Most of the time, it will be barely noticeable, but some of the time it could lock up the player for as long as 19 minutes or so (32-bit timer at 3.6864mhz) -- which is pretty much what people are reporting here.

Yeah, it is a bit spooky how everyone whose clock lurches, lurches by the same amount.

But a little more data would be nice to confirm all of this. So, I'll put out v222 this morning, WITHOUT the fix, but with more data collection for when it happens.

By the way, we ought to thank you really for all the work you're putting in to help us debug our own code and fix what's probably the worst outstanding bug in the 2.0 betas. So, thanks!

Peter

Here's the question.... Is the code you're fixing NEW in the 2.0x betas? Because there were no reported lockups back in the 1.03 days... So I would think that whatever the real bug is has to be in code that was introduced for 2.0... Right?

Dunno, but this particular bug has probably been there all along. I wonder if the frequency of FIQs has increased since the 1.03 stream?

Something like this gets influenced by all sorts of subtle behaviours in process priorities and whatnot, and it is entirely possible for the triggering interval to have been much lower in the past. Or not.

Also, does it just seem more common now, with about triple the number of player owners out there and posting to this BBS, happily coinciding with the v2 betas, or did it REALLY never occur before?

Remember, it probably happens a lot more than we think, cuz we've gotta be pushing buttons to notice it.

???

Actually, the FIQ's have a lot to do with the presence of a Tuner module.. I'm not that familiar with all of this, but my understanding is that without a Tuner they are less common (or not at all)? And the v2 betas began not long after Tuner modules began shipping, so there wasn't a lot of overlap time with v1.03 and Tuners (and FIQs?).

And without FIQs, the bug I'm looking at won't occur.

So.. what are FIQs used for, anyway?

Eh... I dunno, I agree that that the 2.0 betas coincided with a massive increase in the population, but we had a pretty large sample size to begin with during the 1.x series, and I don't think there were *ever* any reports of any sort of lockups of this nature. That was a long period of time, too... However, the other possibilities you mention (higher frequency of FIQs, changes in various timings/processes, etc) are all quite plausible. I'm 100% in favor of any effort to try to fix this problem, and this sounds promising. The lockups and the buggy info mode setting are the only two things that make running 2.0b11 painful for me, so solving this problem would be a huge win.

The price drop and 2.00beta3 both happened in December right? That is when the membership skyrocketed AND 2.00beta hit. I remember making a post along those lines back then.

Okay, Hijack v222 is out.

slightly improved "lockup messages". These now say stuck jiffies 5 or unstuck jiffies 8. The numbers will change, but they provide important information. Please report them here if you ever see either of these messages.

the Reboot Machine menu item now remounts all drives read-only before rebooting. This has been missing for a while, and I finally remembered to implement it. The remote "REBOOT" command (FTP,HTTP,/proc) has had this since day-1.

the bugfix for the jiffies reset is not included yet, but it will be in v223 later on. I suspect this will fix nearly all player lockups, but we'd like to collect a little more data first.

Upcoming features (sometime in March'2000):

ir_translate capability for Sony Stalk (this one's been simmering for a while, as I'm still not 100% sure how to present it, and have been distracted by the barbarians pounding on the gates).

The jiffies bugfix

Tuner ID detection at power-on, plus a menu extension to make use of it for AC/DC override.

-ml

>By the way, we ought to thank you really for all the work you're
>putting in to help us debug our own code and fix what's probably
>the worst outstanding bug in the 2.0 betas. So, thanks!

Thanks, Peter. But what I really want is a serial port command to "append FID" and another to "insert FID", similar to the existing "replace FID" command (#xxxx). So we would have:

   #xxxx     == replace currently playing FID (as today)

   #xxxx-    == insert FID to current running playlist

   #xxxx+    == append FID to current running playlist

If this ends up solving the lockups, I think that would be a fair trade.

Cheers

-ml

Upcoming features (sometime in March'2000):

• ir_translate capability for Sony Stalk (this one's been simmering for a while, as I'm still not 100% sure how to present it, and have been distracted by the barbarians pounding on the gates).
• The jiffies bugfix
• Tuner ID detection at power-on, plus a menu extension to make use of it for AC/DC override.
  

