The `reftable_block` structure associates a byte slice with a block
source. As such it only holds the data of a reftable block without
actually encoding any of the details for how to access that data.
Rename the structure to instead be called `reftable_block_data`. Besides
clarifying that this really only holds data, it also allows us to rename
the `reftable_block_reader` to `reftable_block` in the next commit, as
this is the structure that actually encapsulates access to the reftable
blocks.
Rename the `struct reftable_block_reader::block` member accordingly.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code that implements block sources is distributed across a couple of
files. Consolidate all of it into "reftable/blocksource.c" and its
accompanying header so that it is easier to locate and more self
contained.
While at it, rename some of the functions to have properly scoped names.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The license headers used across the reftable library doesn't follow our
typical coding style for multi-line comments. Fix it.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce the `REFTABLE_UNUSED` annotation and replace all existing
users of `UNUSED` in the reftable library to use the new macro instead.
Note that we unconditionally define `MAYBE_UNUSED` in the exact same
way, so doing so unconditionally for `REFTABLE_UNUSED` should be fine,
too.
Suggested-by: Toon Claes <toon@iotcl.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We use `xmmap()` to map reftables into memory. This function has two
problems:
- It causes us to die in case the mmap fails.
- It ties us to the Git codebase.
Refactor the code to use mmap(3p) instead with manual error checking.
Note that this function may not be the system-provided mmap(3p), but may
point to our `git_mmap()` wrapper that emulates the syscall on systems
that do not have mmap(3p) available.
Fix `reftable_block_source_from_file()` to properly bubble up the error
code in case the map(3p) call fails.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The `block_source_read_block()` function and its implementations return
an integer as a result that reflects either the number of bytes read, or
an error. As such its return type, a signed integer, isn't wrong, but it
doesn't give the reader a good hint what it actually returns.
Refactor the function to return an `ssize_t` instead, which is typical
for functions similar to read(3p) and should thus give readers a better
signal what they can expect as a result.
Adjust callers to better handle the returned value to avoid warnings
with -Wsign-compare. One of these callers is `reader_get_block()`, whose
return value is only ever used by its callers to figure out whether or
not the read was successful. So instead of bubbling up the `ssize_t`
there, too, we adapt it to only indicate success or errors.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Adapt the name of the `strbuf` block source to no longer relate to this
interface, but instead to the `reftable_buf` interface.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Convert the reftable library to use the `reftable_buf` interface instead
of the `strbuf` interface. This is mostly a mechanical change via sed(1)
with some manual fixes where functions for `strbuf` and `reftable_buf`
differ. The converted code does not yet handle allocation failures. This
will be handled in subsequent commits.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
The reftable blocksource provides a generic interface to read blocks via
different sources, e.g. from disk or from memory. One of the block
sources is the malloc block source, which can in theory read data from
memory. We nowadays also have a strbuf block source though, which
provides essentially the same functionality with better ergonomics.
Adapt the only remaining user of the malloc block source in our tests
to use the strbuf block source, instead, and remove the now-unused
malloc block source.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The reftable code uses a lot of virtual function pointers, but many of
the concrete implementations do not need all of the parameters.
For the most part these are obviously fine to just mark as UNUSED (e.g.,
the empty_iterator functions unsurprisingly do not do anything). Here
are a few cases where I dug a little deeper (but still ended up just
marking them UNUSED):
- the iterator exclude_patterns is best-effort and optional (though it
would be nice to support in the long run as an optimization)
- ignoring the ref_store in many transaction functions is unexpected,
but works because the ref_transaction itself carries enough
information to do what we need.
- ignoring "err" for in some cases (e.g., transaction abort) is OK
because we do not return any errors. It is a little odd for
reftable_be_create_reflog(), though, since we do return errors
there. We should perhaps be creating string error messages at this
layer, but I've punted on that for now.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Similar to the preceding commit, let's carry over macros to allocate
arrays with `REFTABLE_ALLOC_ARRAY()` and `REFTABLE_CALLOC_ARRAY()`. This
requires us to change the signature of `reftable_calloc()`, which only
takes a single argument right now and thus puts the burden on the caller
to calculate the final array's size. This is a net improvement though as
it means that we can now provide proper overflow checks when multiplying
the array size with the member size.
Convert callsites of `reftable_calloc()` to the new signature and start
using the new macros where possible.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The blocksource interface provides an interface to read blocks from a
reftable table. This interface is implemented using read(3P) calls on
the underlying file descriptor. While this works alright, this pattern
is very inefficient when repeatedly querying the reftable stack for one
or more refs. This inefficiency can mostly be attributed to the fact
that we often need to re-read the same blocks over and over again, and
every single time we need to call read(3P) again.
A natural fit in this context is to use mmap(3P) instead of read(3P),
which has a bunch of benefits:
- We do not need to come up with a caching strategy for some of the
blocks as this will be handled by the kernel already.
- We can avoid the overhead of having to call into the read(3P)
syscall repeatedly.
