Hi all. Wanted to post this to the group to see if anyone else has had a similar problem.

After reading a load of posts about Displayserver, I wanted to fire it up on my mk2. (Background: have had empegs since Mk1 serial: 00042. Recently started getting back into the cool extras of the hardware / software). I originally had Developer image 1.01 and installed ds2b2 with no problem (didn't add custom kernel). I found that I was having a problem with the getfid function, which are what the Index test1 and test2 links call. I upgraded to the developer image 1.03 and tried again (with the custom kernel), got the same results. Haven't tried the 2.0 beta images yet... Here are details of the problem.

When I click on Index test1 or Index test2, I get a blank page in the browser and error messages on the empeg over the serial line as follows:

ds2: memory violation at pc=0x40073894, lr=0x40073638 (bad address=0x00000000, c

ode 2)

pc : [<40073894>]    lr : [<40073638>]

sp : bfffd688  ip : 400ff764  fp : bfffd69c

r10: 00000000  r9 : 00000000  r8 : 00000000

r7 : 00000000  r6 : 00000000  r5 : 00000000  r4 : 00000000

r3 : 00000000  r2 : 020086c0  r1 : 00000000  r0 : 00000000

Flags: nZCv  IRQs on  FIQs on  Mode USER_32  Segment user

Control: C823D17D  Table: C823D17D  DAC: 00000015

Function entered at [<40073624>] from [<02004414>]

 r5 = 00000000  r4 = 00000000

Function entered at [<0200416c>] from [<02006808>]

 r10 = 400FF128  r9 = 02007F08  r8 = 02008230  r7 = 02000CB0

 r6 = 4000C2A8  r5 = 00000001  r4 = 02011EE4

Function entered at [<02004a14>] from [<020073a4>]

 r10 = 400FF128  r9 = 02007F08  r8 = 02008230  r7 = 02000CB0

 r6 = 4000C2A8  r5 = 00000001  r4 = 02011EE4

Function entered at [<020072a4>] from [<02007eb0>]

Function entered at [<02007b54>] from [<02008148>]

Function entered at [<02007f08>] from [<40037040>]

 r4 = BFFFFE24

Function entered at [<40036ee4>] from [<0200111c>]

 r10 = 4001D858  r9 = 00000000  r8 = 00000000  r7 = 00000000

 r6 = 020010F8  r5 = 00000000  r4 = 4001E5EC

I'm attaching the html output (actually just javascript). Interestingly, the next Playlist that was to be setup in the script was called
_Singles. I wondered about the _, so I took it out, then the output changed, in fact, it ended even earlier at fidinfo[18][8], which it got past before, not sure if that helps.

Thanks,
Chris

Crocklobster,

Which MkII do you have? Is it one of the original Empeg branded devices? If so, the problem is that you don't have enough memory. Some (or all) of the original Empeg branded MkIIs shipped with 12MBs of RAM as opposed to the Rio branded models that came with 16MBs.

The bad news is that there isn't a lot you can do. I've tried reducing the number of buffer space required by the player to their absolute minimum but no joy. I'd be interested to hear if you manage to get it working.. :)

I had the earlier version of Displayserver (1.1) working on my Mk2 player with the lower amount of memory. I haven't tried displayserver 2.0 on it yet.

Is Frank still offering Displayserver 1.1 for download from his web site? If so, Crocklobster could try that.

Note: try getting Displayserver working by itself first, without the replacement kernel. If that works, then move on to the kernel mod.

I do have the empeg version of the Mk2. So, is it not physically possible to up the ram? Would the empeg guys want to retrofit for cash? :-)

Are there any other differences between empeg Mk2 and the Rio? I might want to sell this one and get more ram.

Okay, going back to 1.1 gave me different results, but in some ways, the same.

I now actually get an html page, with a few playlists on it, but the page hangs and I get the following on the serial line:

displayserver: memory violation at pc=0x40073894, lr=0x40073638 (bad address=0x0

0000000, code 2)

pc : [<40073894>]    lr : [<40073638>]

sp : bfffefb4  ip : 400ff764  fp : bfffefc8

r10: 400ff128  r9 : 02003604  r8 : 02004020

r7 : 02000b50  r6 : 4000c2a8  r5 : 00000000  r4 : bffffe14

r3 : 00000000  r2 : 02004148  r1 : 00000000  r0 : 00000000

Flags: nZCv  IRQs on  FIQs on  Mode USER_32  Segment user

Control: D022D17D  Table: D022D17D  DAC: 00000015

Function entered at [<40073624>] from [<02002214>]

 r5 = 00000001  r4 = BFFFFE14

Function entered at [<02001e50>] from [<020024a0>]

Function entered at [<02002414>] from [<02003e5c>]

Function entered at [<02003604>] from [<40037040>]

 r4 = BFFFFE14

Function entered at [<40036ee4>] from [<02000f6c>]

 r10 = 4001D858  r9 = 00000000  r8 = 00000000  r7 = 00000000

 r6 = 02000F48  r5 = 00000000  r4 = 4001E5EC

It hangs on the playlist right after the "Singles" playlist. That list has 328 tunes, if that helps. And actually, the reason I get the html page just
seems
that it's setup differently, there's not a large javascript block at the begining of it.

