kernel.git/fs/bcachefs/io_write.c, branch linux-6.15.yHosts the 0x221E linux distro kernel.
https://git.0xinfinity.dev/distro/kernel.git/atom?h=linux-6.15.y2025-05-01T10:19:58Zbcachefs: check for inode.bi_sectors underflow2025-05-01T10:19:58ZKent Overstreet2025-05-01T03:56:00Zurn:sha1:3a72e369412d6bf7f6a8af410ac4bdd7d48deb62
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Kill btree_iter.trans2025-04-02T14:24:34ZKent Overstreet2025-03-21T19:18:51Zurn:sha1:9180ad2e161b7030e0af78fd2266cbcefe81e652
This was planned to be done ages ago, now finally completed; there are
places where we have quite a few btree_trans objects on the stack, so
this reduces stack usage somewhat.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Split up bch_dev.io_ref2025-04-02T14:24:34ZKent Overstreet2025-03-30T03:11:08Zurn:sha1:dcffc3b1ae3251d796a25c673f614e3099ca83d3
We now have separate per device io_refs for read and write access.
This fixes a device removal bug where the discard workers were still
running while we're removing alloc info for that device.
It's also a bit of hardening; we no longer allow writes to devices that
are read-only.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Fix null ptr deref in bch2_write_endio()2025-03-31T21:39:27ZKent Overstreet2025-03-31T20:19:04Zurn:sha1:de399658588931506e1b3f4cc63cb3c9134a692e
This was previously hard to hit since it requires racing with device
removal, but splitting up io_ref uncovered it.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Don't unnecessarily decrypt data when moving2025-03-26T20:26:35ZKent Overstreet2025-03-24T20:25:53Zurn:sha1:19ff84b20d62e62ff527a384b35102bf745a70d1
There's various checks for "are we going to compress this" - but we're
not going to compress if we know it's incompressible.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Fix duplicate checksum error messages in write path2025-03-25T15:49:43ZKent Overstreet2025-03-24T20:40:22Zurn:sha1:a76db26a96982857a306d3e9dbc7f44d5ae45d9d
Also, improve the message in prep_encoded_data() - it now prints
good/bad checksums, and checksum type.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Fix nonce inconsistency in bch2_write_prep_encoded_data()2025-03-25T15:40:35ZKent Overstreet2025-03-25T15:40:35Zurn:sha1:5af61dbd96275e184adcfe615507b0f04ed7b328
If we're moving an extent that was partially overwritten,
bch2_write_rechecksum() will trim it to the currenty live range.
If we then also want to compress it, it'll be decrypted - but the nonce
has been advanced for the overwritten start of the extent that we
dropped, and we were using the nonce we calculated before rechecksum().
Reported-by: Gabriel de Perthuis <g2p.code@gmail.com>
Fixes: 127d90d2823e ("bcachefs: bch2_write_prep_encoded_data() now returns errcode")
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: EIO cleanup2025-03-24T13:50:36ZKent Overstreet2025-03-20T14:16:48Zurn:sha1:8a9f3d058279ed0f99114e0449d129fb5abc5eca
Replace these with proper private error codes, so that when we get an
error message we're not sifting through the entire codebase to see where
it came from.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: bch2_write_prep_encoded_data() now returns errcode2025-03-24T13:50:36ZKent Overstreet2025-03-20T14:30:51Zurn:sha1:127d90d2823ef45dd37246aaf7bb0392e6231c38
Prep work for killing off EIO and replacing them with proper private
error codes.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bcachefs: Simplify bch2_write_op_error()2025-03-24T13:50:36ZKent Overstreet2025-03-20T14:25:15Zurn:sha1:2fe208303a114012d7ca035dcb37edcee93b64f6
There's no reason for the caller to do the actual logging, it's all done
the same.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>