Message ID | cover.1620145501.git.matheus.bernardino@usp.br (mailing list archive) |
---|---|
Headers | show |
Series | Parallel Checkout (part 3) | expand |
On 5/4/2021 12:27 PM, Matheus Tavares wrote: > This is the last part of the parallel checkout series. It adds tests and > parallel checkout support to `git checkout-index` and > `git checkout <pathspec>`. > > I rebased this version onto `master`, as part-2 was merged down to > `master`. I read the range-diff and gave the patches another pass. This version looks good to me. Thanks, -Stolee
Derrick Stolee <stolee@gmail.com> writes: > On 5/4/2021 12:27 PM, Matheus Tavares wrote: >> This is the last part of the parallel checkout series. It adds tests and >> parallel checkout support to `git checkout-index` and >> `git checkout <pathspec>`. >> >> I rebased this version onto `master`, as part-2 was merged down to >> `master`. > > I read the range-diff and gave the patches another pass. This version > looks good to me. Thanks. Your assessment matches mine.
This is the last part of the parallel checkout series. It adds tests and parallel checkout support to `git checkout-index` and `git checkout <pathspec>`. I rebased this version onto `master`, as part-2 was merged down to `master`. Changes since v2: * Patch 1: Aligned function parameters in make_transient_cache_entry(), and renamed the `struct mem_pool *` argument from `mp` to `ce_mem_pool`. This is more consistent with the other functions using in read-cache.c and with how we call this function in the next patch. * Patch 2: Removed unnecessary 'Note: ' from commit message. * Patch 3: Rewrote commit message to better explain why we unified the exit paths for `checkout_all()` and `checkout_file()` modes, and changed `git checkout-index --all`s error code from 128 to 1. Matheus Tavares (8): make_transient_cache_entry(): optionally alloc from mem_pool builtin/checkout.c: complete parallel checkout support checkout-index: add parallel checkout support parallel-checkout: add tests for basic operations parallel-checkout: add tests related to path collisions t0028: extract encoding helpers to lib-encoding.sh parallel-checkout: add tests related to .gitattributes ci: run test round with parallel-checkout enabled builtin/checkout--worker.c | 2 +- builtin/checkout-index.c | 24 ++- builtin/checkout.c | 20 ++- builtin/difftool.c | 2 +- cache.h | 14 +- ci/run-build-and-tests.sh | 1 + parallel-checkout.c | 18 ++ read-cache.c | 14 +- t/README | 4 + t/lib-encoding.sh | 25 +++ t/lib-parallel-checkout.sh | 45 +++++ t/t0028-working-tree-encoding.sh | 25 +-- t/t2080-parallel-checkout-basics.sh | 229 ++++++++++++++++++++++++ t/t2081-parallel-checkout-collisions.sh | 162 +++++++++++++++++ t/t2082-parallel-checkout-attributes.sh | 194 ++++++++++++++++++++ unpack-trees.c | 2 +- 16 files changed, 732 insertions(+), 49 deletions(-) create mode 100644 t/lib-encoding.sh create mode 100644 t/lib-parallel-checkout.sh create mode 100755 t/t2080-parallel-checkout-basics.sh create mode 100755 t/t2081-parallel-checkout-collisions.sh create mode 100755 t/t2082-parallel-checkout-attributes.sh Range-diff against v2: 1: f870040bfb ! 1: bf6c7114aa make_transient_cache_entry(): optionally alloc from mem_pool @@ cache.h: struct cache_entry *make_empty_cache_entry(struct index_state *istate, - * Create a cache_entry that is not intended to be added to an index. - * Caller is responsible for discarding the cache_entry - * with `discard_cache_entry`. -+ * Create a cache_entry that is not intended to be added to an index. If `mp` -+ * is not NULL, the entry is allocated within the given memory pool. Caller is -+ * responsible for discarding "loose" entries with `discard_cache_entry()` and -+ * the mem_pool with `mem_pool_discard(mp, should_validate_cache_entries())`. ++ * Create a cache_entry that is not intended to be added to an index. If ++ * `ce_mem_pool` is not NULL, the entry is allocated within the given memory ++ * pool. Caller is responsible for discarding "loose" entries with ++ * `discard_cache_entry()` and the memory pool with ++ * `mem_pool_discard(ce_mem_pool, should_validate_cache_entries())`. */ struct cache_entry *make_transient_cache_entry(unsigned int mode, const struct object_id *oid, const char *path, - int stage); -+ int stage, struct mem_pool *mp); ++ int stage, ++ struct mem_pool *ce_mem_pool); -struct cache_entry *make_empty_transient_cache_entry(size_t name_len); -+struct cache_entry *make_empty_transient_cache_entry(size_t len, struct mem_pool *mp); ++struct cache_entry *make_empty_transient_cache_entry(size_t len, ++ struct mem_pool *ce_mem_pool); /* * Discard cache entry. @@ read-cache.c: struct cache_entry *make_empty_cache_entry(struct index_state *ist } -struct cache_entry *make_empty_transient_cache_entry(size_t len) -+struct cache_entry *make_empty_transient_cache_entry(size_t len, struct mem_pool *mp) ++struct cache_entry *make_empty_transient_cache_entry(size_t len, ++ struct mem_pool *ce_mem_pool) { -+ if (mp) -+ return mem_pool__ce_calloc(mp, len); ++ if (ce_mem_pool) ++ return mem_pool__ce_calloc(ce_mem_pool, len); return xcalloc(1, cache_entry_size(len)); } @@ read-cache.c: struct cache_entry *make_cache_entry(struct index_state *istate, - const char *path, int stage) +struct cache_entry *make_transient_cache_entry(unsigned int mode, + const struct object_id *oid, -+ const char *path, int stage, -+ struct mem_pool *mp) ++ const char *path, ++ int stage, ++ struct mem_pool *ce_mem_pool) { struct cache_entry *ce; int len; @@ read-cache.c: struct cache_entry *make_transient_cache_entry(unsigned int mode, len = strlen(path); - ce = make_empty_transient_cache_entry(len); -+ ce = make_empty_transient_cache_entry(len, mp); ++ ce = make_empty_transient_cache_entry(len, ce_mem_pool); oidcpy(&ce->oid, oid); memcpy(ce->name, path, len); 2: e2d82c4337 ! 