In a well-formed gvar the per-glyph data ranges [offset[i],
offset[i+1]) are disjoint, so the retained glyphs' data add up to at
most the table. gvar::subset()'s avar2 culling pass used the offsets
as they are, so malformed offsets (0, D, 0, D, ...) could point
thousands of glyphs at one 128 KB block; the pass parsed, rewrote and
kept a heap copy of it once per glyph, repeated on each of the
serializer's buffer-doubling retries (26 s and 1.3 GiB from a 251 KB
font before it was rejected).
Restore the invariant where the pass is driven: charge each retained
glyph's byte length to a budget equal to the source blob length with
hb_budget_spend(), and on exhaustion raise HB_SERIALIZE_ERROR_OTHER
so _hb_subset_table_try() does not retry. A spec-conforming table
can never reach the limit, so there is no tuning constant and valid
fonts subset byte-identically; a font whose retained glyphs' gvar
ranges overlap by more than the table length now fails avar2 partial
instancing.
Adds test_avar2_culling_aliased_gvar_data to test/api/test-subset-avar2.c.
The non-avar2 instantiate() branch trusts the same offsets and is
not touched here.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-NJZQ4XFP.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Budget the glyf points a plan keeps for partial instancing
For partial instancing the subset plan decodes every retained glyph
through Glyph::get_all_points_without_var() and keeps all the points in
plan->new_gid_contour_points_map until the plan is destroyed. That
caller passed no work budget, and nothing capped the total. A simple
glyph of repeated flags declares ~65536 points in ~526 bytes, and a
loca whose offsets alias one another lets many glyph ids share those
bytes, so a 25 KB font drove plan creation to 2.2 GB.
PR #6256 added a charge-before-allocate budget to
SimpleGlyph::get_contour_points() and passed it from the draw/extents
path, but left this subsetter caller unbudgeted. Give the loop one
plan-wide budget, the larger of HB_BUDGET_SUBSET_CONTOUR_POINTS and the
source glyf length, and charge each glyph's decoded points after it is
decoded. One glyph's decode is bounded by the format (about 786 KB),
so the transient is proportional to the input; the budget bounds what
the plan retains. Exhausting it fails plan creation, the same outcome
as any other decode failure in this loop. Ordinary glyph data carries
at least one byte per point, so the size-derived bound does not
constrain well-formed fonts.
Adds an API regression test (test_subset_gvar_shared_glyph_data) with
an aliased loca over a 65536-point simple and composite glyph.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-O2ZFVVUI.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Budget the avar2 reachable-range computation
_compute_avar2_reachable_ranges() bounded only the breakpoint grid
(MAX_GRID = 16384 vertices), while its real work is
axes x vertices x active row columns x tents. A delta-set row may list
the same region tens of thousands of times, and a clamped
DeltaSetIndexMap lets every fvar axis share that one row, so a 50 KB
font cost about two minutes of CPU per hb_subset_plan_create_or_fail().
Charge one shared allowance per plan, HB_BUDGET_SUBSET_AVAR2_RANGES, at
the three places the work is created: scanning the row's columns into
the active set, collecting each active region's tents, and the exact
grid enumeration (charged up front as grid_size x (active + tents)).
Degradation reuses states the surrounding code already handles: an axis
whose exact enumeration does not fit takes the existing
interval-arithmetic bounds (still conservative, so the instance stays
correct), and once the allowance is gone the remaining axes are left
unconstrained. The hostile fonts now finish in well under a second;
real avar2 fonts use a small fraction of the budget and their
output is unchanged.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-3VDGZ019.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Guard against null glyf blob in contour-point budget
get_instance_glyphs_contour_points() sizes its op budget from the
source glyf table length. When head.glyphDataFormat (or
indexToLocFormat) is out of range the glyf accelerator is disabled and
reference_glyf_table() returns no blob, so hb_blob_get_length() then
dereferenced a null pointer and crashed during partial instancing.
Take the length only when the blob is non-null. Add a regression test
that builds a variable font with an invalid glyphDataFormat and plans a
partial instance.
Reported by Anthropic's OSS Scanner.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Document that the avar2 range budget fallback is safe
Running out of HB_BUDGET_SUBSET_AVAR2_RANGES returns early from
_compute_avar2_reachable_ranges() and records nothing for the
remaining axes. That does not produce a wrongly instanced font:
avar2_reachable_ranges is only used to cull variation data that
cannot be reached, and an axis without a range is never culled
(_hb_avar2_region_is_dead() skips it, the same as the VARC path that
resets the map). A pinned axis that is not detected as
self-contained stays in fvar as a hidden pin, which is what CFF2
fonts always get. The output is correct, only larger. Likewise the
exact-grid fallback uses interval bounds, which are conservative.
Say so in the comments instead of "left unconstrained".
Tested: meson test -C build (278/278); hb-subset --instance
wght=400 and wght=300:700 on GoogleSansFlex, NotoSansKR and
SourceSerif4 variable fonts, all glyphs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* [subset/cff] Bound parsed subroutines by the table size before allocating
subr_subsetter_t::subset() resizes one parsed_cs_str_vec_t per Font DICT
to that DICT's localSubrs->count, assuming the sum is bounded by the
table. It is not: any number of Font DICTs may point at the same
Private DICT and Subrs INDEX, so a 500 KB font with 16 Font DICTs all
sharing one 1000-subr INDEX cost about 1.6 GiB (measured: control 10
MiB, shared 1651 MiB; 2.8 GiB with hb_subset_preprocess).
Sum the global plus per-Font-DICT local subroutine counts in 64-bit
before any per-Font-DICT allocation, and fail the subset when the
records would exceed 256 bytes per byte of the source table, the same
ratio _hb_subset_table_try() uses to stop growing the output buffer.
A font whose Font DICTs do not share their Subrs needs at least one
table byte per subroutine, so it stays far below the limit; every
CFF/CFF2 font in the test corpus is at most a couple of bytes per table
byte. Over the limit the subset fails, as any CFF planning failure has
since dde7da182e; the shared font now peaks at 7 MiB instead of 1.6 GiB.
The check sits before the allocations in the non-accelerator branch, so
it also covers the hb_subset_preprocess path and both CFF1 and CFF2
instantiations of the template; the --desubroutinize flattener path
does not allocate per subroutine and is unaffected.
Adds test_subset_cff1_shared_local_subrs: a 16-Font-DICT CID font with
1000 local subrs subsets when the subrs are unshared and fails when
every Font DICT shares them.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-HPFG1IAV.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Share one CFF charstring work budget across a subsetting pass
Every per-glyph charstring loop in the subsetter (subr_subsetter_t::
subset, subr_flattener_t::flatten, the two CFF2 partial-instancing
passes, the CFF2 instance-metrics loop, and the CFF1 seac closure in
the plan and in the depend builder) created a fresh interpreter bounded
only by HB_CFF_MAX_OPS per charstring. Glyphs can share subroutines,
so a font where every glyph makes one cheap call into a deeply nested
shared subr cost num_glyphs x ~200,000 ops: a 121 KB, 30,000-glyph font
took 33 s to subset (control 0.05 s).
Give each pass one cumulative counter, max(HB_BUDGET_SUBSET_CFF = 2^24,
32 ops per table byte), threaded through cs_interpreter_t::interpret()'s
existing budget argument exactly as the outline and extents paths do.
When it runs out, interpret() fails as it does for a single over-limit
charstring, so the CFF/CFF2 table fails to subset (the subset fails, as
any CFF planning failure has since dde7da182e); the seac closure fails
the plan and the depend builder fails. The 32-ops-per-byte allowance
was measured against the in-tree corpus, where real fonts interpret at
most a few operations per CFF byte, so they stay far below it and
produce byte-identical output; the attacker font now fails in 0.17 s.
