| Message ID | pull.930.v2.git.1618139107203.gitgitgadget@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] merge-ort: only do pointer arithmetic for non-empty lists | expand |
On Sun, Apr 11, 2021 at 4:05 AM Andrzej Hunt via GitGitGadget <gitgitgadget@gmail.com> wrote: > > From: Andrzej Hunt <ajrhunt@google.com> > > versions could be an empty string_list. In that case, versions->items is > NULL, and we shouldn't be trying to perform pointer arithmetic with it (as > that results in undefined behaviour). > > Moreover we only use the results of this calculation once when calling > QSORT. Therefore we choose to skip creating relevant_entries and call > QSORT directly with our manipulated pointers (but only if there's data > requiring sorting). This lets us avoid abusing the string_list API, > and saves us from having to explain why this abuse is OK. Oh, nice, I like that. > > Finally, an assertion is added to make sure that write_tree() is called > with a valid offset. > > This issue has probably existed since: > ee4012dcf9 (merge-ort: step 2 of tree writing -- function to create tree object, 2020-12-13) > But it only started occurring during tests since tests started using > merge-ort: > f3b964a07e (Add testing with merge-ort merge strategy, 2021-03-20) > > For reference - here's the original UBSAN commit that implemented this > check, it sounds like this behaviour isn't actually likely to cause any > issues (but we might as well fix it regardless): > https://reviews.llvm.org/D67122 > > UBSAN output from t3404 or t5601: > > merge-ort.c:2669:43: runtime error: applying zero offset to null pointer > #0 0x78bb53 in write_tree merge-ort.c:2669:43 > #1 0x7856c9 in process_entries merge-ort.c:3303:2 > #2 0x782317 in merge_ort_nonrecursive_internal merge-ort.c:3744:2 > #3 0x77feef in merge_incore_nonrecursive merge-ort.c:3853:2 > #4 0x6f6a5c in do_recursive_merge sequencer.c:640:3 > #5 0x6f6a5c in do_pick_commit sequencer.c:2221:9 > #6 0x6ef055 in single_pick sequencer.c:4814:9 > #7 0x6ef055 in sequencer_pick_revisions sequencer.c:4867:10 > #8 0x4fb392 in run_sequencer revert.c:225:9 > #9 0x4fa5b0 in cmd_revert revert.c:235:8 > #10 0x42abd7 in run_builtin git.c:453:11 > #11 0x429531 in handle_builtin git.c:704:3 > #12 0x4282fb in run_argv git.c:771:4 > #13 0x4282fb in cmd_main git.c:902:19 > #14 0x524b63 in main common-main.c:52:11 > #15 0x7fc2ca340349 in __libc_start_main (/lib64/libc.so.6+0x24349) > #16 0x4072b9 in _start start.S:120 > > SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior merge-ort.c:2669:43 in > > Signed-off-by: Andrzej Hunt <ajrhunt@google.com> > --- > merge-ort: only do pointer arithmetic for non-empty lists > > V2 removes relevant_entries as per René's suggestion, and adds an > assertion as per Junio's question. > > Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-930%2Fahunt%2Fmerge-ort-ubsan-v2 > Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-930/ahunt/merge-ort-ubsan-v2 > Pull-Request: https://github.com/gitgitgadget/git/pull/930 > > Range-diff vs v1: > > 1: d52ce7110446 ! 1: 48c1a0ad6f46 merge-ort: only do pointer arithmetic for non-empty lists > @@ Commit message > NULL, and we shouldn't be trying to perform pointer arithmetic with it (as > that results in undefined behaviour). > > + Moreover we only use the results of this calculation once when calling > + QSORT. Therefore we choose to skip creating relevant_entries and call > + QSORT directly with our manipulated pointers (but only if there's data > + requiring sorting). This lets us avoid abusing the string_list API, > + and saves us from having to explain why this abuse is OK. > + > + Finally, an assertion is added to make sure that write_tree() is called > + with a valid offset. > + > This issue has probably existed since: > ee4012dcf9 (merge-ort: step 2 of tree writing -- function to create tree object, 2020-12-13) > But it only started occurring during tests since tests started using > @@ Commit message > > ## merge-ort.c ## > @@ merge-ort.c: static void write_tree(struct object_id *result_oid, > - * We won't use relevant_entries again and will let it just pop off the > - * stack, so there won't be allocation worries or anything. > - */ > + size_t maxlen = 0, extra; > + unsigned int nr = versions->nr - offset; Should we remove the initialization here and just define nr only, and then... > + struct strbuf buf = STRBUF_INIT; > +- struct string_list relevant_entries = STRING_LIST_INIT_NODUP; > + int i; > + > +- /* > +- * We want to sort the last (versions->nr-offset) entries in versions. > +- * Do so by abusing the string_list API a bit: make another string_list > +- * that contains just those entries and then sort them. > +- * > +- * We won't use relevant_entries again and will let it just pop off the > +- * stack, so there won't be allocation worries or anything. > +- */ > - relevant_entries.items = versions->items + offset; > - relevant_entries.nr = versions->nr - offset; > -+ if (versions->nr) { > -+ relevant_entries.items = versions->items + offset; > -+ relevant_entries.nr = versions->nr - offset; > -+ } > - QSORT(relevant_entries.items, relevant_entries.nr, tree_entry_order); > +- QSORT(relevant_entries.items, relevant_entries.nr, tree_entry_order); > ++ assert(offset <= versions->nr); Define nr here after this assert? The