As subject says, there's no more support for supermount in 2.6.8 kernel of the gentoo-dev branch, and I think this is definitely bad cause now supermount has showed to be stable, quick and full functionally. Please re-add it to the gentoo-dev-sources branch (if there are no big problem, in which case please explain what's happening)
Steps to Reproduce:
> Removed supermount patch, am told that udev makes this obsolete now
This is not true at all. Please bring it back.
Well, the patch doesn't apply to 2.6.8, so that was the first reason to
The second is that udev is supposed to be able to support everything that
supermount did. If this is not true, please give me details about what
is not supported.
Thirdly, this patch has been rejected by upstream many different times, and is
going no where anytime soon. I don't want to support it, and none of the other
main kernel.org developers do either. It's time to start weaning yourself off of it :)
at first - sorry for reopening this bug :-(
i too use supermount and do not know how to tell udev to let me remove mounted
devices without unmounting them before. so far about me and trying to use super-
mount to do what supermount does. would be glad if someone could tell me how to
replace supermount with udev & co - found nothing on google expect ivman using
hal/d-bus - for now it is not really usable for me but looks promising...
i have a patch against linux-22.214.171.124. it is tested on x86 - running smoothly.
there where only some minor changes i had to apply to the existing patch for
2.6. hope it works well for you all. would be glad to have supermount back on
gentoo unless someone can figure out how to emulate the functionality in udev :-)
so far from me - the patch is attached to this message :-)
Created attachment 37570 [details, diff]
SuperMount patch for linux-126.96.36.199
pls try and report :-) hopefully i made no mistakes - it is my first
There is submount, which is in portage. But it did not work as well as supermount for me.
On using udev for what supermount did. I have got udev to make device nodes appear where i want them, and i could probably make udev mount devices when they appaear and umount them when they disapear. What makes supermount nessessary is that it will mount the devices on access - so that u can use it for floppy drives and cdrom drives, and it ensures that buffers are flushed so that you do not loose data when the device is removed.
supermount does some voodoo magic in the kernel (hence its fragile nature, and why many kernel devs hate it) so that your cdrom drive _never_ locks, and does it properly (ie. it doesn't constantly umount the filesystem, like say automount). That alone is something I've yet to see in other cdrom drive management solutions. And if there is one thing I cannot stand, it's pressing the eject button on my cdrom drive and seeing it NOT open. That's just stupid. Windows does it properly for over a decade now, and supermount is still the only thing I know of under Linux that provides comparable support, and works.
With the correct settings, you can tell supermount to lock drives when it is writing/reading to them. So that the drive is only locked while supermount is using the drive. Ive used supermount for a while at home and at work. It does exactly what is requierd of it. The only problem i have had is things like KDE or FAM continuing to monitor files/directories that are not longer being used.
so pls guys test my patch and gimme some feedback. :-)
perhaps someone is willing to help me maintaining this
patch for the next few kernelversions? that is perhaps
the only way of having supermount in here...
In the a
A lot of people here (in India) still use floppies extensively. I've been using supermount ever since 2.4.16-mdk IIRC and have set up three media labs in slum clusters in Delhi which run on Gentoo and depend on supermount being provided in the kernel.
An easy and transparent way for normal users to reliably store and retreive files on floppies has been critical for our operations and remains so.
I'd be glad to help the gentoo kernel maintainers test the supermount patches. Alternatively, udev documentation on transparent floppy/cdrom/usb device access will be extremely welcome.
What about using autofs and HAL? That should work, right?
Please have a look at bug #4285 - have posted a comment there about
dbus/hal/ivman and subfs. In short:
None of them worked for me. I really, really tried to replace super-
mount with the other solutions. I NEED supermount. Please put it back
into at least gentoo-dev-sources. a good start may be 2.6.8-gentoo-r4
Created attachment 40142 [details]
supermount-ng205-1 for gentoo-dev-sources-2.6.8-r4
This supermount patch should apply to gentoo-dev-sources-2.6.8-r4 with some
line offsets and work fine (at least with my configuration). Perhaps this is
a good starting point for the gentoo kernel guys to put supermount back into
the 2.6.x series of the kernel. There are many people who NEED supermount.
Sorry, but no, I'm not going to add it.
HAL and autofs should work, and if not, please let those developers know what
does not work for you.
*** Bug 74134 has been marked as a duplicate of this bug. ***
Pretty please, with sugar on top, add super mount back to gentoo-dev-sources.
Non of the other solutions provides a way to eject a media with a live mounted filesystem, without trashing the mount table with mount -f ( u know this evil fam that keeps locks open everywhere ).
* There is no functional replacement for it now. *
Please consider adding it back. It can be left diabled in the default .config,
so no harm for "plain" systems will be done.
As for now, if someone needs to have a civilized distro, here is a patch for gentoo-dev-sources 2.6.9, applies cleanly and works smoothly.
And one question. Why gentoo developers ignore gentoo users? Isn't it odd?
There are many many votes on bug.gentoo.org sugesting, that users need
supermount. We users just want the cutting edge distro.
Find someone willing to fix it up and submit it into the mainline kernel and we will include it. As it stands right now, it has an ugly design, has masses of problems with NFS, has some nasty race conditions making it easy to crash your kernel, and hasn't been developed for almost 1 year.
We aim to keep our patchset slim and feature patches that we include are rapidly progressing towards mainline inclusion. Supermount would be the opposite of this.
It will not be included in its current state- sorry. If you insist on using it, you can apply it each time.
submount works reasonably well. you still can't eject cds mounted with submount.. . but apart from that, it works pretty well.
use udev and ivman. That supports ejecting just like supermount does, and it's maintained.
Well but until 2 weeks ago or so it didn't support usb stick, for example, making it quite useless for a modern desktop system. And anyway I think we'll have still to wait some months before getting a 99% proof ivman system.
Btw, I know supermount has bugs and problems (i burnt tons of CDs before realizing it was due to supermounting empty medias I got problem), and I think NOW (as 2.6.10) it's normal to throw it away from gentoo-kernel, but this bug was opened months ago, and then the situation was really different.
a) use ivman
b) if you dont want that for whatever reason, use submount
submount does the same as supermount w/o a kernel patch, but external module, it i a full replacement.
These options where there even months ago, so please dont argue.
They all do what supermount does (yes, even unlocking and ejecting w/o umounting)
ivman did not do usbsticks untill 2 weeks ago, but submount ever did ..
(though only ivman has this cool automatic configuration where you dont need to care about /dev/sd* but just get the stick mountet ;)
Submount, agai, was until few weeks ago hard masked, I think this is not a good sign if you're trying to switch from a broken method to another...
I know ivman is the right choice now, using hal and dbus it gets tons of advantages over the other methods, but in AGOUST it wasn't really that usable, and that's the reason why I opened this bug
Submount was never broken .. it was masked because of broken config-kernel, which had nothing to do with submount, and the mask is now gone, as you can see. (for ~archs as no dev marked it stable yet)
uhm... I was using usb-storage with ivman when I tested it for addition to portage in October, so that functionality has been there from the beginning... I'm not quite sure what you mean by that... but rest assured, supermount is NOT going back into the kernel as the code is crap, prone to spinlock, unmaintained, deprecated, etc, etc.