Adds test_subset_cff1_shared_subrs: four shared-subr glyphs subset
fine, while a hundred of them exhaust the budget and the subset (and
the dependency builder) fail.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-Z5NLRFA6.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* [test] Force the ot shaper in the layout budget tests
test_ot_layout_gpos_mark_base_rescan_budget calls hb_shape() with no
shaper forced, then asserts an OT-specific max_ops trip via
hb_buffer_allocation_successful(). Under the HarfRust shaper
(HB_SHAPER_LIST=harfrust, as the rust CI job re-runs the api suite)
it aborts, because harfrust shapes independently and never reaches
the C-side max_ops paths it tests.
Shape with hb_shape_full(..., {"ot", NULL}) so it exercises the OT
shaper regardless of the configured shaper list, as
test_shape_lookup_budget already does.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [layout] Charge the skip, skipped-record and ligature walks
The per-pass `len / 16` charge does not bound per-position inner work:
the skipping iterator's rescans over default-ignorables, apply_lookup's
walk over a rule's skipped LookupRecords, and match_input /
ligate_input's walk over a ligature's components and marks. Each is a
CPU-DoS a crafted font triggers under the default max_ops.
Charge each on its work path only, so honest shaping is unaffected:
skipping_iterator_t::next/prev charge one op per skip (a first-try
match charges nothing); apply_lookup charges each skipped record;
match_input charges the ligbase backward walk and ligate_input the
mark re-adjust walk, once each.
Reproducers drop from 20 s / 6.8 s / 3.8 s to 0.05 s / 0.003 s /
0.02 s; in-house and aots shaping tests pass with no output change
(the pre-existing text-rendering-tests failures are unchanged).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [layout] Stop matching once the skip/walk budget is spent
The previous commit charged the walks, but the skipping iterator's skip
arm only decremented max_ops; within one pass a tiny aliased font could
still re-run a walk per candidate, per rule or per subtable without bound
(dfrg's harfrust #604 review found the same gaps here).
Flag the buffer unsuccessful when the skip arm spends the budget -- it
still never early-returns, so *unsafe_to stays set for the kern machine --
and stop the loops that repeat walks once it clears: apply_backward,
Ligature::apply, the reverse-chain and mark-base/lig subtable applies,
and context_apply_lookup / chain_context_apply_lookup. apply_forward
already checks successful.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [layout] Enforce the budget in the shared subtable loop
dfrg's re-review (harfrust #604): the per-kind entry guards miss paths
that scan before reaching them -- context format 3, the context/ligature
pre-match scans, cursive and mark-to-mark -- so a lookup aliasing thousands
of subtables restarts a buffer scan at each one. Guard the shared
accelerator apply instead (hb_applicable_t::apply), and drop the
now-redundant reverse-chain and mark-base/lig entry guards it subsumes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
cff2_instancing_plan_t::create() builds k identity tuples of k cells
(then keeps a k-float column per surviving solver tuple) for every
entry of the CFF2 ItemVariationStore's VarData offset array, and
nothing bounded k (a 16-bit region count), the number of entries, or
their aliasing of one VarData. The work is therefore quadratic in a
font field, so a 12 KB font cost 2.2 GB.
Charge those cells, before they are allocated, to a per-table budget
HB_BUDGET_SUBSET_CFF2_CELLS (2^24 cells) via the existing
hb_budget_spend helper, at both allocation sites (the k*k identity
tuples and the surviving solver columns). Exhausting it fails create()
like any other planning failure, so the partial instance fails safely
(bounded at ~175 MB worst case) instead of exhausting memory.
Adds test_instance_cff2_vardata_budget: a 6.8 KB font with 40 offsets
aliasing one unused 800-region VarData must fail to partially instance
rather than build a region transform for each; it succeeds (keeping
CFF2 at hundreds of MB) before this change.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-I7KWU2X6.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
hb_depend_context_t::recurse() only refused to re-enter a lookup already
on the call stack, so a chain of distinct Context lookups recursed once
per lookup (stack overflow around 8600 lookups on an 8 MB stack) and
lookups invoking each other in every order were walked once per ordering
(a factorial number of interned paths, gigabytes of memory).
recurse() is where every sibling context (closure, closure-lookups,
apply) enforces HB_MAX_NESTING_LEVEL, so add the same nesting_level_left
guard here; shaping never applies lookups nested deeper, so no edge the
closure can reach is lost (a chain's final edge drops at exactly the
depth the closure stops, 65->66). Bound the number of distinct
recursion paths with a counter kept over the whole table (not cleared by
reset_recurse_cache, which runs per top-level lookup) against a new
HB_MAX_DEPEND_RECURSE_PATHS limit in hb-limits.hh, so it can be tuned
on its own; exceeding it fails the build so
hb_subset_depend_from_face_or_fail() returns nullptr instead of emitting
a truncated graph. The largest whole-table path count among the test
fonts is 744 (NotoNastaliqUrdu-Bold), so 35000 leaves ample headroom.
Builds only with experimental_api. Adds a small permutation-font fuzz
seed and a test-ot-depend test; test-ot-depend, the depend fuzzer and
the depend-closure-parity tests pass. Related to PR #6185.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-N28924W8.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MarkBasePosFormat1 and MarkLigPosFormat1 find a mark's base with a
cache (c->last_base / last_base_until) that is only amortized while
buffer->idx moves forward. A (Chain)Context lookup that recurses into
the mark lookup at input positions out of order resets the cache, so
each application rescans the glyphs from last_base_until to idx from
scratch; on a long run of marks with no base this is quadratic and was
uncharged.
Charge the rescanned span (idx - j) to buffer->max_ops after the
backward search in both lookups, failing the shape when the budget runs
out. A real mark-heavy font pays a negligible amount. Adds a
regression test shaping a Context lookup that recurses to MarkBasePos
at positions 1,0,1,0,...
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-FJEXEDJ4.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Lookups can share subtables, features can share lookups, and each
shaping stage applies its lookups anew, so the number of passes over
the buffer is not bounded by the font size. An earlier lookup can
also grow the buffer far beyond the text the budget was sized from:
a font that grows one character into a long buffer and then walks it
with thousands of lookups (as in harfbuzz/harfrust#571) made hb_shape()
spend CPU out of proportion to the text, and none of it was charged
to buffer->max_ops.
Charge each pass in apply_string() len / 16 for walking the buffer,
even if it applies at no glyph (skipping a glyph is cheap, hence the
division; applying at one is charged only through the buffer
operations and nested lookups it does), and stop the stage loop in
hb_ot_map_t::apply() once the buffer has failed. Nothing is charged
in the matching hot loops (skipping iterator, RuleSet, ChainRuleSet,
LigatureSet): charging there cost real fonts most of the budget in
HarfRust (Gulzar spent about half of max_ops on ordinary Urdu prose,
Noto Sans Duployan ran out on its own sample text). Aliased rules and
ligatures stay bounded by the sanitizer, including the ChainRule array
charge. Matches harfbuzz/harfrust#601.
test_shape_lookup_budget grows one 'a' to 65025 glyphs before applying
aliased lookups, and checks that the pass charge stops the shape.
