Volunteering for BeFS maintainership
Hi all,
Now that the 4.7 cycle is finished, I want to volunteer myself and Salah Triki for maintainership of the BeFS filesystem.
Salah and myself have been cleaning up the code for the last few months, and since the current status of it is Orphan, it would be great if we could take care of it.
Personally, besides clean ups, I would like to add more documentation of how the file system works, eg. including more function descriptions, so the Linux implementation can be the ideal practical match to Dominic Giampaolo's free book "Practical File System Design with the Be File System".
This book is the most in-depth I've found about any file system, which is why I think having linux/fs/befs well documented would be very valuable to eveybody learning about how file systems work.
I attach below what the change in MAINTAINERS would look like. If this is approved I will send a proper formatted patch mail.
Thanks,
Luis
PD: I don't know if there is a formal process for maintainership in the Linux kernel community. I am happy to follow that if it exists.
Also, if I'm missing anybody who should be in CC, please let me know.
--- MAINTAINERS | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
MAINTAINERS +3 -1
@@ -2321,7 +2321,9 @@ S: Supported F: drivers/media/platform/sti/bdisp BEFS FILE SYSTEM -S: Orphan +M: Luis de Bethencourt <[email protected]> +M: Salah Triki <[email protected]> +S: Mantained F: Documentation/filesystems/befs.txt F: fs/befs/
--
2.5.1
Quoted 2 lines · Luis de Bethencourt
PD: I don't know if there is a formal process for maintainership in the Linux kernel community. I am happy to follow that if it exists.
If no one objects, and you want to do this, send a real patch and away you go!
Good luck!
greg k-h
Quoted 11 lines · Greg KH
Quoted 2 lines · Luis de Bethencourt
PD: I don't know if there is a formal process for maintainership in the Linux kernel community. I am happy to follow that if it exists.
If no one objects, and you want to do this, send a real patch and away you go!
Good luck!
greg k-h
Thanks Greg! :)
I will wait a few days in case other people want to comment before.
Luis
Quoted 3 lines · Luis de Bethencourt
I will wait a few days in case other people want to comment before.
Sounds great! Do you have a git tree set up for your befs development?
And if you haven't made plans to use xfstests, I would certainly commend that for your consideration. Don't forget to subscribe to [email protected] to get xfstests announcements. Also, if you're interested in a framework to easily run xfstests in a GCE (Google Compute Engine) VM, feel free to contact me.
Cheers,
- Ted
Quoted 7 lines · Theodore Ts'o
Quoted 3 lines · Luis de Bethencourt
I will wait a few days in case other people want to comment before.
Sounds great! Do you have a git tree set up for your befs development?
Thank you :)
Yes, I have the following in github (if that is OK):
https://github.com/luisbg/linux-befs
I have two branches there based on Linus' master:
- befs-linus: with patches Andrew Morton has approved
- befs-next: with patches I've tested but that remain under review
Quoted 10 lines
And if you haven't made plans to use xfstests, I would certainly commend that for your consideration. Don't forget to subscribe to [email protected] to get xfstests announcements. Also, if you're interested in a framework to easily run xfstests in a GCE (Google Compute Engine) VM, feel free to contact me.
Cheers,
- Ted
I've subscribed to [email protected].
It would be amazing to have a framework to run xfstests in a GCE VM.
Big thanks,
Luis
Quoted 9 lines · Luis de Bethencourt
Quoted 2 lines
Sounds great! Do you have a git tree set up for your befs development?
Yes, I have the following in github (if that is OK):
https://github.com/luisbg/linux-befsI have two branches there based on Linus' master:
- befs-linus: with patches Andrew Morton has approved
- befs-next: with patches I've tested but that remain under review
So it sounds like you plan to send patches through Andrew's tree. That works fine, although if you end up sending a larger number of patches through the linux-mm tree, it might make sense for you to send patches to Linus directly. So if you have a chance to get a GPG key which is signed by people in the Kernel keyring, that would be a good preparation for that eventuality. That will require face-to-face verification of your identity by people who are already in the GPG web of trust, so it's good to plan for that in advance.
Quoted 1 line
It would be amazing to have a framework to run xfstests in a GCE VM.
Please see:
https://thunk.org/gce-xfstests
and
https://github.com/tytso/xfstests-bld/blob/master/README.md
for more information.
I plan to do some work to make it simpler to get started using gce-xfstests. (Specifically, so you don't have to build the tree and generate your own GCE image, but instead using a premade one.)
Are there userspace tools available to create and consistency check BeFS file systems? If so, I can try to get those included into the test appliance image. (Better yet, if you can arrange to have someone create a debian package for BeFStools, that would be great.)
- Ted
Quoted 4 lines · Theodore Ts'o
Quoted 3 lines · Luis de Bethencourt
Quoted 2 lines
Sounds great! Do you have a git tree set up for your befs development?
Quoted 8 lines
Quoted 6 lines
Yes, I have the following in github (if that is OK):
https://github.com/luisbg/linux-befsI have two branches there based on Linus' master:
- befs-linus: with patches Andrew Morton has approved
- befs-next: with patches I've tested but that remain under reviewSo it sounds like you plan to send patches through Andrew's tree.
Not sure. Whatever Andrew, Al, Greg and yourself think is better.
