Hijack has just turned 300. Let's party!

v300 includes:

-- Mike Comb's 2.2.17 kernel upgrade patch
-- Mike Comb's ext3 patch (off by default in Hijack)
-- genixia's buffer overflow fix for the remount function
-- new "insert" and "append" options in the khttpd playlists

I think the mk1 kernels are slightly larger now, and mk2 kernels are smaller.
The .patch file is MASSIVE (and still uploading.. gotta think about a smaller diff next time).

Enjoy! It's party time!

-ml

Happy Birthday Hijack!

Hope this doesn't make it a short birthday, but: What about JaHarkes' patches that make the long presses work properly under his GPSapp?

*puts on a party hat and waves*

"drunk from too much punch, thinks laura's waving at self.
Waves back, starts to walk over then realizes Laura was waving at Mark.
Makes sharp right to cake table... "

*puts some brownies next to the cake*

Can't have a party without brownies.

Alright!

I'm gonna get some Guinness and celebrate by converting to ext3...

The button fix for gpsapp is in v300 as well.

Party on, Dude!

Rock on. Thanks, Mark.

And for those who are curious, the REAL birthday of Hijack was on 19-Oct-2001, with this posting:

http://empeg.comms.net/php/showthreaded.php?Cat=&Board=empeg_general&Number=41436&page=&view=&sb=&o=

And why I ever thought to use the word "Hijack" for later names, I'll never know.. meanwhile, I get hand searched every time I get onboard an aircraft in the USA.. 28 flights in a row and counting..

Doh!

I suppose this could be extended considerably, to "hijack" the front panel from the player (while the player is running!), to allow use to add our own custom menus and whatever in a more integrated fashion..


...those words look so.....innocent.

meanwhile, I get hand searched every time I get onboard an aircraft in the USA

Same here on my flights for the empeg meet. It seems the metal detectors are set so high now, that an average pair of jeans set them off.

http://www.theonion.com/onion3838/faa_passenger_ban.html

Cheers! Thanks for getting all my changes integrated into your tree. For anybody who wants to give ext3 a shot you can grab kernels compiled with the ext3 option on as well as the other needed tools and instructions from my site.

On a related note can somebody tell me what if any sources I should be hosting along with those binaries to comply with the GPL? I am a little fuzzy on how that works, am I supposed to provide a copy of hijack, the kernel, and e2fsutils or can I just link to them?

Thanks,
-Mike

Heh, you could just take down all of your copies and let Mark worry about it.

In reply to:

---

Heh, you could just take down all of your copies and let Mark worry about it.

---

Well, somebody has to host the binaries with ext3 enabled since Mark isn't turning it on by default. As long as I have those binaries I believe I have some obligations to post or link to the source as well. I am sure nobody will ever bother downloading it from me, but I still want to be compliant with the GPL.

-Mike

Thanks for another great tweak.

How does one enable the ext3 patch?

Thank you,

Anything you haven't patched yourself (i.e. things you got in binary format) simply provide the url where the source came from. So you could have a link to www.debian.org for the e2fsutils stuff.

If you patched something, you just have to promise to give anyone who asks for the source a way to obtain it, but you are allowed to charge for the cost of media/distribution. So you could offset to send a CD-R with the sources for 5 bucks, or whatever could be considered a reasonable cost for the media/duplication/distribution.

A lot of GPL apps don't really follow it completely to the letter. f.i. any interactive application 'must' display an appriopriate copyright notice, that there is no warranty, that users may redistribute the program under the conditions of the GPL and how to obtain a copy of said GPL. However, I would consider something like that an annoying and useless 'nag' screen, and most interactive applications that are GPL'd do not bother either.

Does that invalidate the license? Perhaps, but without the license normal copyright law applies, so nobody except for the original author would be allowed to modify or redistribute the application. And as soon as multiple people contribute to a project, there is no single original author. So implicily everyone agrees that the license is in fact valid and applies.

When using the logo editor to update, i get

found empeg unit: entering program mode
manufacturer=0089, product=88c1
waiting for prompt
starting erase [100%] erase ok
starting program at 0x10000 [Failed to get echo-back on lock/unlock command.

Error: lockpage(10000,1) got code 1

Press the Enter key to continue.


Logo editor is version 1.4

Flashes back to 299 no problem

In reply to:

---

How does one enable the ext3 patch?

---

If you are compiling your own kernel just add CONFIG_EXT3_FS=y to your kernel config. Otherwise, follow the link for binaries and (not that great) setup instructions.

-Mike

Forgot to note that it updates with jemplode fine

Happy Birthday for Hijack !!!!


He mak : What a long way since the 100th HJ .....

Birthday greetings!

are the e2fsutils backwards compatible? so i don't have to worry if my partitions are ext2 still.