With the font from harfbuzz/harfrust#571 (one 'A' grown to 30,000
glyphs, then 2,000 lookups), hb_shape() takes about 10 ms instead of
0.6 s, stopping after about 35 passes with the glyphs kept. Gulzar
and Noto Nastaliq Urdu spend under 0.5% of max_ops on Urdu and Persian
prose. Noto Sans Duployan spends up to 45% on its sample text, as on
main: nearly all of it is the _infos_set_glyph_flags() charge from
#5522; the per-pass charge adds about 5%.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-6A9E6OEX.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-K2T2LOJH.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-DVG5XAU0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merge main's active-glyph fast paths and precomputed ClassDef caches
with the dependency and GSUB closure budgets. Charge the fast paths
for the glyphs they visit and the class caches for their ClassDef
scans, preserving both the optimizations and the work limits.
The full Meson suite passes with experimental APIs enabled (300 tests)
and with the default configuration (281 tests). Also syntax-checked
hb-ot-layout.cc with HB_NO_BEYOND_64K.
Assisted-by: Codex
SingleSubstFormat2 already picks, per closure visit, between walking its
whole Coverage and iterating the (usually small) active glyph set. The
Multiple/Alternate/Ligature/ReverseChain closure and depend paths, the
format-1 (Chain)Context closure, and Coverage::intersect_set lacked that
choice, so a nested lookup visited many times over a small active set
re-walked a large Coverage every time, giving quadratic GSUB closure
(and thus subsetting) time.
Add the same get_population()-based fast path to those paths and to
CoverageFormat1/2::intersect_set. Output is byte-identical; only the
per-visit work is bounded.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-KNHDK379.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Rule::collect_second_glyphs() resolved the second input glyph of a
class-based contextual rule by scanning the whole input ClassDef once
per rule, so a RuleSet with many rules sharing one large ClassDef was
quadratic in collect_glyphs() (and thus in subsetting).
Collect the needed classes into a set and walk each ClassDef once for
all of them: add ClassDef::collect_classes(), and have
collect_second_glyphs() accumulate the classes and resolve them in a
single pass. The collected sets are identical; only the work is
reduced.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-5FCXNOQF.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Charge Coverage and ClassDef glyph walks to a work budget
Coverage::subset() expands a format 2 Coverage glyph by glyph, up to
numGlyphs, every time an offset to it is subset, and nothing bounded
that work. A 10-byte range Coverage referenced by 30,000 two-byte
offsets in a subtable listed 8 times costs 1.6e10 iterator steps,
while the sanitizer only bounds work by table bytes; a 60 KB font
keeps hb-subset busy ~50 s. GDEF's MarkGlyphSets reaches the same
walk with no feature or lookup needing to survive planning.
Give hb_subset_context_t a per-table work budget
(HB_BUDGET_SUBSET_TABLE, in the hb_budget_spend() idiom of
hb-limits.hh) with a check_ops() helper that fails the serializer once
exhausted, and charge Coverage::subset() its capped population before
the walk. Charging inside Coverage::subset() covers every caller
(ReverseChainSingleSubst, (Chain)ContextFormat3, GDEF MarkGlyphSets,
...) without touching them. The same unbounded per-reference pattern
in ClassDefFormat1_3 / ClassDefFormat2_4::subset() (the range or
glyphset walk and the glyph_filter count) is charged the same way.
A hostile table that exceeds the budget now fails the subset cleanly
(hb_subset_or_fail returns NULL) in a fraction of a second instead of
succeeding slowly; valid-font output is byte-identical and the
heaviest local table spends a small fraction of the budget.
test_subset_shared_coverage_budget builds the reverse-chaining GSUB in
memory and asserts 256 shared references subset while 2048 fail.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-LLIWV9UK.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Charge layout glyph walks to the existing subset budget
Drop the separate HB_BUDGET_SUBSET_TABLE budget and the second
hb_subset_context_t field it needed; check_ops() now charges
budget_remaining (HB_BUDGET_SUBSET_GLYF, 1 << 28), the per-table
budget glyf instancing already uses. Both are per-table and shared
across buffer-size retries, and no table spends both.
The larger budget still bounds the attack (fails after 2^28 glyph
steps, ~3 s), so test_subset_shared_coverage_budget now expects the
failure at 8192 shared offsets instead of 2048.
Subsetting all glyphs with --layout-features='*' of Gulzar,
NotoSansKR-VF and GoogleSansFlex-VF keeps GSUB/GPOS/GDEF with output
byte-identical to before; the heaviest table (Gulzar GPOS) spends
~98K units, under 0.04% of the budget. Full meson test: 278 OK.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
hb_closure_context_t bounded GSUB glyph closure only by counting lookup
visits (HB_MAX_LOOKUP_VISIT_COUNT per round), and forgets which lookups
are done whenever the glyph set grows, so what one visit walks -- through
aliased offsets, many times the table size -- was charged to nothing. A
LigatureSubst with 9000 ligatureSet offsets aliased onto one 64-component
Ligature, re-walked after every one of 1700 Context triggers that each
add a glyph, makes hb-subset --unicodes='*' of an 86 KB font spend ~110 s
of CPU. A ReverseChain/ChainContextFormat3 with aliased backtrack
Coverages is a second way in.
Give the context an int64_t budget (HB_BUDGET_LAYOUT_CLOSURE records, in
the hb_budget_spend() idiom) spent as each closure() implementation walks
its arrays: dispatch() charges a visit its coverage record count;
ligature sets / ligatures / components, sequences, alternates,
single-subst intersections, rule sets and rules, context Coverages (via a
spend_coverages() helper) and the per-rule-set ClassDef walks are each
charged. Coverage and ClassDef gain a get_record_count() helper because
intersects()/intersect_set() are linear in records, not population. Once
the budget is spent, should_visit_lookup()/lookup_limit_exceeded() stop
the closure early, exactly as they do at the visit limit.
The budget lives in the context, so both public closure entry points (also
reached from hb_ot_shape_glyphs_closure() and the subset plan) are covered
with no API/ABI change. Both attack shapes drop to ~1 s and the control
font's closure output is byte-identical; the heaviest real font in the
corpora uses about 1% of the budget. A hostile GSUB now gets a truncated
closure, as one hitting the visit limit already does.
test_ot_layout_lookups_substitute_closure_budget builds the hostile table
in C and drives hb_ot_layout_lookups_substitute_closure() over it.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-J9K92HA9.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
While extracting GSUB dependencies, find_or_create_glyph_set() turned
every Coverage (format 3 contexts) and class (format 2 contexts) that
a contextual rule mentions into an hb_set_t holding every glyph id the
table lists, whether the face has it or not, with no budget on the
expansion or on the second copy kept in the returned graph. In
experimental builds, where the depend API and 24-bit Coverage formats
live together, a 6.5 KB font with 400 backtrack Coverage format 4
tables each covering [0, 0xFFFFFF] drives a build to ~2 GB.
Clip the cached sets to the face in find_or_create_glyph_set(), the
one place every context position goes through (the first input
Coverage and ReverseChainSingleSubst already clip this way), so the
copies kept in the graph shrink with them. For real fonts this
changes nothing; only broken fonts stop carrying glyph ids past the
face.
Clipping alone is not enough: the temporary is still expanded in full,
and the glyph count itself is cheap to inflate with a MAXP table. So
add a work budget for the whole build in the existing hb_budget_spend()
style: HB_BUDGET_DEPEND, budget_remaining on hb_depend_data_builder_t,
and charge_set(), which spends HB_BUDGET_4 per page while expanding a
temporary and HB_BUDGET_32 per page for every set the graph or context
keeps (new_set(), find_or_create_glyph_set(), find_or_create_recurse_set()).
Running out fails the build and hb_subset_depend_from_face_or_fail()
returns NULL, as it already does on allocation failure. A
get_page_count() accessor is added to the bit-set types so the charge
does not reach into set internals; no public API change.