2: e898889787 builtin/checkout.c: complete parallel checkout support @@ Commit message checkout_entry() directly, instead of unpack_trees(). Let's add parallel checkout support for this code path too. - Note: the transient cache entries allocated in checkout_merged() are now + The transient cache entries allocated in checkout_merged() are now allocated in a mem_pool which is only discarded after parallel checkout finishes. This is done because the entries need to be valid when run_parallel_checkout() is called. @@ builtin/checkout.c: static int checkout_worktree(const struct checkout_opts *opt enable_delayed_checkout(&state); + if (pc_workers > 1) + init_parallel_checkout(); - for (pos = 0; pos < active_nr; pos++) { - struct cache_entry *ce = active_cache[pos]; - if (ce->ce_flags & CE_MATCHED) { + + /* TODO: audit for interaction with sparse-index. */ + ensure_full_index(&the_index); @@ builtin/checkout.c: static int checkout_worktree(const struct checkout_opts *opts, &nr_checkouts, opts->overlay_mode); else if (opts->merge) 3: 0fe1a5fabc ! 3: 916f391a46 checkout-index: add parallel checkout support @@ Metadata ## Commit message ## checkout-index: add parallel checkout support - Note: previously, `checkout_all()` would not return on errors, but - instead call `exit()` with a non-zero code. However, it only did that - after calling `checkout_entry()` for all index entries, thus not - stopping on the first error, but attempting to write the maximum number - of entries possible. In order to maintain this behavior we now propagate - `checkout_all()`s error status to `cmd_checkout_index()`, so that it can - call `run_parallel_checkout()` and attempt to write the queued entries - before exiting with the error code. + Allow checkout-index to use the parallel checkout framework, honoring + the checkout.workers configuration. + + There are two code paths in checkout-index which call + `checkout_entry()`, and thus, can make use of parallel checkout: + `checkout_file()`, which is used to write paths explicitly given at the + command line; and `checkout_all()`, which is used to write all paths in + the index, when the `--all` option is given. + + In both operation modes, checkout-index doesn't abort immediately on a + `checkout_entry()` failure. Instead, it tries to check out all remaining + paths before exiting with a non-zero exit code. To keep this behavior + when parallel checkout is being used, we must allow + `run_parallel_checkout()` to try writing the queued entries before we + exit, even if we already got an error code from a previous + `checkout_entry()` call. + + However, `checkout_all()` doesn't return on errors, it calls `exit()` + with code 128. We could make it call `run_parallel_checkout()` before + exiting, but it makes the code easier to follow if we unify the exit + path for both checkout-index modes at `cmd_checkout_index()`, and let + this function take care of the interactions with the parallel checkout + API. So let's do that. + + With this change, we also have to consider whether we want to keep using + 128 as the error code for `git checkout-index --all`, while we use 1 for + `git checkout-index <path>` (even when the actual error is the same). + Since there is not much value in having code 128 only for `--all`, and + there is no mention about it in the docs (so it's unlikely that changing + it will break any existing script), let's make both modes exit with code + 1 on `checkout_entry()` errors. Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br> 4: f604c50dba = 4: 667777053a parallel-checkout: add tests for basic operations 5: eae6d3a1c1 = 5: dcb3acab1d parallel-checkout: add tests related to path collisions 6: 9161cd1503 = 6: 6141c46051 t0028: extract encoding helpers to lib-encoding.sh 7: bc584897e6 = 7: 5350689a30 parallel-checkout: add tests related to .gitattributes 8: 1bc5c523c5 ! 8: 7b2966c488 ci: run test round with parallel-checkout enabled @@ parallel-checkout.c: static const int DEFAULT_NUM_WORKERS = 1; else if (*num_workers < 1) ## t/README ## -@@ t/README: and "sha256". - GIT_TEST_WRITE_REV_INDEX=<boolean>, when true enables the - 'pack.writeReverseIndex' setting. +@@ t/README: GIT_TEST_WRITE_REV_INDEX=<boolean>, when true enables the + GIT_TEST_SPARSE_INDEX=<boolean>, when true enables index writes to use the + sparse-index format by default. +GIT_TEST_CHECKOUT_WORKERS=<n> overrides the 'checkout.workers' setting +to <n> and 'checkout.thresholdForParallelism' to 0, forcing the