No no.. not the same at all. I pass through the security checkpoints with no problems -- no metal to set off detectors.

It's when I try to actually board the plane that I get pulled aside, 100% of the time. Actually, it really happens at check-in -- my boarding card gets flagged with some kind of special marking indicating that I must be hand-checked.

A bloody nuisance when trying to catch a tight connection in, say, Chicago -- I get searched on boarding my first flight to ORD, and then again as I try to board my connection in ORD. Same thing on the return flights a few days later.

100%. Not just a few flights in a row, but 28 (and counting.. 8 more next month).

Stupid bloody buggers. They could be searching somebody else, instead of wasting their time on me four times a week.

-ml

What is it that makes them suspicious?

Probably his hard hat, the huge amounts of rope he carries and that sheepish grin.


Z~

Oh man...I was hoping that you were exaggerating. That sucks.

When using the logo editor to update, i get (...)
starting program at 0x10000 [Failed to get echo-back on lock/unlock command.

Please see the extensive thread in "Bug Reports" on this exact error.

Mark, if you have a moment, could you peek at it and see if you can offer any insight?

Hijack is a top ten google result for "hijack". I'm actualy fairly impressed that they're together enough to connect you with it every time you fly. However, I'm not too impressed by the fact that they find it a reason to search you every time you fly... Have you looked into what number to call to get yourself off the list of known terrorists?

Matthew

>Have you looked into what number to call to get yourself off the list of known terrorists?

I don't have any idea of where to even starting looking for such.

Though I will try to pry another tidbit of info from the check-in clerks next month.

Cheers

That would be similar to trying to unsubscribe from a spam list.

The .patch file is MASSIVE

Why don't you bzip2 it? That takes 89% off.

Pim

Funny,
I just had to verify that hijack really shows up on googles first page.
And what happens ?
Five minutes later I get an (automated) email from security about:
"Reconsidering if researching certain terms [hijack] on the web can really be justified by business needs."
(The mail was _much_ longer...)

I guess they're just a little paranoid.
And I should bypass that proxy.

On a second thought, I wonder if they will log this message, since it contains the word [hijack]....


Oh yes.
@hijack: Happy Birthday !
@Mark: keep up the good work!

In reply to:

---

are the e2fsutils backwards compatible? so i don't have to worry if my partitions are ext2 still.

---

Yes. I still run my root and programs partitions as ext2 with no problems.

-Mike

Hey Mike, I upgraded my music and /usr partitions to ext3 no problem. I was thinking about doing the same with my root partition but I got this nagging feeling that it might be a bad idea. Any reason not to convert the root partition too? I know that future .upgrades will be laying down an ext2 filesystem, so they'd have to be converted after each upgrade. I don't mind doing that... Is there any other reason that upgrading the root partition to ext3 is a bad idea?

In reply to:

---

Is there any other reason that upgrading the root partition to ext3 is a bad idea?

---

Good question. I haven't tried it myself, but I don't think it will work. The root partition is mounted directly by the kernel IIRC and it does not seem to use the normal mount call that I tweaked to get the other partitions to magically mount as ext3. I think you would have to hack up the kernel a bit more to find where and how it mounts / and change that call. I didn't bother since that partition is small enough that an fsck doesn't take any real time. Also, since the root partition is so small I wouldn't want to waste the space that would be taken up by the journal file. Anyway, YMMV but I wouldn't recommend it.

-Mike

Okay, that makes sense. Does a journal file really take up that much space, BTW?

The only reason I wanted to try is that my root partition and my /usr partition seem to be the ones I always screw up and forget to mount read-only.

In reply to:

---

Does a journal file really take up that much space, BTW?

---

The amount of space used is relative to the size of the partition, but I don't know if there is a minimum amount that is always used. As a reference tune2fs created a 30Meg journal file for my 30Gig drive in my main empeg.

-Mike

<slightly tardy>
mirror at Empeg-Hijack.com updated for anyone who has trouble getting to sourceforge.
</slightly tardy>

In reply to:

---

Stupid bloody buggers. They could be searching somebody else, instead of wasting their time on me four times a week.

---

But look it from their [computer systems] point of view:

You are a known "Hijacker", you associate with "hijackers" and you're a foreigner [Canadian] who travels a lot in USA.

Therefore according to their "profile assessments" of what average "hijacker" looks like - you must rate right up there with Osama and his mates in terms of security risk!

Naw, I figure they're just hassling me cuz I'm "taking jobs 'way from 'mericans!", or somethin' like that.

Cheers

It's when I try to actually board the plane that I get pulled aside, 100% of the time. Actually, it really happens at check-in -- my boarding card gets flagged with some kind of special marking indicating that I must be hand-checked.

Ahh, ok. I got randomly selected flying out of COS, but never on the way home. My friend who flew recently got randomly picked for this search at the gate as well. Thats really odd that you have been nailed every single time.