- We do not need to allocate returned blocks repeatedly, but can
instead hand out pointers into the mmapped region directly.
Using mmap comes with a significant drawback on Windows though, because
mmapped files cannot be deleted and neither is it possible to rename
files onto an mmapped file. But for one, the reftable library gracefully
handles the case where auto-compaction cannot delete a still-open stack
already and ignores any such errors. Also, `reftable_stack_clean()` will
prune stale tables which are not referenced by "tables.list" anymore so
that those files can eventually be pruned. And second, we never rewrite
already-written stacks, so it does not matter that we cannot rename a
file over an mmaped file, either.
Another unfortunate property of mmap is that it is not supported by all
systems. But given that the size of reftables should typically be rather
limited (megabytes at most in the vast majority of repositories), we can
use the fallback implementation provided by `git_mmap()` which reads the
whole file into memory instead. This is the same strategy that the
"packed" backend uses.
While this change doesn't significantly improve performance in the case
where we're seeking through stacks once (like e.g. git-for-each-ref(1)
would). But it does speed up usecases where there is lots of random
access to refs, e.g. when writing. The following benchmark demonstrates
these savings with git-update-ref(1) creating N refs in an otherwise
empty repository:
Benchmark 1: update-ref: create many refs (refcount = 1, revision = HEAD~)
Time (mean ± σ): 5.1 ms ± 0.2 ms [User: 2.5 ms, System: 2.5 ms]
Range (min … max): 4.8 ms … 7.1 ms 111 runs
Benchmark 2: update-ref: create many refs (refcount = 100, revision = HEAD~)
Time (mean ± σ): 14.8 ms ± 0.5 ms [User: 7.1 ms, System: 7.5 ms]
Range (min … max): 14.1 ms … 18.7 ms 84 runs
Benchmark 3: update-ref: create many refs (refcount = 10000, revision = HEAD~)
Time (mean ± σ): 926.4 ms ± 5.6 ms [User: 448.5 ms, System: 477.7 ms]
Range (min … max): 920.0 ms … 936.1 ms 10 runs
Benchmark 4: update-ref: create many refs (refcount = 1, revision = HEAD)
Time (mean ± σ): 5.0 ms ± 0.2 ms [User: 2.4 ms, System: 2.5 ms]
Range (min … max): 4.7 ms … 5.4 ms 111 runs
Benchmark 5: update-ref: create many refs (refcount = 100, revision = HEAD)
Time (mean ± σ): 10.5 ms ± 0.2 ms [User: 5.0 ms, System: 5.3 ms]
Range (min … max): 10.0 ms … 10.9 ms 93 runs
Benchmark 6: update-ref: create many refs (refcount = 10000, revision = HEAD)
Time (mean ± σ): 529.6 ms ± 9.1 ms [User: 268.0 ms, System: 261.4 ms]
Range (min … max): 522.4 ms … 547.1 ms 10 runs
Summary
update-ref: create many refs (refcount = 1, revision = HEAD) ran
1.01 ± 0.06 times faster than update-ref: create many refs (refcount = 1, revision = HEAD~)
2.08 ± 0.07 times faster than update-ref: create many refs (refcount = 100, revision = HEAD)
2.95 ± 0.14 times faster than update-ref: create many refs (refcount = 100, revision = HEAD~)
105.33 ± 3.76 times faster than update-ref: create many refs (refcount = 10000, revision = HEAD)
184.24 ± 5.89 times faster than update-ref: create many refs (refcount = 10000, revision = HEAD~)
Theoretically, we could also replicate the strategy of the "packed"
backend where small tables are read into memory instead of using mmap.
Benchmarks did not confirm that this has a performance benefit though.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor `reftable_block_source_from_file()` to match our coding style
better.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are calls to pread(3P) and read(3P) where we don't properly handle
interrupts. Convert them to use `pread_in_full()` and `read_in_full()`,
respectively.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
1214aa841b (reftable: add blocksource, an abstraction for random
access reads, 2021-10-07), makes the assumption that it is ok to
free a reftable_block pointing to NULL if the size is also set to
0, but implements that using a memset call that at least in glibc
based system will trigger a runtime exception if called with a
NULL pointer as its first parameter.
Avoid doing so by adding a conditional to check for the size in all
three identically looking functions that were affected, and therefore,
still allow memset to help catch callers that might incorrectly pass
a NULL pointer with a non zero size, but avoiding the exception for
the valid cases.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This would be triggered in the unlikely event of fstat() failing on an
opened file.
Signed-off-by: Han-Wen Nienhuys <hanwen@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The reftable format is usually used with files for storage. However, we abstract
away this using the blocksource data structure. This has two advantages:
* log blocks are zlib compressed, and handling them is simplified if we can
discard byte segments from within the block layer.
* for unittests, it is useful to read and write in-memory. The blocksource
allows us to abstract the data away from on-disk files.
Signed-off-by: Han-Wen Nienhuys <hanwen@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Patrick Steinhardt