Add the ability to enable simple quotas from mkfs with '-O squota'
There is some complication around handling enable_gen while still
counting the root node of an fs. To handle this, employ a hack of doing
a no-op write on the root node to bump its generation up above that of
the qgroup enable generation, which results in counting it properly.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
Allow for RAID levels 0, 1 and 10 on zoned devices if the RAID stripe tree
is used.
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We're getting more features and the string size limit will not be
sufficient, so extend enough that we won't have to care for some time.
Signed-off-by: David Sterba <dsterba@suse.com>
The accelerated crc32c needs to check for two CPU features, the crc32c
instructions is in SSE 4.2 and 'pclmulqdq' is a separate. There's still
old hardware used that does not have the PCLMUL instructions. Detect it
and make it the condition.
The pclmul is not supported on old compilers so also add a
configure-time detection and leave the SSE 4.2 only implementation as
the accelerated one if possible.
Issue: #676
Signed-off-by: David Sterba <dsterba@suse.com>
Aligning with the kernel's struct btrfs_fs_devices:fs_list, rename
btrfs_fs_devices::list to btrfs_fs_devices::fs_list.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Copy faster implementation of crc32c from linux kernel as of 6.5-rc7
(x86_64, arch/x86/crypto/crc32c-pcl-intel-asm_64.S). This needs
assembler build support, so detect target architecture so
cross-compilation still works.
Add a special CPU flag so the old and new implementations can be
benchmarked and verified separately.
Sample benchmark:
CPU flags: 0x1ff
CPU features: SSE2 SSSE3 SSE41 SSE42 SHA AVX AVX2 CRC32C_PCL
Block size: 4096
Iterations: 1000000
Implementation: builtin
Units: CPU cycles
NULL-NOP: cycles: 77177218, cycles/i 77
NULL-MEMCPY: cycles: 226313072, cycles/i 226, 62133.395 MiB/s
CRC32C-ref: cycles: 24418596066, cycles/i 24418, 575.859 MiB/s
CRC32C-NI: cycles: 1188335920, cycles/i 1188, 11833.073 MiB/s
CRC32C-PCL: cycles: 463193456, cycles/i 463, 30358.037 MiB/s
XXHASH: cycles: 851606646, cycles/i 851, 16511.916 MiB/s
SHA256-ref: cycles: 74476234956, cycles/i 74476, 188.808 MiB/s
SHA256-NI: cycles: 34198637428, cycles/i 34198, 411.177 MiB/s
BLAKE2-ref: cycles: 14761411664, cycles/i 14761, 952.597 MiB/s
BLAKE2-SSE2: cycles: 18101896796, cycles/i 18101, 776.807 MiB/s
BLAKE2-SSE41: cycles: 12599091062, cycles/i 12599, 1116.087 MiB/s
BLAKE2-AVX2: cycles: 9668247506, cycles/i 9668, 1454.418 MiB/s
The new implementation is about 2.5x faster.
Note: there new version does not work on musl because of linkage
problems (relocations in .rodata), so it's still using the old
implementation.
Signed-off-by: David Sterba <dsterba@suse.com>
In some places we want to read a single u64 value from a sysfs path, or
from fsid directory. Add helpers that do that in one go.
Signed-off-by: David Sterba <dsterba@suse.com>
The sysfs could use more convenience helpers so move the current code to
own file before adding more helpers.
Signed-off-by: David Sterba <dsterba@suse.com>
API for extensible array of pointers for covenience. A simple wrapper
around a (void *) array with length.
Signed-off-by: David Sterba <dsterba@suse.com>
This is a potentially breaking change to json output. An all zeros uuid
was printed as "-" but we can utilize native json type null for that.
Note the va_copy must be used as va_arg advances the pointer.
{
"nulluuid": null
}
Signed-off-by: David Sterba <dsterba@suse.com>
Make the timestamp format more descriptive what is actually printed. We
may need separate date or time in the future.
Signed-off-by: David Sterba <dsterba@suse.com>
The json spec allows numeric values and it's recommended to use them
instead of the stringified numbers. This is a potentially breaking change
if some tools relied on the string value.
As most formats we now have are '%llu' and it's convenient to just pass
it to vprintf, don't add a special type for ints. Any new int type must
be added to the list.
{
"number": 1234
}
Signed-off-by: David Sterba <dsterba@suse.com>
The 'str' type was added in ecbb6a7fcd ("btrfs-progs: add json
formatter for escaped string") but not documented. It should be used
e.g. for paths or strings from unknown origin.
Signed-off-by: David Sterba <dsterba@suse.com>
For null or boolean values the "..." quoting must not be done, add
support for that. This is detected internally for each printed value.
Signed-off-by: David Sterba <dsterba@suse.com>
A newline character in option description text will break line and then
indent the text properly, can be used for lists or paragraphs.