definition of nr also depended implicitly on this requirement, so if we're making this requirement explicit, it would make sense to define nr after stating the requirement. > ++ if (versions->nr) > ++ QSORT(versions->items + offset, nr, tree_entry_order); > > /* Pre-allocate some space in buf */ > + extra = hash_size + 8; /* 8: 6 for mode, 1 for space, 1 for NUL char */ > > > merge-ort.c | 15 +++------------ > 1 file changed, 3 insertions(+), 12 deletions(-) > > diff --git a/merge-ort.c b/merge-ort.c > index 5e118a85ee04..702eb5bf362d 100644 > --- a/merge-ort.c > +++ b/merge-ort.c > @@ -2493,20 +2493,11 @@ static void write_tree(struct object_id *result_oid, > size_t maxlen = 0, extra; > unsigned int nr = versions->nr - offset; > struct strbuf buf = STRBUF_INIT; > - struct string_list relevant_entries = STRING_LIST_INIT_NODUP; > int i; > > - /* > - * We want to sort the last (versions->nr-offset) entries in versions. > - * Do so by abusing the string_list API a bit: make another string_list > - * that contains just those entries and then sort them. > - * > - * We won't use relevant_entries again and will let it just pop off the > - * stack, so there won't be allocation worries or anything. > - */ > - relevant_entries.items = versions->items + offset; > - relevant_entries.nr = versions->nr - offset; > - QSORT(relevant_entries.items, relevant_entries.nr, tree_entry_order); > + assert(offset <= versions->nr); > + if (versions->nr) > + QSORT(versions->items + offset, nr, tree_entry_order); > > /* Pre-allocate some space in buf */ > extra = hash_size + 8; /* 8: 6 for mode, 1 for space, 1 for NUL char */ > > base-commit: 89b43f80a514aee58b662ad606e6352e03eaeee4 > -- > gitgitgadget Otherwise, this patch looks good to me; thanks!
Elijah Newren <newren@gmail.com> writes: > Otherwise, this patch looks good to me; thanks! Queued, manually amended, and here is what I have. $ git range-diff @{1}... 1: ca4cc8d182 ! 1: c1ea48a8f7 merge-ort: only do pointer arithmetic for non-empty lists @@ Commit message SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior merge-ort.c:2669:43 in Signed-off-by: Andrzej Hunt <ajrhunt@google.com> + Reviewed-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com> ## merge-ort.c ## @@ merge-ort.c: static void write_tree(struct object_id *result_oid, + size_t hash_size) + { size_t maxlen = 0, extra; - unsigned int nr = versions->nr - offset; +- unsigned int nr = versions->nr - offset; ++ unsigned int nr; struct strbuf buf = STRBUF_INIT; - struct string_list relevant_entries = STRING_LIST_INIT_NODUP; int i; @@ merge-ort.c: static void write_tree(struct object_id *result_oid, - relevant_entries.nr = versions->nr - offset; - QSORT(relevant_entries.items, relevant_entries.nr, tree_entry_order); + assert(offset <= versions->nr); ++ nr = versions->nr - offset; + if (versions->nr) + QSORT(versions->items + offset, nr, tree_entry_order);
Elijah Newren <newren@gmail.com> writes: >> /* Pre-allocate some space in buf */ >> extra = hash_size + 8; /* 8: 6 for mode, 1 for space, 1 for NUL char */ >> >> base-commit: 89b43f80a514aee58b662ad606e6352e03eaeee4 >> -- >> gitgitgadget > > Otherwise, this patch looks good to me; thanks! By the way, I noticed the post-context comment and got curious. /* Pre-allocate some space in buf */ extra = hash_size + 8; /* 8: 6 for mode, 1 for space, 1 for NUL char */ for (i = 0; i < nr; i++) { maxlen += strlen(versions->items[offset+i].string) + extra; } strbuf_grow(&buf, maxlen); Because "6 for mode" is wrong if it means "%06o", but the format used in the code is "%o" so there is no correctness issues in the code (Phew). This "grow" is to avoid having to repeatedly reallocate "nr" times in the loop that comes, but (1) s/6 for mode/up to 6 for mode/ to make it less misleading. (2) does this preallocation really help performance? (3) it is really disturbing to find a custom tree-writing code that is exercised only by ort here. Tree-writing has been one major source of broken objects in various reimplementation of Git--- all known ones made a mistake sometime in the past---and I'd strongly prefer to see a single helper that formats a tree object, given a set of <mode, object name, path> in the longer code health.
diff --git a/merge-ort.c b/merge-ort.c index 5e118a85ee04..702eb5bf362d 100644 --- a/merge-ort.c +++ b/merge-ort.c @@ -2493,20 +2493,11 @@ static void write_tree(struct object_id *result_oid, size_t maxlen = 0, extra; unsigned int nr = versions->nr - offset; struct strbuf buf = STRBUF_INIT; - struct string_list relevant_entries = STRING_LIST_INIT_NODUP; int i; - /* - * We want to sort the last (versions->nr-offset) entries in versions. - * Do so by abusing the string_list API a bit: make another string_list - * that contains just those entries and then sort them. - * - * We won't use relevant_entries again and will let it just pop off the - * stack, so there won't be allocation worries or anything. - */ - relevant_entries.items = versions->items + offset; - relevant_entries.nr = versions->nr - offset; - QSORT(relevant_entries.items, relevant_entries.nr, tree_entry_order); + assert(offset <= versions->nr); + if (versions->nr) + QSORT(versions->items + offset, nr, tree_entry_order); /* Pre-allocate some space in buf */ extra = hash_size + 8; /* 8: 6 for mode, 1 for space, 1 for NUL char */