles; all you'd lose is the bookmarks. Or do later versions of the patch do more than that?
Peter
|
|
Top
|
|
|
|
|
#289019 - 30/10/2006 20:04
Re: Hijack v464: Required for drives larger than 132GB
[Re: peter]
|
carpal tunnel
Registered: 29/08/2000
Posts: 14484
Loc: Canada
|
Quote:
Quote:
The only real maybe-issue, is that the on-disk locations of things like the "seek profiles" for each track will change.
Is that really true? I thought all the patch did was move the last two bookmarks into an unused area, extend the first two bookmarks into the space freed-up by that, and extend the per-track data at the end? None of that would invalidate existing seek profiles; all you'd lose is the bookmarks. Or do later versions of the patch do more than that?
Peter
I think your memory is better than mine (and I'm too busy/lazy right now to go and re-read what it actually does!).
So, all we lose are the current bookmarks, then. I think.
Cheers
|
|
Top
|
|
|
|
|
|
Previous Topic
Index