---------- Forwarded message ---------- From: Hy Che <cvghy...@gmail.com> Date: Sun, Jun 11, 2017 at 10:45 PM Subject: Re: Questions about how BtrFS works. To: Nikolay Borisov <n.borisov.l...@gmail.com>
Sorry for sending privately, I did't know clicked on 'Replay' is not sent to the mailing list. Some additional info. 4. It's not that I'm lazy reading code, I just don't know where to start. 5. I understand why it is now, the 100 bytes is header size, and the offset only counted at the end of the header. On Sun, Jun 11, 2017 at 3:00 PM, Nikolay Borisov <n.borisov.l...@gmail.com> wrote: > > > > On 10.06.2017 17:37, Hy Che wrote: > > Hello everyone, > > My name is Hy (ugen on freenode), I am a student attending GSoC for > > Haiku[1]. Because I have a project that support write features for > > BtrFS[2], there are some questions, since the documents on wiki are > > not enough so I have to ask here. > > > > 1. Where can I find to read a complete btree manipulations in detail > > (split, insert, etc) for BtrFS except the code base ? I need to get > > the idea first. > > Have you read this paper: https://dl.acm.org/authorize?N80838 > The original idea came by Ohad Rodeh while he was at IBM. > > I believe this was the initial paper: https://dl.acm.org/authorize?N80839 Have read both of them before, but I need more details. There is a useful stuff here on https://www.spinics.net/lists/linux-btrfs/msg05421.html Is it work the same in current version ? > > > 2. About the reference, " ...with a refcount associated to each tree > > node but stored in an ad-hoc free map structure ... " - wikipedia. > > What is a free map structure, and where does it stored ? > > I believe you are looking for this : > https://btrfs.wiki.kernel.org/index.php/Btrfs_design#Reference_Counted_Extents > > > http://www.spinics.net/lists/linux-btrfs/msg33415.html Thanks, exactly what I want. > > > > > 3. How BtrFS allocate block and extent when cow-ed ? As I tested after > > making a new file, the root tree root, extent root and csum root move > > forward 65536 bytes, the fs root move 49152 bytes and the data extent > > is allocated next to the previous one. Where those numbers come from ? > > Is data extent is cow-ed ? I see it doesn't because extent-item still > > contains the old offset. > > You'd have to read the code, have you taken a look at : __btrfs_cow_block() I have take a look, but it is kinda long and complex. If i traced correcly, It would take to find_free_extent(http://elixir.free-electrons.com/linux/latest/source/fs/btrfs/extent-tree.c#L7408), where the infomation need is key "ins", and I don't know how it is changed in which case. > > > > > > 4. How BtrFS handle transactions ? > > Correctly me if I'm wrong, the transaction collect all requests in 30 > > seconds and then write back to disk. The transid increments when new > > request appeared and genid is asigned to this one. > > I don't think there is anything written per-se. You'd again have to resort to > reading the > code I need a rough idea before reading code because it would be taking lots of time. > > > > > 5. Why there is unused space (100 bytes) at the end of each node ? > > > 6. How does BtrFS calculate checksum ? > > It uses a 32bit CRC. The actual function which is used to calc > the csum is csum_tree_block you can check its callers and internals to > in which code paths the crc is used. But in general all it does is call > btrfs_csum_data > on the extent buffer which holds the particular block. > Thanks Hy -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