Quoted 7 lines
That works fine, although if you end up sending a larger number of patches through the linux-mm tree, it might make sense for you to send patches to Linus directly. So if you have a chance to get a GPG key which is signed by people in the Kernel keyring, that would be a good preparation for that eventuality. That will require face-to-face verification of your identity by people who are already in the GPG web of trust, so it's good to plan for that in advance.
I will be at ELCE in 3 months [0]. It will be a good opportunity to get my GPG key signed by people in the Kernel keyring. For example Maxime Ripard, who is one of the speakers and in the GPG web of trust [1].
Quoted 5 lines
Quoted 2 lines
It would be amazing to have a framework to run xfstests in a GCE VM.
Please see:
Quoted 7 lines
https://thunk.org/gce-xfstests
and
https://github.com/tytso/xfstests-bld/blob/master/README.md
for more information.
Very nice. Looking forward to playing with it soon, and adding it to my development process. I have learned a lot reading the ext4 code so I am also interested in running the test suite with that.
Quoted 4 lines
I plan to do some work to make it simpler to get started using gce-xfstests. (Specifically, so you don't have to build the tree and generate your own GCE image, but instead using a premade one.)
I will watch out for updates on this on fstests@vger.
Quoted 7 lines
Are there userspace tools available to create and consistency check BeFS file systems? If so, I can try to get those included into the test appliance image. (Better yet, if you can arrange to have someone create a debian package for BeFStools, that would be great.)
- Ted
Support for BeFS in Linux is read-only. So there are no tools to create BeFS file systems. I have a bunch of BeFS images created from Haiku OS that cover most things.
There is no reason a consistency check tool can't be written. I think this could be a fun exercise, I am tempted to add it to my task list :) However this tool can only inform if the file system is consistent, and not really fix it. Similar to "e2fsck -n".
Salah told me he is planning to slowly work on adding read support in the future. But don't want to make any promises/plans on his behalf.
Thanks,
Luis
[0] http://events.linuxfoundation.org/events/embedded-linux-conference-europe/program/schedule
[1] https://www.kernel.org/doc/ksmap/
Quoted 3 lines · Luis de Bethencourt
Support for BeFS in Linux is read-only. So there are no tools to create BeFS file systems. I have a bunch of BeFS images created from Haiku OS that cover most things.
Ah, well, pretty much all of the xfstests assume the ability to write into the file system. So it may be a while before using xfstests will make sense for BeFS.
Quoted 4 lines
There is no reason a consistency check tool can't be written. I think this could be a fun exercise, I am tempted to add it to my task list :) However this tool can only inform if the file system is consistent, and not really fix it. Similar to "e2fsck -n".
That's all which is necessary for xfstests --- the idea is that the test will make various changes to the file system, and then correctness is checked both by whether the expected output is printed as the test probes changed the file system state, and by the consistency checker confirming that the file system is in a consistent sane state after each test completes.
Quoted 2 lines
Salah told me he is planning to slowly work on adding read support in the future. But don't want to make any promises/plans on his behalf.
I assume you mean write support in the above paragraph. :-)
Cheers,
- Ted
Quoted 14 lines · Theodore Ts'o
Quoted 3 lines · Luis de Bethencourt
Support for BeFS in Linux is read-only. So there are no tools to create BeFS file systems. I have a bunch of BeFS images created from Haiku OS that cover most things.
Ah, well, pretty much all of the xfstests assume the ability to write into the file system. So it may be a while before using xfstests will make sense for BeFS.
Quoted 4 lines
There is no reason a consistency check tool can't be written. I think this could be a fun exercise, I am tempted to add it to my task list :) However this tool can only inform if the file system is consistent, and not really fix it. Similar to "e2fsck -n".
Quoted 7 lines
That's all which is necessary for xfstests --- the idea is that the test will make various changes to the file system, and then correctness is checked both by whether the expected output is printed as the test probes changed the file system state, and by the consistency checker confirming that the file system is in a consistent sane state after each test completes.
Right. It will still be valuable for me to run xfstests for ext4 and btrfs.
Thanks for the advice.
Quoted 5 lines
Quoted 2 lines
Salah told me he is planning to slowly work on adding read support in the future. But don't want to make any promises/plans on his behalf.
I assume you mean write support in the above paragraph. :-)
You assume correctly :) Sorry for the typo.
Quoted 4 lines
Cheers,
- Ted
Thanks,
Luis
Quoted 13 lines · Theodore Ts'o
Quoted 3 lines · Luis de Bethencourt
I will wait a few days in case other people want to comment before.
Sounds great! Do you have a git tree set up for your befs development?
And if you haven't made plans to use xfstests, I would certainly commend that for your consideration. Don't forget to subscribe to [email protected] to get xfstests announcements. Also, if you're interested in a framework to easily run xfstests in a GCE (Google Compute Engine) VM, feel free to contact me.
Quoted 3 lines
Cheers,
- Ted
Thanx :)
I'm also interested in the GCE VM.
best regards,
salah
Quoted 10 lines · Greg KH
Quoted 2 lines · Luis de Bethencourt
PD: I don't know if there is a formal process for maintainership in the Linux kernel community. I am happy to follow that if it exists.
If no one objects, and you want to do this, send a real patch and away you go!
Good luck!
greg k-h
Thanx greg :)