I have upgraded to Hijack v.3.00 and It say here that
"the journaling file system stuff is included with this version",
but "not turned on by default"

How do I "turn it on"?
I followed the link to http://macgeek.dyndns.org/empeg/ext3/
and it reads "All the patches below have been integrated into hijack"

So I am assuming that there is nothing to apply to the player,
I just don't seem to be able to figure out how to actually see the benefit.
Is it automatically setup up for me by applying Hikack v3.00
or do I have to issue a bunch of commands to the player to actually implement this type of filesystem?
If so, what, specifically are those commands?

Did I miss a FAQ entry or do I smell a new one coming?

Also, what, if any are the benefits?
I know that fsking is cut out, but are there any other performance advantages?
Does this have anything to do with beta 13's new way of organizing FID's ?

P.S. Sorry for all the questions but I am a linux newbie, so please, use kid gloves, thanks

If you're a newbie to Linux, I recommend that you do nothing with the new ext3 features in the kernel. There will not be any immediate benefit to you.

The biggest benefit is "no disk checks", but that is already there in Hijack because it has a menu option "Filesystem check on sync=Disabled" that you can use.

As a software developer, there might be some benefits to using ext3 on the player, but if you're just a regular user then you shouldn't convert your file system.

At the moment I agree with Tony. But ext3 might have benefits for non-developers in the future. If we ever develop any applications that want to write to disk whilst in the car, then ext3 has major benefits over ext2.

I smell something brewing

I don't think you quite follow that there's a good amount of difference between the don't check option and what ext3 provides. Even if you do, it's worth pointing out in more detail.

The normal filesystem, ext2, keeps a log of how many times it's been mounted. Once it reaches a certain number, it ``requests'' that it be checked on the assumption that it might have become corrupted without it noticing. For many applications, the empeg included, this is useless. The don't check option turns this off, so you won't get the occasional slow empeg start.

Ext3, on the other hand, is a journalling filesystem. This means that every time a change is made to the filesystem, a log is written that says the filesystem is about to be modified, then the change is made, then the log is erased. (I'm sure that this is not totally accurate, but it's generally the idea.) The benefit of this is that if the filesystem is forcibly unmounted, as from the power being removed, the filesystem driver can see if there are any logs that weren't erased. If there were, then it can find the partial change and back it out. (Again, probably not quite accurate.) This means that filesystem checks needn't be performed at all. Mostly. Of course, in order for changes to occur, the filesystem must be mounted read-write, and the filesystem driver for both filesystems understands this, so it's never checked due to a forcible unmount of a read-only filesystem.

Since, for non-hackers, the empeg's filesystems are only mounted read-write when syncs are being performed, this occurs fairly seldom -- only when the power is removed during a sync, or crashes. But they can happen.

But you're right that it's much more likely to help a hacker-type. But, IMHO, what would be more likely to help is just to remember to unmount the filesystems cleanly.

Tony, I think you are seriously understating the deal with ext3.

The reason there are no file system checks isn't as simple as the case for hijack, where they're just turned off.

There actually are file system checks with ext3, they just happen a lot faster because
it's a journaling file system and it's much harder to get it into a state where there is real
corruption. It's a much more robust solution.

I realize you probably know this stuff, but I don't know if everyone does.

Anyway, if indeed everything you need is built into the latest hijack (I haven't checked
myself), you just need to:

1) run tune2fs -j on the partitions
2) make sure they are mounted as ext3 -- it's been awhile since I poked around on the player, so I can't remember if there is an fstab to tweak in this regard.

As someone else stated, it may be tricky to get the root partition mounted ext3, I am not really sure.

I should think Mark would chime in, being the file system stud he is :-)

Cheers,
rjf&

In reply to:

---

How do I "turn it on"?

---

As has been pointed out by a few other people, if you aren't comfortable with this and don't know what ext3 is you probably don't want to try this just yet. If you are feeling brave (or foolish) what you need to do is go back to my site that you linked and download the appropriate kernel binary and follow the direction on my site to install it and configure ext3.

The kernel binary that you can download at Mark Lord's site does not support ext3. What he meant in his original post is that if you download the hijack SOURCE from his site you don't have to patch the ext3 stuff from my site in. But, the binary he distributes does not have ext3 enabled.

-Mike

In reply to:

---

I should think Mark would chime in, being the file system stud he is :-)

---

I don't want to understate Mark's studliness, but I believe he is more of a driver stud that a filesystem stud. At least I didn't see his name that much while poking around all the fs code to get ext3 working.

As far as who should consider ext3, well it is probably most useful to developers, but I think it will get more and more important to regular users as hard drives get bigger. We've already got people on this board with 120G empegs. How long does that take to fsck? 3, maybe 4 hours?

