lkml.org 
[lkml]   [2014]   [Jul]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: BTRFS hang with 3.16-rc5 (and also with 3.16-rc4)
From
Hey Hugo,
Thanks for the advice. I will look into this today :).
Cheers Nick

On Fri, Jul 25, 2014 at 4:02 AM, Hugo Mills <hugo@carfax.org.uk> wrote:
> On Thu, Jul 24, 2014 at 11:06:34PM -0400, Nick Krause wrote:
>> On Thu, Jul 24, 2014 at 10:32 PM, Duncan <1i5t5.duncan@cox.net> wrote:
> [snip]
>> Hey Duncan and others ,
>> I have read this and this seems to need some working on.
>> If you want my help please ask , I am new to the kernel
>> so I may ask a dumb question or two but if that's fine with
>> you I have no problem helping out here. I would like
>> a log of printk statements leading to the hand if that's
>> not too much work in order for me to trace this back.
>
> Note that btrfs is complex -- there's something around 100k lines
> of code in it. My first piece of kernel work in btrfs was simply
> documenting the way that the on-disk data structures related to each
> other[1]. That on its own took me two to three weeks of solid
> full-time effort, reading the code to find where each structure was
> used and how its elements related to other structures. You can't just
> wander up and dive in without putting in the effort of learning first.
> Whilst people will help you (come over to #btrfs on Freenode for more
> real-time interaction), they can't do the basic work of sitting down
> and understanding the code in detail for you.
>
> Chris, who designed and wrote the filesystem, has spent the last
> couple of weeks tracking down this particular problem. Do you think
> it's appropriate to leap into the middle of the discussion on this
> subtle bug as someone with absolutely no experience in the area?
>
> Your first task is to reproduce the bug on your own machine. If you
> can do that, _then_ you might be able to start tracking down its
> cause. But I wouldn't recommend doing that, as (a) it's a nasty subtle
> bug, and (b) Chris seems to be close to tracking it down anyway.
>
> My recommendations for you, if you want to work on btrfs, are:
>
> * Build and install the latest kernel from Linus's git repo
>
> * Read and understand the user documentation [2]
>
> * Create one or several btrfs filesystems with different
> configurations and learn how they work in userspace -- what are the
> features, what are the problems you see? Actually use at least one
> of the filesystems you created for real data in daily use (with
> backups)
>
> * Build the userspace tools from git
>
> * Pick up one of the userspace projects from [3] and implement that.
> If you pick the right one(s), you'll have to learn about some of
> the internal structures of the FS anyway. Compile and test your
> patch. If you're adding a new feature, write an automated xfstest
> for it as well.
>
> * Get that patch accepted. This will probably involve a sequence of
> revisions to it, multiple versions over a period of several weeks
> or more, with a review process. You should also send your test to
> xfstests and get that accepted.
>
> * Do the above again, until you get used to the processes involved,
> and have demonstrated that you can work well with the other people
> in the subsystem, and are generally producing useful and sane code.
> It's all about trust -- can you be trusted to mostly do the right
> thing? (So far on linux-kernel, you've rather demonstrated the
> opposite: your intentions are good, but your execution leaves a lot
> to be desired)
>
> * Use the documentation at [4], and the output of btrfs-debug-tree to
> understand the internal structure of the FS
>
> * Pick up one of the smaller, more self-contained ideas from the
> projects page [5] (say, [6] or [7]) and try to implement it. Again:
> build, write test code, test thoroughly, submit patch for review,
> modify as suggested by reviewers, and repeat as often as necessary
>
> Hugo.
>
> [1] https://btrfs.wiki.kernel.org/index.php/Data_Structures
> [2] https://btrfs.wiki.kernel.org/index.php/Main_Page#Guides_and_usage_information
> [3] https://btrfs.wiki.kernel.org/index.php/Project_ideas#Userspace_tools_projects
> [4] https://btrfs.wiki.kernel.org/index.php/Main_Page#Developer_documentation
> [5] https://btrfs.wiki.kernel.org/index.php/Project_ideas
> [6] https://btrfs.wiki.kernel.org/index.php/Project_ideas#Cancellable_operations
> [7] https://btrfs.wiki.kernel.org/index.php/Project_ideas#Implement_new_FALLOC_FL_.2A_modes
>
> --
> === Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
> PGP key: 65E74AC0 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
> --- But people have always eaten people, / what else is there to ---
> eat? / If the Juju had meant us not to eat people / he
> wouldn't have made us of meat.


\
 
 \ /
  Last update: 2014-07-25 18:21    [W:0.036 / U:0.260 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site