The 400-position font now peaks at ~11 MB instead of ~2 GB, the 64 KB
/ 4000-position font is refused, and hb-subset-depend output is
unchanged for real fonts. Regression tests build the face with
hb_face_builder_create(): one asserts no glyph above the face count is
kept, the other that a 600-range face is refused.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-856K54QG.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
With edge keys hashed by fasthash32 over all fields, a font can no
longer steer its edges onto one slot, so probing is bounded by the
table's load factor; remove HB_DEPEND_MAX_EDGE_PROBES and its
per-probe charge.
HB_DEPEND_MAX_EDGES at 1<<22 was too close to real fonts: macOS's
DecoType Nastaleeq Urdu builds a graph of 1,198,826 edges, 3.5x below
it. Raise the cap to 1<<25, 28x above that. Other large fonts
measured: Noto Nastaliq 234,098; Hiragino Kaku Gothic 120,851; Gulzar
30,377; NotoSans-merged 29,968; NotoSansKR 7,797.
Reaching the raised cap takes 3.7 s and 1.7 GB, too much for a test,
so test_depend_edge_limit now only checks that a modest fan-out is
recorded in full. test_depend_gsub_high_bit_tags still covers the
hash.
Tested: meson test -C build-exp (experimental API, 296/296 OK).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The dependency-graph edge dedup table in
hb_depend_data_builder_t::add_edge() was indexed by the low bits of
an FNV-style hash whose chain of XORs and odd-constant multiplies
only ever carries upwards, so bit j of the result depended only on
bits 0..j of each field. A font controls feature tags (free 32-bit
values) and substitute glyph IDs, so it can pick values that differ
only in their high bits and pile every edge of a lookup onto one
slot. The open-addressed probe loop then does about N^2/2 work and
each resize_edge_slots() repeats it, so a 15 KB font with 680k
colliding edges spent ~10 min of CPU while an ordinary font with the
same edges takes 0.1 s.
hb_depend_edge_key_t::hash() now runs fasthash32() over all identity
fields, so every bit of every field reaches the index; equality and
the probe sequence are unchanged, so real fonts produce the same
graph at about two probes per edge. The now-unused
hb_depend_edge_t::hash() is removed.
A mixed hash alone is not enough: it is unkeyed, so colliding edges
can still be found offline (each one then costs font bytes instead
of coming free from a product of tags and pairs, but the walk stays
quadratic in them). Every occupied slot stepped over in add_edge()
and resize_edge_slots() is therefore charged to a new per-build
counter edge_probes_remaining, seeded from HB_DEPEND_MAX_EDGE_PROBES
(1<<26, in hb-limits.hh and overridable like the other HB_*_MAX_*
limits, following the HB_RASTER_MAX_DRAW_EDGES precedent); exceeding
it fails the build the way an allocation failure does, so
hb_subset_depend_from_face_or_fail() returns NULL. At about two
probes per edge the cap leaves more than 100x headroom over the
largest test font. Honest fonts are already bounded well below it
by the HB_DEPEND_MAX_EDGES cap added in the previous commit; the
probe budget is what stops a crafted all-colliding chain from doing
quadratic work before that cap is reached.
A regression test builds a GSUB with 2048 features whose tags differ
only in bits 20 and up and checks that all 2048 edges are recorded.
test-ot-depend, the api suite and the depend fuzzing/closure-parity
tests pass on the ASan experimental build, and the finding's
reproducer now builds the hostile font in well under a second.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-4XW1PGV5.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
hb_depend_data_builder_t::add_edge() stored one record per
(source, target, feature tag) with no bound. In GSUB a lookup
reached by F feature tags multiplies its N*M substitutions by F, so
a lookup whose covered glyphs all share one AlternateSet, selected
by many feature tags, produces N*M*F edges from a font that grows
only with N+M+F. A 5 KB font yields 33M edges and ~1.4 GB of
builder memory; COLRv0 overlapping layer slices reach the same sink
without any GSUB. The builder also kept walking after fail(), so an
allocation failure still cost the full N*M*F iterations.
This adds HB_DEPEND_MAX_EDGES (1<<22, about 4M, next to
HB_MAX_GRAPH_EDGE_COUNT in hb-limits.hh and overridable the same
way; 16x the largest graph in the test corpus) and enforces it in
add_edge(), the single choke point every table's extractor stores
through, so hb_subset_depend_from_face_or_fail() returns NULL
instead of building an unbounded graph. add_edge() and
add_gsub_lookup() return early once the builder has failed, and
GSUB_accelerator_t::depend()'s lookup loop stops as well, matching
the existing !successful checks in the context walkers, so a failed
build does bounded work.
A regression test builds the fan-out font in memory with
hb_face_builder (no binary test font needed) and checks that a
16x16x4 fan-out is recorded in full and that the 256x256x512 one is
refused. test-ot-depend and the api suite pass on the ASan
experimental build, and the finding's reproducer now returns NULL
in ~1 s.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-9IOHFFER.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
GSUB_accelerator_t::depend() resolved each feature tag separately: it
ran hb_ot_layout_collect_features() once per tag and, for every tag,
walked the whole FeatureVariations table looking for substitute
features. Tables with many features and many FeatureVariations records
were therefore quadratic, work that sanitize() bounds only for a single
pass.
Query all tags at once, then visit each feature once and each
FeatureVariations substitute once (via new visit_substitutes()
helpers), mapping every lookup to the tag of the feature it belongs to.
The emitted dependency graphs are unchanged.
Adds a regression test building a FeatureVariations GSUB whose features
have no direct lookups.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-5C2QXX5N.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
context_closure_recurse_lookups() and context_depend_recurse_lookups()
already hold the rule's input sequence as a width-erased glyph_data_t
view, but for glyph-based rules (ContextFormat::SimpleContext) they
unwrapped the raw pointer and let intersected_glyph() index it as
HBUINT16. That is wrong for the HBUINT24 arrays of
SequenceContextFormat4 / ChainedSequenceContextFormat4: every input
position after the first was read at the wrong bytes, so the nested
lookup was closed over the wrong glyph and its real output never
entered the closure (other glyph ids keep glyphs that can never
appear). Only experimental / beyond-64k builds are affected.
Resolve the glyph via input[seqIndex - 1] for every format and make
intersected_glyph() a plain insert of the value, mirroring its sibling
intersects_glyph() which already takes the resolved glyph and ignores
data. These two call sites were the only users of the
pointer-plus-index convention; 16-bit formats are unchanged, and there
is no API/ABI change.
Adds a regression test under HB_NO_BEYOND_64K: a SequenceContextFormat4
rule "1 2" invoking a SingleSubst 2->3 must close {1, 2} to {1, 2, 3}.
It fails on the unpatched library and passes with the fix.
Reported by Anthropic's OSS Scanner as ordinary bug ANT-2026-IU58QGNT.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* [subset] Keep hinting for fonts that require it
Some legacy fonts have hinting programs that must be run for correct
rendering; dropping them changes glyph advances and outlines. FreeType
refuses to load these fonts without hinting, so do the same in the
subsetter by ignoring HB_SUBSET_FLAGS_NO_HINTING for them.
Only TrueType-outline fonts qualify. Identify them by family name, and
by the length and checksum of their cvt, fpgm, and prep tables. The
lists and the name-to-ASCII semantics are ported from FreeType's
src/truetype/ttobjs.c; fontations has the same logic in
read-fonts/src/tables/glyf/hint_reliant.rs. Keep the detection in
hb-subset-hinting.hh rather than the subsetter, and add tests for both
paths.
Assisted-by: DeepSeek
* [subset] Refine hint-dependent font detection
- Store family names in a fixed-size table to avoid relocations.
- Read the name table directly instead of building the name
accelerator, and select the record with FreeType's platform and
language priority.