[nag]

• mmap() buffer for Hijack display writes
  

[/nag]

Hell yeah... I think that's a wonderful trade.

If this ends up solving the lockups, I think that would be a fair trade.

LOL

ok, it wasn't exactly difficult to implement. It'll be in the next beta.

Peter

Suh WHEET.

Also, does it just seem more common now, with about triple the number of player owners out there and posting to this BBS, happily coinciding with the v2 betas, or did it REALLY never occur before?

I don't know about other people, but I don't remember any lockups occuring on my player with 1.0x, but with 2.0x I was pretty much guranteed a lockup once a day. The difference really was that noticable.

Good. Then we need YOU to run Hijack v223 immediately. Let us know if you still get lockouts.

Hugo has rightly pointed out that this fix doesn't "make sense" in some cases, but perhaps another assumption is wrong somewhere.

Or maybe it won't help at all. Let's find out.

-ml

Already have done, I was sitting there hitting "F5" waiting for it to appear. No lockups yet...

Was this on your Mk1, or your Mk2 ?

-ml

My Mk2a, I have not used my Mk1 much since 2.0xbeta appeared, not enough to see any lockups.

FIQs have always been used on the mk1 for IR (well, definitely since v1).

On the mk2, FIQs are only used when sampling the audio inputs - eg, in aux or tuner mode, to do the visuals. In MP3 mode there is no sampling active and so there should be no fiqs generated.

Hugo

I've also only had lockups since the 2.0 betas. I'm installing Hijack v223 right now. I'll let you know if this cures it.

Edit: btw I get several lockups per day.

OK. I've just got a "stuck jiffies" error come up and the player seems to have frozen. The number after "stuck jiffies" is incrementing. It's still playing music and the hd is being accessed ok but no response from the buttons or the remote. Hope this helps.

Update: it now says "unstuck jiffies" and the counter has stopped at 44602. I can see the visual running in the background now and the player responds to the buttons and most of the remote buttons. Strangely the "Previous" and "Next" track buttons on the remote aren't responding. I've just rebooted (pulled the plug) and the "Previous" and "Next" track buttons on the remote are now working again.

Mark, I'm running 221 that you released last night. Unfortunately, the one day where I want the player to lock up, it hasn't.

More background in case it's been lost in the shuffle of other threads. Player is an MK2a with b11 and two 30GB drives (5950 tunes). I've been running fairly recent versions of Hijack. Player runs in the car randomizing the entire playlist with the least recently played mode. Always running Info:Track display. No tuner connected. Volume locked at 0db. No EQ or other DSP settings. I skip tracks quite a bit in the mornings and evenings during a 1 hour drive each way. So I tend to notice when the player has locked.

The problem has, without question, beein happening more regularly with b11 than it ever did in the past. It also happened with b7, but not with this frequency.

I'm not sure what level of fixes you've decided top put or leave in the latest version, but if you make sure to leave in whatever debug info/display you think will help you, I'll make sure to note it down when it finally locks.

If you do get a permanent fix, you might consider a slightly more substantial trade request from the empeg team... I'm sure you can vertical think of something vertical you might want to vertical aditionally ask for.

Bruno

If you do get a permanent fix, you might consider a slightly more substantial trade request from the empeg team... I'm sure you can vertical think of something vertical you might want to vertical aditionally ask for.

Not a chance.

Rob

As long as soup views are vertical, I'm cool. And I really hope you can drill down through soup entries in Artist and Album (PLEASE! )

Bruno

Stuck Jiffies 44596, about 20 playing hours after installing Hijack (approx 40% in car).

Took about 18-20 minutes before unsticking.

Mk2a, 2.0b11, Hijack v222.

>Stuck Jiffies 44596, about 20 playing hours after installing Hijack
> (approx 40% in car).

>Took about 18-20 minutes before unsticking.

Very useful information. The 18-20 minutes is actually 19 minutes and 53 seconds, the time required for the jiffies counter to wrap around a full 32-bit rotation at 3.6Mhz.

V225 should fix that problem (the bug is not Hijack's, but the fix is). Please report back if you ever see it again while running Hijack >= v225.

Thanks