Signed-off-by: David Sterba <dsterba@suse.com>
To be able to test errors at specific locations, add a simple way to
check for a condition in code and controlled from user space environment
variable INJECT. For now a single value is accepted.
Use like:
if (inject_error(0x1234)) {
do_something();
return -ERROR;
}
This is enabled in debugging build by default (make D=1) and can be
enabled on demand too (make EXTRA_CFLAGS=-DINJECT).
Signed-off-by: David Sterba <dsterba@suse.com>
There's a report that btrfs-find-root does not work as built-in tool in
btrfs.box, while it's advertised in the help:
$ ./btrfs.box help --box
Standalone tools built-in in the busybox style:
- mkfs.btrfs
- btrfs-image
- btrfs-convert
- btrfstune
- btrfs-find-root
Add the support as it might be useful tool sometimes. In the future the
command should be moved to e.g. inspect-internal or rescue.
Issue: #648
Signed-off-by: David Sterba <dsterba@suse.com>
The function check_where_mounted() scans the system for all other btrfs
devices, which is necessary for its operation. However, in certain
cases, devices remaining in the scanned state is undesirable. Introduce
the 'noscan' argument to make devices unscanned before return.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
To prepare for handling command line given devices factor out
btrfs_scan_argv_devices().
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The variable 'is_btrfs' is declared as an integer but should be a boolean
instead.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The send.h for libbtrfs has been separated some time ago so we're now
free to keep up with kernel, 6.4-rc1.
Signed-off-by: David Sterba <dsterba@suse.com>
The following functions accept a buffer for write, which can be marked
as const:
- btrfs_pwrite()
- write_data_to_disk()
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
It's not a common practice to use the same io function for both read and
write (we have pread() and pwrite(), not pio()).
Furthermore the original function has the following problems:
- Not returning proper error number
If we had ioctl/stat errors we just return 0 with errno set.
Thus caller would treat it as a short read, not a proper error.
- Unnecessary @ret_rw
This is not that obvious if we have different handling for read and
write, but if we split them it's super obvious we can reuse @ret.
- No proper copy back for short read
- Unable to constify the @buf pointer for write operation
All those problems would be addressed in this patch.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The fixes involve the following changes:
- Unexport functions which are not utilized out of the file
* print_path_column()
* parse_reflink_range()
* btrfs_list_setup_print_column()
* device_get_partition_size_sysfs()
* max_zone_append_size()
- Include related headers before implementing the function
* change-uuid.c
* convert-bgt.c
* seed.h
- Add missing headers caused by the above header changes
* include <uuid/uuid.h> for tune/tune.h.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We won't actually use the async code in progs, however we call the
helpers and such all over the normal code, so sync this into btrfs-progs
to make syncing other parts of the kernel easier.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Glibc provides an interface to extend the printf formats but this is not
standardized and does not work on musl. The code brought from kernel
uses %pV for varargs and also has own implementation of printk.
As a workaround for musl expand the pV value to a string and then
simply print it. The details are hidden behind macros:
- DECLARE_PV(vaf)
- PV_ASSIGN(vaf, format, args)
- PV_FMT in printf string
- PV_VAL in arguments
Signed-off-by: David Sterba <dsterba@suse.com>
We use the struct va_format to do nested printk's internally with our
message handling. Add the appropriate user space code to make this work
properly so when we start copying this code into btrfs-progs we get the
proper messages.
Note: this breaks build on musl, printf.h is not available.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These are the printk helpers from the kernel. There were a few
modifications, the hi-lights are
- We do not have fs_info::fs_state, so that needed to be removed.
- We do not have discard.h sync'ed yet, so that dependency was dropped.
- Anything related to struct super_block was commented out.
- The transaction abort had to be modified to fit with the current
btrfs-progs code.
- Added a btrfs_no_printk() helper to common/messages.* so that the
print statements still worked.
- The 32bit limit checkers are not needed so are behind __KERNEL__
Additionally there were kerncompat.h changes that needed to be made to
handle the dependencies properly. Those are easier to spot.
Any function that needed to be modified has a MODIFIED tag in the
comment section with a list of things that were changed.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We want to keep this file locally as we want to be uptodate with
upstream, so we can build btrfs-progs regardless of which kernel is
currently installed. Sync this with the upstream version and put it in
kernel-shared/uapi to maintain some semblance of where this file comes
from.
There are some changes that need to be synced back to kernel. A local
definition of static_assert is used to avoid compilation problems on gcc
(< 9) due to mandatory 2nd parameter.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These helpers all do variations on the same thing, so add a helper to
just do the printf part, and a macro to handle the special prefix and
postfix, and then make the helpers just use the macro and new helper.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have different helpers for warning_on and error_on(), which take the
condition and then do the printf. However we can just check the
condition in the macro and call the normal warning or error helper, so
clean this usage up and delete the unneeded message helpers.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Everybody who calls print_trace wraps it around this check, move the
check instead to print_trace and remove the check from all the callers.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
While syncing messages.[ch] I had to back out the ASSERT() code in
kerncompat.h, which means we now rely on the kernel code for ASSERT().