- Use hb_vector_t::set_storage() for the name buffer; treat an
allocation failure as a non-match.
- Pick the family name the way FreeType does: honor the OS/2 WWS bit,
and do not fall through to a lower-priority family name when the
preferred one exists but does not match.
- Add equality operators and use CheckSum::set_for_data() for table
checksums, rejecting unaligned data and handling the trailing bytes.
- Use face->table.glyf and drop the beyond-64K guard.
- Use ISALPHA/TOUPPER for the PDF randomization tag check.
- Add a regression test for the OS/2 WWS family selection.
Assisted-by: DeepSeek
* [perf] Add NO_HINTING subset benchmarks
Add a per-test NO_HINTING flag and enable it for the static TrueType
font benchmarks, so subsetting through the hint-dependent-font
detection path is measured. Benchmark names get a /nohinting suffix.
Assisted-by: DeepSeek
* [subset] Allow unaligned and unpadded checksum data
CalcTableChecksum() can add a trailing partial word. Use that so
_table_checksum() can pass arbitrary table data directly to
CheckSum::set_for_data() instead of rejecting unaligned data and
splitting off the padded prefix.
Add unit coverage for padded, unpadded, and unaligned data.
Assisted-by: DeepSeek
* [subset] Rename and cache the hint-dependent face check
Prefix the entry point with _hb_subset, and cache the answer in face
user data so repeated subsetting of the same face does not repeat the
name and table scans.
Assisted-by: DeepSeek
* [subset] Avoid allocating for the hint-dependent cache
Store the hint_dependent_cache_t value in the face user-data pointer
instead of allocating a bool. UNCACHED is zero, CACHED_FALSE is one,
and CACHED_TRUE is two.
Assisted-by: DeepSeek
* [test] Validate checksums against Python-generated vectors
Generate deterministic buffers and expected OpenType checksums with an
independent Python implementation, covering unaligned offsets and
lengths that are not multiples of four. The generator is committed as
src/generate-checksum-vectors.py and rewrites the marked block in the
unit test.
Assisted-by: DeepSeek
* [test] Drop the checksum vector generator
Keep the Python-generated vectors embedded in test-checksum.cc, but
remove the generator script and its markers. The comment records that
the values were generated independently with Python.
Assisted-by: DeepSeek
* [subset] Use CheckSum::CalcTableChecksum directly
Drop the trivial _table_checksum() wrapper and call
CheckSum::CalcTableChecksum() from the table-ID builder.
Assisted-by: DeepSeek
* [subset] Release the sanitized name table
hb_blob_ptr_t does not destroy its blob in its destructor, so add a
scope guard to release it on every return path.
Assisted-by: DeepSeek
* [subset] Limit cache enum scope
* [subset] Use dagger iteration for hint lists
Replace the family-name and font-ID membership loops with
hb_iter/hb_map/hb_any, and name the family-name capacity used by the
table and its iterator.
Assisted-by: DeepSeek
Drop the separate HB_BUDGET_SUBSET_TABLE budget and the second
hb_subset_context_t field it needed; check_ops() now charges
budget_remaining (HB_BUDGET_SUBSET_GLYF, 1 << 28), the per-table
budget glyf instancing already uses. Both are per-table and shared
across buffer-size retries, and no table spends both.
The larger budget still bounds the attack (fails after 2^28 glyph
steps, ~3 s), so test_subset_shared_coverage_budget now expects the
failure at 8192 shared offsets instead of 2048.
Subsetting all glyphs with --layout-features='*' of Gulzar,
NotoSansKR-VF and GoogleSansFlex-VF keeps GSUB/GPOS/GDEF with output
byte-identical to before; the heaviest table (Gulzar GPOS) spends
~98K units, under 0.04% of the budget. Full meson test: 278 OK.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Coverage::subset() expands a format 2 Coverage glyph by glyph, up to
numGlyphs, every time an offset to it is subset, and nothing bounded
that work. A 10-byte range Coverage referenced by 30,000 two-byte
offsets in a subtable listed 8 times costs 1.6e10 iterator steps,
while the sanitizer only bounds work by table bytes; a 60 KB font
keeps hb-subset busy ~50 s. GDEF's MarkGlyphSets reaches the same
walk with no feature or lookup needing to survive planning.
Give hb_subset_context_t a per-table work budget
(HB_BUDGET_SUBSET_TABLE, in the hb_budget_spend() idiom of
hb-limits.hh) with a check_ops() helper that fails the serializer once
exhausted, and charge Coverage::subset() its capped population before
the walk. Charging inside Coverage::subset() covers every caller
(ReverseChainSingleSubst, (Chain)ContextFormat3, GDEF MarkGlyphSets,
...) without touching them. The same unbounded per-reference pattern
in ClassDefFormat1_3 / ClassDefFormat2_4::subset() (the range or
glyphset walk and the glyph_filter count) is charged the same way.
A hostile table that exceeds the budget now fails the subset cleanly
(hb_subset_or_fail returns NULL) in a fraction of a second instead of
succeeding slowly; valid-font output is byte-identical and the
heaviest local table spends a small fraction of the budget.
test_subset_shared_coverage_budget builds the reverse-chaining GSUB in
memory and asserts 256 shared references subset while 2048 fail.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-LLIWV9UK.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
#6521 made hb_raster_draw_render() skip the scanline sweep when the
estimated pixel work exceeds the budget, but it still returned a
cleared image, so callers could not tell a budget failure from an
outline that legitimately covers nothing.
Check the budget before allocating the image and return NULL, the
same way render already reports allocation and configuration
failures. The accumulated edges are still cleared by the existing
scope guard. All in-tree callers (raster paint masks, hb-raster
utilities) already handle a NULL result.
test_budget_sweep now expects NULL for the over-budget case. Full
meson test passes.
Follow-up to #6521
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#6499 added caches for the vertical advance and origin, but also
changed what the origin is: its x became half the horizontal
advance (as hb-ot does) instead of FreeType's
horiBearingX - vertBearingX, and a glyph FreeType fails to load got
a cached origin of 0 instead of the callback returning false. That
changed shaping output for fonts whose vertical bearings are not
centered (e.g. bitmap-only fonts), and stopped the fallback to
h_origins on load failure.
Keep the caches, but cache both the x and y FreeType computes, the
same way as before #6499, and return false on load failure without
caching anything. Values that don't fit the unsigned cache after
sign normalization are simply not cached.
Testing: meson test -C build passes. test-ft now checks the
batched origins against FreeType's bearings and that a failed load
returns false on repeat queries. hb-shape --font-funcs=ft
--direction=ttb output over NotoSansKR plus 80 shape/api test fonts,
at several sizes, a variation, and negative scale (810 runs) is
identical to the commit before #6499; on main 30 of those differed
(aat-feat, chromacheck-cbdt, chromacheck-sbix).
Follow-up to #6499
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#6500 made the CBDT accelerator build a per-face (ppem, glyph range)
-> record index at load time, with a strike set/map, a sorted vector,
a walk_ppems fallback and an HB_CBDT_MAX_INDEX_ENTRIES cap, to avoid
walking a huge strike array on every glyph query. Real fonts have a
handful of strikes, so that machinery costs every CBDT face memory and
load time for the benefit of pathological fonts only.
Remove it and restore the simple CBLC::choose_strike() and
BitmapSizeTable::find_table() lookup path, but only consider the first
HB_CBLC_MAX_STRIKES (256) strikes when choosing one; later ones are
ignored. ppem is 8-bit, so this never drops a distinct size a real
font could need. The sbix strike index, the CBDT subsetter's record
map and its image-copy budget from #6500 are kept unchanged.