Thanks for your input so far. I'd love to get this straightened out.


I've attached the resulting html page.

Chris

Have you done a hardware test, just to rule out bad memory?
I think the sequence of getting to the hardware test is Ctrl-T on empeg bootup, but I believe that Tony can validate this for me.

I had one of the original MkII's with the bad batch of Ram, and it actually didnt interfere with the player, only when synching (and that was intermittent until the memory completely went), so the memory could be affecting displayserver.

My guess, but then my guess may be out to lunch.

Details for performing a RAM test are here.

Thanks Tony and Smoker_Man for the suggestions.

It appears that my hardware is okay.

empeg-car bootstrap v1.00 20000601 ([email protected])
If there is anyone present who wants to upgrade the flash, let them speak now,
or forever hold their peace...
0000 empeg-car board test version 0.04
0100 ram test starting
0110 testing ic 1 (0-3mb, low word)
0120 testing ic 2 (0-3mb, high word)
0130 testing ic 3 (4-7mb, low word)
0140 testing ic 4 (4-7mb, high word)
0150 testing ic 19 (8-11mb, low word)
0160 testing ic 32 (8-11mb, high word)
01f0 ram test done
0200 ide detect
0210 command issued
0220 drive id TOSHIBA MK1814GAV
02f0 ide ok
0300 dsp/i2c test
03f0 dsp/i2c ok
0400 temperature detect
0410 temperature is 37
04f0 temperature ok
0500 usb detect
0510 usb rev 1012
05f0 usb ok
0600 ethernet detect
0610 ethernet rev 334b
06f0 ethernet ok
0700 cs4231 detect
0710 cs4231 rev a0
07f0 cs4231 ok
00f0 tests complete

I tried editing my config.ini file to reduce the size of the cache. I found some old info that said I could set
[startup]
ReduceCache=48

not sure if that's still valid or not, but that's what I tried. I didn't look at my system info before, but I'm attaching my current system info for review. Doesn't appear to change the displayserver results.

Starting to look like I might be ool, but that's strange, when others have been able to get it running. I wonder if it's my certain playlists or number of tunes (2556)

Well, it is certainly not the amount of files on your empeg. (I have a little bit more, ok a lot, and DSv1.1 worked for me)
Just to be clear, you are running player image version 1.03 developer, and DisplayServer 2b2 ??
I am having a funny feeling that DS2 was built for the 2.x series of empeg images, Frank can tell us more, but he seems to have returned to his homeworld for a while, so he is unavailable for comment. Perhaps try displayserver v1.1 ?? I know for a fact that it works, the install is relativly simple, and it works. The custom html stuff doesnt work, as that is a DS2 implementation. But for simple m3u streaming, use DS1.1

For the record, I did use Empeg Dev Image 1.03 and DSv1.1 ever since Dec/Jan 2000/2001, and have only recently upped to the 2.07Beta, hence losing DSv1.1
I have tried DS2, but well, we all know that emplode will not accept a mock *.mp3 file anymore, and untarring the archive reveals far too many files to upload (SecureCRT will do this nicely however). So I am still using DSv1.1 with the 2.07 Beta, which for me seems to be rock solid, so hopefully we will be going gold soon.

Oops, after re-reading the whole thread, I think what is happening is that DS is trying to generate the html pages, and is sort of puking at its size. This did happen to me, and I am going to list a few things that I checked and fixed to resolve this.
#1 Chache size, seems like you have already done this
#2 Check to make sure that DS put the drives back in ro, and check the init script to make sure it is trying to mount the proper partitions, there was an issue with this when Frank first published DS, he only had a single drive empeg, ad others with dual drives were having all sorts of funky issues, spreading across every conceivable problem. Turns out, Frank's mount/init script didnt have the proper partitions mounted for dual drives, so that caused a lot of grief. Also he estimated the numerical convention of partition names/numbers into his mount/init script, and empeg didnt follow natural convention (or did and we didnt, whatever)
#3 See it DS is launching it's process on every reboot, not a big deal since it hardly uses much memory (we would all like it to use more, but the greedy player likes ram, perhaps mlord and his hijack tweak can help Frank with the memory issues)
But it seems that if its in the startup script, it has a better chance of grabbing memory before the player software does. And if ds pukes, simply reboot the player.

