If the root inode/subvolume is unreadable we can repair automatically -
but only if we're still in recovery, so that we can rewind to the
appropriate recovery pass.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Instead of going emegency read only with a bch2_fs_inconsistent() call,
log the error and recovery pass appropriately.
If we're still in recovery it'll be repaired immediately, otherwise
it'll be repaired on the next mount.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
bch2_print_string_as_lines() is a low level helper that allows messages
longer than 1k to be printed without truncation.
But we should always be printing with the helpers that take a filesystem
object, if we're in fsck they direct output to the userspace process
controlling fsck instead of the dmesg log.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Part of the ongoing project to kill off bch2_(fs|trans)_inconsistent
calls - they generally need to be replaced with either
- a fsck_err() call that can repair the error, or
- logging an error of the appropriate type in the superblock, and
flagging the appropriate recovery pass to repair the error
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
We prefer helpers that emit log messages to printbufs rather than
printing them directly; that way, we can ensure that different log
messages from the same event are grouped together and formatted
appropriately in the dmesg log.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
version_upgrade is now a runtime option.
In the future we'll want to add compatible upgrades at runtime, and call
the full check_version_upgrade() when the option changes, but we don't
have compatible optional upgrades just yet.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The helpers are now:
- bch2_opt_hook_pre_set()
- bch2_opts_hooks_pre_set()
- bch2_opt_hook_post_set
Fix a bug where the filesystem discard option would incorrectly be
changed when setting the device option, and don't trigger rebalance
scans unnecessarily (when options aren't changing).
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Single device filesystems are now identified by the block device name,
not the UUID - and single device filesystems with the same UUID can be
mounted simultaneously, without any special options.
This allocates a new bit in the superblock, BCH_SB_MULTI_DEVICE, which
indicates whether a filesystem has ever been multi device.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
On single device filesystems, c->name contains the block device name,
not the UUID.
Initialize this earlier, so that single device mode can use it for
initializing sysfs/debugfs.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Make the output slightly clearer, and include a counter for "nodes we
couldn't free because we would have gone under our reserve".
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kill 'opts.very_degraded', and make 'opts.degraded' a persistent option,
stored in the superblock.
It's now an enum, with available choices ask/yes/very/no.
"ask" mode will be handled by the mount helper, for prompting the user
(on a machine used interactively) for whether to do a degraded mount.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This patch uses printbuf_indent_add_nextline() to set a consistent
indentation level for error messages of invalid compression.
In my previous patch [1], the newline is added by using '\n' in
the argument of prt_str(). This patch replaces prt_str() with
prt_printf() to make indentation level work correctly.
[1] Link: https://lore.kernel.org/20250406152659.205997-2-integral@archlinuxcn.org
Signed-off-by: Integral <integral@archlinuxcn.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
When an invalid compression type or level is passed as an argument
to `--compression`, two error messages are squashed into one line:
> bcachefs format --compression=lzo bcachefs-comp.img
invalid option: invalid compression typecompression: parse error
> bcachefs format --compression=lz4:16 bcachefs-comp.img
invalid option: invalid compression levelcompression: parse error
To resolve this issue, add a newline character at the end of the
first error message to separate them into two lines.
Signed-off-by: Integral <integral@archlinuxcn.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Currently, when passing a negative integer as argument, the error
message is "too big" due to casting to an unsigned integer:
> bcachefs format --block_size=-1 bcachefs.img
invalid option: block_size: too big (max 65536)
When negative value in argument detected, return early before
calling bch2_opt_validate().
A new error code `BCH_ERR_option_negative` is added.
Signed-off-by: Integral <integral@archlinuxcn.org>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Defer memory allocations only needed in RW mode until we actually go RW.
This is part of improved support for RO images.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
_init_early() is for initialization that cannot fail, and often must
happen for teardown partway through initialization to work.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Add a better helper for check_snapshot_exists().
create_snapids() can't be changed to use this, unfortunately, because
the transaction that creates new snapshot will also be inserting other
keys (e.g. root inode) that reference that snapshot ID, and they expect
the snapshot table to already be updated.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
More stack usage improvements: instead of creating a new alloc_request
(currently on the stack), save/restore just the fields we need to reuse.
This is a bit tricky, because we're doing a normal alloc_foreground.c
allocation, which calls into ec.c to get a stripe, which then does more
normal allocations - some of the fields get reused, and used
differently.
So we have to save and restore them - but the stack usage improvements
will be well worth it.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Add a struct for common state for satisfying an on disk allocation,
instead of passing the same long list of items to every function.
This will help with stack usage, performance, and perhaps enable some
code cleanups.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
We're occasionally seeing the WARN_ON() for bump allocator usage
exceeding BTREE_TRANS_MEM_MAX; add some tracing so we can see what's
going on.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This was achieved before by zero-ing out the source buffer and then
copying the bytes into the destination buffer. This can also be done with
memcpy_and_pad which will zero out only the destination buffer if its
size is bigger than the size of the source buffer. This is already used in
the same way in journal_transaction_name().
Moreover, zero-ing the source buffer was done twice, first in
__bch2_fs_log_msg() and then in bch2_trans_log_msg(). And this method
may also require allocating some extra memory for the source buffer.
In conclusion, using memcpy_and_pad is better even tough the result is
the same because it brings uniformity with what's already used in
journal_transaction_name, it avoids code duplication and reallocating
extra memory.
Signed-off-by: Roxana Nicolescu <nicolescu.roxana@protonmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