Adds test_hb_ot_color_png_cbdt_max_strikes, which places the only
usable strike at index 255 / 256 and checks it is found / ignored; it
fails with the cap removed. Full meson test passes.
Follow-up to #6500
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
cff2_subset_plan::create() cast acc.fdSelect (a CFF2FDSelect) to the
CFF1 FDSelect type for hb_plan_subset_cff_fdselect(), whose
get_fd_range() dispatch knows only formats 0 and 3. A CFF2-only
format-4 FDSelect therefore reported FD 0 for every glyph, so the
planner kept only Font DICT 0: every retained glyph belonging to
another Font DICT lost that DICT's Private DICT and local subrs, and
when it called a local subr it either lost its outline or drew through
FD 0's subrs with the wrong shape. The call still succeeded, writing a
single-Font-DICT font. Format 4 is the only FDSelect format with a
16-bit FD index, so every CFF2 font with more than 256 Font DICTs is
affected.
The planner only needs get_fd_range() and format, so give CFF2FDSelect
its own get_fd_range() dispatching formats 0, 3 and 4, make
hb_plan_subset_cff_fdselect a template over the FDSelect type with
explicit instantiations for CFF1FDSelect and CFF2FDSelect (same pattern
as other explicitly instantiated subset helpers), and drop the cast.
The existing format-4 writer in hb_serialize_cff_fdselect is now
reachable. Because a CFF2 font with >256 retained FDs can now
legitimately plan format 4, serialize_cff2_to_cff1() rejects that case
(CFF1 has no format 4) instead of emitting an invalid CFF table; with
fewer FDs the downgrade is unchanged.
Adds test_subset_cff2_fdselect_format4, which subsets the format-3 and
format-4 variants of a two-Font-DICT font and requires identical CFF2
output and glyph 2's extents; it fails before this change.
Reported by Anthropic's OSS Scanner as ordinary bug ANT-2026-7RSDR0F0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cff1_subset_plan renumbers every custom SID it emits through sidmap
(charset via plan_subset_charset, Top DICT names via reassignSIDs, FD
names via sidmap[]), and the output String INDEX is serialized from
that map. plan_subset_encoding was the one emitter that pushed the
source font's SID unchanged into subset_enc_supp_codes, so once an
earlier custom-named glyph was dropped a supplement code named the
wrong glyph, or a string past the end of the renumbered String INDEX.
Look the SID up in sidmap where the supplement entry is recorded. The
source SID is still used for get_supplement_codes(), which searches the
source Encoding. This is safe on every path that reaches it:
subset_encoding implies a non-CID font, so plan_subset_charset has
already added each visited glyph's name (sidmap.add is idempotent), and
when the charset is predefined and plan_subset_charset is skipped all
glyph SIDs are standard strings, which sidmap[] returns unchanged.
Adds test_subset_cff1_encoding_supplement with a generated CFF font
(custom names, format-0 Encoding with supplement 98 -> 'second') and
its expected B,C subset; the test fails before this change.
Reported by Anthropic's OSS Scanner as ordinary bug ANT-2026-JORVXUQZ.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
closure_glyphs(), depend() and varc_subset_plan_t::collect_glyphs()
each decode the full glyph record of every glyph id they reach,
assuming every glyph has its own record. Nothing enforces that:
CFFIndex::sanitize() accepts overlapping glyphRecords offsets, and a
Coverage table may map many glyph ids to the same INDEX entry, so N
glyph ids can resolve to one large record and the walk does N times
that record's work -- quadratic in the font size (44 s for a 245 KB
font), charged to no budget.
In a well-formed table the records are disjoint and each glyph is
visited once, so the bytes of all records walked cannot exceed the
glyphRecords INDEX. Enforce exactly that invariant where the records
are fetched, in all three walks, failing through each site's existing
idiom. The remaining work is linear: a component decodes only if its
record holds at least one byte per 64 counted axis values.
In collect_glyphs() this also replaces the saturating sum and the
post-loop data_size != UINT_MAX test, which a 64-bit build bypasses
because hb_unsigned_add_saturate() returns size_t and the sum truncates
into the unsigned data_size (a near-4 GiB total stored as a small
value); the running total is now checked after every addition.
Overlapping VARC records, already malformed per the CFF INDEX spec, now
fail subsetting instead of being accepted; shaping and drawing are
unaffected and valid VARC fonts subset to byte-identical output.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-71PGEDL2.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The subsetter cannot rewrite tables that refer to glyphs. That includes
legacy kern, AAT (ankr, bsln, just, kern, kerx, lcar, mort, morx, opbd,
prop), Graphite (Glat, Gloc, Silf), JSTF, SVG, LTSH, EBDT/EBLC, and
Apple's Zapf, bdat/bloc, gcid, and hvgl/hvpm tables. When one of these
is retained it is passed through unchanged, so renumbering glyphs makes
its glyph ids stale. Implicitly enable retain-gids in that case.
Make the plan use its own flags after this adjustment so every consumer
sees the effective retain-gids state, and add regression tests covering
AAT, Graphite, and the other unsubsettable glyph tables.
Assisted-by: DeepSeek
hb-cairo hands the actual rendering to cairo, so the work done per
paint operation (clip, gradient, compositing) is cairo's, not ours.
Any count we keep is a guess. The guesses we had were not good:
022a9e475 charged each paint
(1 + clip + path segments) x max(weight, segments), with a weight of
up to 8192, against HB_BUDGET_GLYPH. Real fonts blow through that.
For example, hb-view renders the Spain flag in Noto-COLRv1 (U+1F1EA
U+1F1F8) as a blank image, and about ten glyphs in that font go over.
Remove the budget accounting from hb-cairo:
- Revert 022a9e475 "[cairo] Charge compositing operations to the
glyph budget", with its test font and generator.
- Revert 8f3ab5ab8 "[cairo] Share work across nested glyph
rendering": the per-scaled-font hb_cairo_budget_t, the segment
charges in the draw callbacks, and the set/get budget callbacks
on the cairo draw and paint funcs.
Keep paint_color_glyph declining (fd8c9209b), so COLR paints nested
glyphs inline under the decycler and nesting limits of the walk in
progress. Going through cairo_show_glyphs() again would bring back
the PaintColrGlyph cycle crash, which test_hb_cairo_color_glyph_cycle
covers. Only the code comment loses its mention of work limits.
Testing: meson test -C build: 299 OK, 1 skipped. The Spain flag
renders again, and the PaintColrGlyph cycle glyphs in
test_glyphs-glyf_colr_1.ttf render without crashing.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
CPAL::subset() copies the retained colors once per distinct
colorRecordIndices start and later writes distinct-blocks x
retained-colors into the 16-bit numColorRecords, but nothing bounded
that product. CPAL palettes may overlap in the color-record array, so
a well-formed 280 KB font with 20000 palettes of 20000 colors made a
single hb_subset_or_fail() hold ~1.5 GiB of copies and burn ~29 s of
CPU (rebuilt on each out-of-room retry) before failing; smaller
products silently wrapped numColorRecords and the per-palette start
indices in the emitted table. This is a regression from 95ba91721
(#6285, "Share color record blocks between identical palettes").
Fail before copying, at the top of the per-start loop, once one more
block of retained colors could no longer be counted in the 16-bit
numColorRecords, via check_assign() with HB_SERIALIZE_ERROR_INT_
OVERFLOW. This caps the copy-and-hash work, the heap and the
serialized size at 65535 records, and makes the two 16-bit
assignments provably safe, which resolves and removes the
"limit total final size" TODO in serialize(). Palettes that do not
overlap can never reach the limit, since their distinct starts times
numColors already fit in the source numColorRecords.