This one isnt much of a fix, but in the root/parent playlist folder, I have about 5-8 choices; Artist, Genre, Mood, Party, etc. Then below that level it goes crazy, especially in the Artists playlist as all artists are in there, rather than in Segmented sections, A-D, E-H, etc.
However, even when I pop into the artists playlist, it does take a whole while to display the whole html page, and displayserver usually does output some sort of puke on the serial line.
But I usually dont give it much notice, as even if ds really does die, it seems that fids still fall out the ethernet port.

My thoughts, not yours. Please excuse the incoherent ramblings of an overworked man.

In reply to:

---

So I am still using DSv1.1 with the 2.07 Beta, ...

---

So let me get this straight: You're using version 1.1 of DisplayServer with 2.0b7 of the player software? I thought this didn't work with the 2.0bX versions. I prefer the earlier version, as you say it's rock solid. I've been having problems with the latest version of DS. I don't use it personally but the guys at work like to listen to my collection over the network. If the old version works then this is great news .

OK, I've tried DS1.1 with 2.0b7 and although it kinda works it's still unreliable. This wasn't the case when I was using it with 1.03. I'm having problems with Winamp timing out and then I'm not able to get any response from DS at all unless I reboot the player. Oh well I thought the problems I was having was with DS2 bugs - it appears not. I hope Frank comes back and does some more work on DS or the Empeg guys make some changes to enable the two to work together.

What I was actually running was developer image 1.03 with DS 1.1. Tony had said earlier that he had DS 1.1 working on 12 Mb ram units, so I switched to 1.1 after 2b2 failed. I got pretty much the same results with DS 1.1 that I did with 2b2.
(At the begining of the thread, I was using 2b2, hence the subject).

I guess with Smoker_Man's extra hints, I might try this again this weekend. Currently, I was playing with a debian installation on it (got ftp working, yet to get telnet working).

Now, what are the details of making the partitions writable and read only? When is it absolutely necessary to make the partitions ro again? Does the unit make them ro by itself sometimes? I have been just making them rw / rwm when I need to, and I don't worry much about making them ro again.

Smoker_Man, do you have a 12Mb unit? I'm glad I picked up a deep discount unit, so I can get that extra 4Mb.

Also, is the only "tuning" parameter the ReduceCache option? I saw somewhere else on the board mention of ReserveCache. Anyone know about that? Is the max value of ReduceCache really 48? I've seen some people trying 64.

I'll check the things you mentioned. Your last tip intrigued me. When I try to get my artists playlist up, that's when it bombs. Of course, I don't have anything encompasing my artists playlist, it's right under the empeg root. It didn't look at all like the fids were still coming down, and that I just had to wait for the page. I'll check that though. I thought what I saw on the serial line was a seg fault and hence a program crash, which would abort that html page send. You can look at some of my earlier attachments for what I get through the browser.

Chris

Yes, I have a 12mb unit, and damn if I want another as a backup/second unit for mixing. And afaik returning the partitions to ro makes things happier, I will not pretend to know why, but that seems to be the way the player software likes things. *shrug*
I too did get a bunch of stuff streaming off the serial line when I run across long playlists, but I am thinking that putting ds in the startup seems to keep it running/persistent even when it dies (I am likely very, very wrong on this, but *shrug*).
I dont know why, but ds seemed to puke on me only when A) 4+ people where browsing the playlists B) 4+ poeple were listening/browsing and I was doing stuff like quickly skipping songs. So ds was/is fairly stable for my use.

But I do agree with presenting this stuff to Frank, whenever he returns.

Hi.
Also, is the only "tuning" parameter the ReduceCache option? I saw somewhere else on the board mention of ReserveCache. Anyone know about that? Is the max value of ReduceCache really 48? I've seen some people trying 64.

The parameter in question is not ReduceCache, but ReserveCache. This is an information that Hugo (altman) confirmed in a private message to me. Check riocar.org developer info for details. The (theoretical) maximum value for this parameter is 179, effectively disabling precaching of music data (leaving only a single Cache chunk, at least on a MkII (12MB) with a 1.03 developer image). Any number higher than 80 is probably too high. Each cache chunk is (at least for 1.03) about 32kBytes in size, so ReserveCache=32 reduces the cache size by (a bit more than) 1MB.

cu,
sven