-Mike

as a side note... if you decide to apply the hijack-ext3 kernel... fscking the root partition (with the updated e2fsck ) or stock fsck.ext2 gives the error:
The superblock could not be read or does not describe a correct ext2
filesystem. If the device is valid and it really contains an ext2
filesystem (and not swap or ufs or something else), then the superblock
is corrupt, and you might try running e2fsck with an alternate superblock:
e2fsck -b 8193 <device>

i had to use the stock fsck to do it.

We've already got people on this board with 120G empegs.

Right.
And the next jump in mobile hard drive capacity is probably going to be 80GB or 100GB GB.
Put those in your empeg and fsk 'em.
It would take d_mn near all day.

In reply to:

---

fscking the root partition (with the updated e2fsck ) or stock fsck.ext2 gives the error:

---

Hmmm, I have not had that problem and I am sure I have fsck'd my root partition quite a few times since I started running ext3. Can anybody else confirm this? If you fsck it again with the newer fsck do you still get that error?

-Mike

its the newer fsck that gives problems. i have to use the backup fsck.old to do it.

In reply to:

---

its the newer fsck that gives problems. i have to use the backup fsck.old to do it.

---

Works for me...

empeg:/bin# fsck -f /dev/hda5
e2fsck 1.29 (24-Sep-2002)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/hda5: 1526/4096 files (0.5% non-contiguous), 11331/16384 blocks

Are you doing 'fsck /' maybe? Does that work with the stock fsck?

-Mike

So are they searching you at random and you just happened to be picked every time, or is it because you're canadian, or did they somehow actually connect your name to "Hijack"?

I believe the admin can specify the desired size for the journal file to just about any size wanted.

In reply to:

---

Works for me...

---

yeah, tried it and it works for me too.

mr. tfabris, may i suggest changing the faq entry from fsck -fay / (which works using the old fsck) to what it should be.

may i suggest changing the faq entry

I don't understand. I just used that FAQ text recently, and it worked perfectly.

yeah, it works with the stock fsck utility on a stock filesystem... but if you decide to use an updated version of fsck (like the one on mcomb's ext3 page), it wont work anymore. gives the error above.

This is from the e2fsck manpage on my machine,

-a This option does the same thing as the -p option. It is provided for backwards compatibility only; it is suggested that people use -p option whenever possible.

Perhaps his version is newer and has dropped the -a support.

Edit: clearly I haven't been reading the thread at all.

"fsck -fay /" fails, because /etc/fstab still contains 'ext2' as the filesystem. However when you use "fsck -fay /dev/hda5" it won't look at the fstab, but use whatever filesystem it can derive from the superblock, i.e. ext3.

but if you decide to use an updated version fo fsck

Ah, I see. You do a nonstandard hack to the player, and put a nonstandard filesystem in place, and replace one of the base system files of the player with a different version, then you want me to document that for you.

Sorry, I have to draw the line somewhere, or my FAQ will soon turn into a full Linux FAQ.

And you wonder why I recommend the average joe avoid the EXT3 conversion.

heh, i don't mind if its in the faq or not.... i just found it as a source of confusion when i referred to it.

but here's for another argument for it.
why not just forget the umount /drive0 and /drive1, and just have it fsck -fay /drive0 and /drive1. that will work also.
its a matter of consistency =). if you're going to fsck mount points, do it for all mount points. if your going to fsck device paths, then do it for all device paths.

how about that?

The thread that generated the text for that FAQ entry was long and had many different opinions and arguments in it. The text that currently resides in the FAQ was reached by a consensus of several experts here on the BBS as being absolutely correct.

I will not change it again unless that same group of experts comes in here an fights it out with you and you can still convince them that they are wrong.

How about just noting that if you're living on the cutting edge and ust ext3 then to read this thread!

It's not like there isn't anything non-standard mentioned in the FAQs...there's one basically devoted to hijack for instance.

In reply to:

---

"fsck -fay /" fails, because /etc/fstab still contains 'ext2' as the filesystem. However when you use "fsck -fay /dev/hda5" it won't look at the fstab, but use whatever filesystem it can derive from the superblock, i.e. ext3.

---

Sounds good in theory, but the fs he is trying to fsck is the root partition which (assuming he followed the directions on my page) is still ext2. If you look at the usage info for the version of fsck I provide it does not support the 'fsck /mountpoint' syntax.

empeg:/empeg/bin# fsck
Usage: fsck [-panyrcdfvstFSV] [-b superblock] [-B blocksize]
[-I inode_buffer_blocks] [-P process_inode_size]
[-l|-L bad_blocks_file] [-C fd] [-j ext-journal]
[-E extended-options] device

I am not sure why this is different that the stock empeg fsck, but I will add a note to my info page noting the difference.

-Mike

If anyone here wants access to the Developer Info section on RioCar for this type of info, let me know.