Trade-off: the bound counts distinct starts, not distinct dedup
blocks, so a font whose many overlapping palettes would all collapse
into a few shared blocks is also rejected past the limit. Pre-#6285
code wrapped on those inputs anyway.
Adds test_subset_cpal_overlapping_palettes asserting the 218-vs-219
palette boundary (65400 records accepted, next rejected).
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-DBBMME51.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* [font] Saturate synthetic bold and glyph origin arithmetic
#6154 made hb_position_t arithmetic saturate and #6296 did the same
for synthetic_glyph_extents(), but the rest of hb_font_t still used
plain signed arithmetic:
- changed() computed the embolden strength with abs (x_scale), which
overflows for INT32_MIN, and converted the float product to int
without a range check. A strength of INT32_MIN could also not be
negated later on, for negative scales.
- get_font_[hv]_extents(), the advance, origin and contour point
getters added the strength with += and could overflow.
- add_offset(), subtract_offset() and apply_offset() overflow when
they apply a large origin, and so do the origin fallbacks in
apply_glyph_[hv]_origins_with_fallback().
- em_multf() converted its float result to hb_position_t without a
range check, which is undefined for a value such as an ascender
above the em at a scale close to INT32_MAX.
Use hb_saturate_add(), hb_saturate_sub() and hb_clamp_to() in these
places, and keep the strength away from INT32_MIN so that negating it
stays defined. Results that fit do not change.
The shift variables that were int are now hb_position_t, since the
saturating helpers deduce one type and int32_t is not int everywhere.
Tested with test-font under UBSan: the new tests fail on the base
commit at hb-font.hh:365, :848, :1154 and :1167, and pass here.
Assisted-by: Claude
* [test] Cover plural origins and synthetic slant saturation
Review feedback: the plural origin getters and the slant
clamp in get_glyph_contour_point were not covered by the
tests, so removing their saturation would still pass.
test_synthetic_bold_saturates now also calls
hb_font_get_glyph_h_origins and hb_font_get_glyph_v_origins
once each and checks both coordinates for INT32_MAX. A new
test_synthetic_slant_saturates sets a slant of 1 on a contour
point one short of INT32_MAX in both axes and checks that x
saturates to INT32_MAX.
Verified by mutation: replacing the plural hb_saturate_add
calls with plain addition makes the first test fail (the
value wraps), and dropping the hb_clamp_to in the slant path
makes the second fail (x comes out as -2). Full
meson test -C build: 270 ok, 0 fail, 7 skipped
(subset+repack, missing repacker dependency; same as the
baseline run for this PR).
Assisted-by: Claude
* [font] Apply the synthetic slant shift before clamping
get_glyph_contour_point clamped the slant shift to
hb_position_t before adding it to x. When x and the shift
have opposite signs and partly cancel, clamping first
changes the result: a contour point at x = -2000000000,
y = 2000000000 with a slant of 2 should land at
x = 2000000000, but came out as 147483647, because the
shift was clamped to INT32_MAX before the addition.
Add the shift first and clamp the sum instead. The
existing saturation case is unchanged: hb_clamp_to
widens to double to do the clamp, so an out-of-range sum
still saturates instead of hitting an undefined
float-to-int conversion.
The new test_synthetic_slant_cancelling covers the
cancelling case. Verified by mutation: restoring the
clamp-first order makes it fail with 147483647, the
reported value. Full meson test -C build: 270 ok,
0 fail, 7 skipped, same as baseline.
Assisted-by: Claude
cmap_accelerator_t::closure_glyphs() in the DMAP path expanded, for
every retained variation selector, all of the selector's Default UVS
ranges and resolved each code point with get_variation_glyph(). Any
number of 11-byte selector records may share one Default UVS table and
a 4-byte range covers 256 code points, so the work was bounded by
selectors x covered code points rather than by the table size: a 223 KB
font with 20000 selectors on one 168-range table drove ~860M lookups
and ~49 s of CPU in hb_subset_plan_create_or_fail().
A default mapping can only ever resolve to the nominal glyph, which the
subset plan already retains with the unicode itself, so the closure only
needs the code points that carry a Non-Default UVS mapping in cmap or
DMAP; get_variation_glyph() still decides which table wins for each.
This is exactly what the non-DMAP path (NonDefaultUVS::closure_glyphs)
already does. A collect_variation_non_default_unicodes() accessor is
added alongside the existing collect_variation_unicodes() chain
(VariationSelectorRecord, CmapSubtableFormat14_15, CmapSubtable,
cmap::accelerator_t) and used from the DMAP closure, so the work is one
lookup per 5-byte mapping visited.
No public API or ABI change; hb_face_collect_variation_unicodes() keeps
expanding Default UVS ranges as before, and fonts without DMAP take the
unchanged first branch. The reproducer's plan drops from ~49 s to 0.01 s
with the same 257-glyph result. The separate, older per-record
DefaultUVS::copy() serialization cost (~5.2 s for this font, independent
of DMAP) is left in place.
Adds test_subset_DMAP_shared_default_uvs to test/api/test-subset-cmap.c.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-8RP452C0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
When rewriting cmap, the subsetter built one hb_set_t of covered code
points per encoding record. SubtableUnicodesCache keyed those sets by
the record's own offset and collected each with num_glyphs = UINT_MAX,
so a font with many records sharing one full-range format 12 subtable
built one ~1.1M-code-point set per record: ~41 s of CPU, or ~2.1 GB of
memory for a 96 KB input, with nothing but the record count (bounded
only by the sanitizer) limiting the work.
Three changes bound it, all matching how nearby cmap code already works:
1. Key cached_unicodes by record->subtable, so records sharing a
subtable share one set (both set_for overloads).
2. Give the cache a num_glyphs member and collect with it instead of
UINT_MAX, as the cmap accelerator already does for
hb_face_collect_unicodes; cmap::serialize fills it from
plan->source->get_num_glyphs() and create_filled_cache() takes it
as a parameter, passed as face->get_num_glyphs() for
hb_subset_preprocess.
3. Process only the first encoding record of each (platformID,
encodingID) pair in both cmap::serialize and create_filled_cache
(variation-selector records exempt). Lookup already sees one record
per pair and EncodingRecord::copy links later records of a format to
the first serialized subtable, so this bounds both the number of
sets built and the _can_drop walk regardless of how many records a
malformed font repeats, independent of whether they share a subtable.
No public API or ABI change. Output is unchanged for well-formed fonts:
the plan's unicodes are already restricted to gids below num_glyphs, and
valid fonts never repeat a (platformID, encodingID) pair. The attached
reproducer's hostile fonts drop from ~41 s / 2.1 GB to ~0 s / 12 MiB
with byte-identical output.
Adds test_subset_cmap_shared_subtable_many_records and
test_subset_cmap_repeated_encoding_records to
test/api/test-subset-cmap.c.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-FSXRDPBT.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
CmapSubtableFormat4::accelerator_t::collect_unicodes() and
collect_mapping() trusted the format's invariant that segments are
sorted and disjoint, so a subtable whose 32766 segments all cover
U+0000..U+FFFE was walked once per segment: ~2.1 billion insertions
(about 30 s of CPU) for a single hb_subset() or
hb_face_collect_nominal_glyph_mapping() call on a 262 KB font.
Both walkers now keep a cursor `next` past the last collected code
point and clamp each segment's start to it, skipping segments that end
before the cursor. Total work per subtable is therefore at most 65536
code points plus one step per segment, mirroring the ordering guard
CmapSubtableLongSegmented::collect_mapping() already has for formats
12/13. The glyphIdArray index is still computed from the raw
startCount[i], so range-offset segments map exactly as before, and
get_glyph() is untouched. Behaviour changes only for malformed
subtables: where sorted segments overlap the earlier segment now wins
the shared code points, and a segment lying wholly inside an earlier
one is ignored.