In order to maintain some semblance of separation introduce UASSERT()
and use that in all the purely userspace code.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The block-group-tree used to be under experimental flag in -R but now
that we've deprecated -R it does not make sense to leave
block-group-tree there for compatibility, this has never been exposed to
users.
Signed-off-by: David Sterba <dsterba@suse.com>
The feedback from the community on block group tree is very positive,
the only complain is, end users need to recompile btrfs-progs with
experimental features to enjoy the new feature.
So let's move it out of experimental features and let more people enjoy
faster mount speed.
Also change the option of btrfstune, from `-b` to
`--enable-block-group-tree` to avoid short option.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The option -R|--runtime-features was introduced to support features that
don't result in a full incompat flag change, thus things like
free-space-tree and quota features are put here.
But to end users, such separation of features is not helpful and can be
sometimes confusing.
Thus we're already migrating those runtime features into -O|--features
option under experimental builds.
I believe this is the proper time to move those runtime features into
-O|--features option, and mark the -R|--runtime-features option
deprecated.
For now we still keep the old option as for compatibility purposes.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's a report that a static build fails when there's a static version
of libudev:
/usr/lib/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../x86_64-pc-linux-gnu/bin/ld: /usr/lib/libudev.a(path-util.o): in function `path_is_mount_point':
path-util.c:(.text+0xbc0): multiple definition of `path_is_mount_point'; common/path-utils.o:path-utils.c:(.text+0x290): first defined here
There's a helper path_is_mount_point in libudev too but not exported so
dynamic library is fine, unlike static build. The static build of
libudev is not common but we can support that with a simple rename.
Issue: #611
Signed-off-by: David Sterba <dsterba@suse.com>
On a 32bit host the split qgroupid is wrong due to the way the numbers
are passed to the formatter as variable length arguments. The level is
u16, promoted to int and then parsed as u64. This means that the values
are shifted and some stack data are printed instead.
Example error messages from yast2-bootloader:
SystemCmd.cc(addLine):569 Adding Line 7 " "qgroupid": "21474836480/23885859321282560","
The value 21474836480 = 0x5000000 is 0x5 shifted by 32 bits,
23885859321282560 is 0x54dc1000000000 and shifting by 32 does not
lead to a valid value which should be 0 in this case.
Bugzilla: https://bugzilla.suse.com/show_bug.cgi?id=1209136
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
Currently cli/009 test case failed with different exit number:
====== RUN CHECK /home/adam/btrfs-progs/btrfstune --help
usage: btrfstune [options] device
[...]
failed: /home/adam/btrfs-progs/btrfstune --help
test failed for case 009-btrfstune
[CAUSE]
In tune/main.c, we have the following call on usage():
static void print_usage(int ret)
{
usage(&tune_cmd);
exit(ret);
}
However usage() itself would always call exit(1):
void usage(const struct cmd_struct *cmd)
{
usage_command_usagestr(cmd->usagestr, NULL, 0, true, true);
exit(1);
}
This makes prevents any caller of usage() to modify its exit number.
[FIX]
Add a new argument @error for print_usage(), so we can properly return 0
for -h/--help usage.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add CPU_FLAG_NONE that's first in the list and represents a CPU without
any acceleration features. Can be used for reference/fallback
implementation.
Signed-off-by: David Sterba <dsterba@suse.com>
Add support for run-time detection of CPU features on x86_64 to allow
selection of accelerated implementations of hash algorithms.
When possible use the compiler builtin (works on gcc and clang).
The SHA extensions can't be detected by __builtin_cpu_supports and the
__cpuid/__cpuidex macros are not consistently provided in all supported
gcc and clang versions. Copy the __cpuidex and call it manually for the
SHA extensions. Complete list https://en.wikipedia.org/wiki/CPUID .
Signed-off-by: David Sterba <dsterba@suse.com>
[FALSE ALERT]
Unlike gcc, clang doesn't really understand the comments, thus it's
reportings tons of fall through related errors:
cmds/reflink.c:124:3: warning: unannotated fall-through between switch labels [-Wimplicit-fallthrough]
case 'r':
^
cmds/reflink.c:124:3: note: insert '__attribute__((fallthrough));' to silence this warning
case 'r':
^
__attribute__((fallthrough));
cmds/reflink.c:124:3: note: insert 'break;' to avoid fall-through
case 'r':
^
break;
[CAUSE]
Although gcc is fine with /* fallthrough */ comments, clang is not.
[FIX]
So just introduce a fallthrough macro to handle the situation properly,
and use that macro instead.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Use the configured widths and print padding directly instead of the
embedded printf format and fixed width strings.
Signed-off-by: David Sterba <dsterba@suse.com>