Adds test_collect_unicodes_format4_overlapping_segments to
test/api/test-collect-unicodes.c.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-GV7L48CF.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The subset plan turned every variation index collected by the COLR,
GDEF and BASE closures into a row to keep, checking only the outer
index against the subtable count and never the inner index against
VarData::itemCount. A font could therefore name tens of thousands of
rows in a VarData whose itemCount is 2, and partial instancing then
sized its work from that inflated row count, allocating
regionIndexCount x rows x 5 bytes (3.2 GB from a 240 KB COLRv1 font).
generate_varstore_inner_maps() now takes the store and keeps only rows
that ItemVariationStore::has_delta_set() confirms exist, and
remap_variation_indices() maps such an index to
HB_OT_LAYOUT_NO_VARIATIONS_INDEX without consuming a new row. That
preserves the runtime meaning of an out-of-range index (delta 0) and
keeps the two functions in step; all consumers already handle the
sentinel. This covers the COLR, GDEF and BASE callers. Well-formed
fonts are byte-identical.
The HVAR/VVAR route noted in the report (its planner builds row maps
itself, one per retained glyph) is left for a follow-up: folding the
missing rows into a shared zero row changes the output form of a
map-less HVAR whose glyph count exceeds the store's itemCount, which
is better landed on its own.
Adds an API regression test (test_subset_colr_phantom_rows) that fails
on the unpatched tree.
Discovered by Anthropic's OSS Scanner, as vulnerability ANT-2026-0TLE41I0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Anchor coordinates are scaled as float, and apply_cursive_attachment()
and the mark attachment code converted them to hb_position_t with a
plain cast or an implicit conversion. A value out of range for the
target type makes that conversion undefined; on x86 it gives INT32_MIN.
#6154 made the additions around these conversions saturate, but not
the conversions themselves.
With NotoNastaliqUrdu-Regular.ttf at a scale of INT32_MAX, a cursive
anchor scales to 2.23e9, and the glyph ended up with an x_advance of
-2147483648.
Convert with hb_clamp_to(), and negate the cursive cross-direction
offsets with hb_saturate_neg(), since they can now be INT32_MIN.
Results that fit do not change.
Tested with test-shape under UBSan: the new test fails on the base
commit at CursivePosFormat1.hh:141 and passes here. No test reaches
the two conversions in MarkArray.hh; they are changed to match.
Assisted-by: Claude
CmapSubtableLongSegmented::collect_unicodes() and collect_mapping()
clamp the end of a group so that its glyph ids stay below num_glyphs.
That only makes sense for format 12, where the glyph id increases
across the group. In format 13 every code point of a group maps to the
same glyph, so the clamp cut each group down to a couple of code
points (Adobe Blank 2: 38 code points collected instead of 1,111,998).
Only clamp when the glyph id increases, and fix the clamp itself: it
kept one code point too many, mapping it to glyph num_glyphs.
Fixes https://github.com/harfbuzz/harfbuzz/issues/6591
Assisted-by: Claude
* [test] Expand CFF number font parsing to subsetter encoding roundtrip
* [subset] Support longintdict encoding
It was a TODO for the encoder and this implements it.
* [test] Move roundtrip parse encode to its own function
* [cff] Extract nibble_buffer_t to avoid heap allocated hb_vector_t
* [cff] Use '(size + 1) / 2' buffer size instead of asserting evenness
Address review
* [cff] Minor touches
* [cff] Use idempotent ensure_error instead of set_error
As review
* [subset] Reuse sanitized tables cached on the face
Use the face table list and SFINAE to borrow sanitized tables from
the source face. Add borrow_sanitized_table() to suitable accelerators
and reuse the COLR accelerator during closure. Keep fallback tables
in the subset cache without taking references for each access.
Cache full GSUB/GPOS sanitization separately from shaping's lazy
sanitization. Inline their no-op blocklist hooks so the subset library
can instantiate the accelerators.
Add regressions for repeated subsets and GPOS data that passes lazy
sanitization but fails full validation.
Tested with meson test -C build: 299 passed, 1 skipped. Also ran
reduced-feature syntax checks and ASan/UBSan with failing malloc.
Assisted-by: Codex
* [ot-layout] Remove unused GSUB/GPOS blocklist hooks
Both hooks always return false. Remove the hooks and their dead
conditional from accelerator initialization.
Tested with meson test -C build: 299 passed, 1 skipped.
Assisted-by: Codex
* [subset] Borrow GDEF before shaping's blocklist
Keep ownership of the unfiltered sanitized GDEF table in the face
accelerator, alongside the potentially blocklisted shaping view.
Subsetting can borrow it without repeating sanitization or taking
references for each access.
Allow static borrowing accessors to initialize accelerators inside
libharfbuzz. GDEF uses this to keep its hidden blocklist implementation
out of the subset library's initialization path. Other accelerators
retain their instance accessors.
Add a synthetic blocklisted-font regression checking that shaping
still ignores GDEF while repeated subsets preserve its glyph classes
without rereading it.
Tested with meson test -C build: 299 passed, 1 skipped. Also ran GCC
and reduced-feature syntax checks, plus ASan/UBSan with failing malloc.
Assisted-by: Codex
hb_utf_offset_to_pointer() walked `start` using `while (offset-- > 0)`
and `while (offset++ < 0)` without checking whether `start` had
reached `text + text_len` or `text`. In addition, post-decrementing
`offset` in the first loop's condition left `offset == -1` after any
non-negative walk, causing the backward loop to execute one extra
`utf_t::prev()` step. In hb_cairo_glyphs_from_buffer(), cluster
deltas in the `utf8_clusters` path could also advance `start` past
`utf8 + utf8_len`, or overflow `signed` when cast for
hb_utf_offset_to_pointer().
Check `start < text + text_len` and `start > text` in
hb_utf_offset_to_pointer() while adjusting `offset` inside the loop
bodies, and clamp each cluster delta in hb_cairo_glyphs_from_buffer()
to the remaining bytes in `utf8`.
Extend test_hb_cairo_glyphs_from_buffer_clusters in
test/api/test-cairo.c to cover multi-byte UTF-8 inputs on exact-sized
heap buffers and cluster deltas that exceed the string length or
overflow signed 32-bit integers.
Co-authored-by: Gemini <noreply@google.com>
* [buffer] Add deserializers with byte offsets
Buffer deserializers return a pointer into caller-owned input. GIR
describes it as an owned UTF-8 string, so bindings can free an
interior pointer or read beyond explicitly sized input.
Add _with_offset variants with bounded byte-array inputs and optional
byte counts. Preserve the existing return values and partial buffer
contents, including on failure. Cross-check both variants against the
existing deserializers in the buffer API tests.
Testing: full Meson suite, buffer tests under Valgrind, and GIR and
PyGObject checks.
Fixes: #6446
Assisted-by: OpenAI Codex
* [buffer] Rename new deserializers to use 2 suffix
Use the established 2 suffix for the new buffer deserializers,
following hb_shape_plan_create2(). Update API docs, NEWS, and tests.
Testing: full Meson suite, generated GIR, and PyGObject checks.
Assisted-by: OpenAI Codex
Slice the blob at dataZ, then slice the remaining bytes at the glyph
start offset. Chaining sub_array avoids a wrapping unsigned addition
and lets each slice clamp against the remaining length.
Keep test_subset_gvar_valid_data_offsets_control and
test_subset_gvar_data_offset_overflow_rejected unchanged.
Assisted-by: an AI coding assistant