fix(insert-sync): track explicit local address provenance#948
fix(insert-sync): track explicit local address provenance#948tonibohnlein wants to merge 5 commits into
Conversation
There was a problem hiding this comment.
Code Review
This pull request introduces the AddressProvenance enum to distinguish between root-relative, known absolute, and unknown/dynamic absolute local memory addresses, improving the precision of synchronization analysis. It also adds overflow checks when calculating absolute byte offsets and ranges, conservatively treating overflowing or dynamic addresses as potentially aliasing. The reviewer identified an issue in UpdateAliasBufferInfo where only the first element of baseAddresses is updated with the offset, which could lead to incorrect tracking for multi-buffer allocations, and suggested updating all elements in the vector.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
|
Addressed the remaining correctness review in
Added compiler regressions for dynamic reinterpret, contiguous-subview-plus-reinterpret, segmented and mixed back-edges, safe dynamic-subview disjointness, chained dynamic subviews, and strided dynamic subviews. Local verification (maximum two workers): affected Werror build, 11 focused lit files, all 21 |
4817de8 to
b083a9c
Compare
|
/run a5 |
|
已接收
页面会自动刷新,可以直接看当前阶段、排队情况和最近结果。 |
b083a9c to
2295d73
Compare
|
/run a3 |
|
已接收
页面会自动刷新,可以直接看当前阶段、排队情况和最近结果。 |
A3 板测失败
失败用例
|
A3 板测失败详情:PR #948comm_p2p_binding_variants
comm_p2p
|
|
/run a3 |
|
已接收
页面会自动刷新,可以直接看当前阶段、排队情况和最近结果。 |
A3 板测失败
失败用例
|
A3 板测失败详情:PR #948comm_p2p_binding_variants
comm_p2p
|
A5 板测失败
失败用例
|
A5 板测失败详情:PR #948xors
xor
vectorAddition
vadd_validshape_hyper
vadd_pto_ir
ttri
trap
trans
tquant_mx
tprefetch_async_binding
tprefetch
tpows
tpow
tinsert
tinsert_fp
tileSetGetValue
syncall_binding
tmatmulk_autosync_a5
test_tmov_row_major_1x16_control_a5
test_tmov_col_major_16x1_align_a5
test_set_wait_unified_api
test_mem_inject_sync_basic
test_intercore_sync_a5
test_intercore_sync_a5_ptoisa_vec
test_intercore_sync_a5_functional
test_intercore_sync_a5_dyn
test_inject_sync_two_event_id
test_inject_sync_loop
test_inject_sync_loop_nest
test_inject_sync_intra_pipe_barrier
test_inject_sync_if
test_inject_sync_if_else
|
A5 板测失败详情:PR #948test_dynamic_valid_shape
test_barrier_sync
test_auto_sync_tail_hint
test_a5_buf_sync
sync
syncHigh
rmsnorm_incore_0
rar_optimization_test
nested_loop_confliect
matmul
decode_projection_incore_0
compensation_test
add_double_dynamic
vadd_pto_pingpong
subview_tsubs
subview
subview_boxed_dynamic
subsc
subs
subc
sub
sqrt
sort32
shrs
shr
shls
shl
set_validshape
sels
sel
scatter
scalar_ptr
rsqrt
|
A5 板测失败详情:PR #948rowsum
rowprod
rowmin
rowmax
rowexpandsub
rowexpandmul
rowexpandmin
rowexpandmax
rowexpandexpdif
rowexpanddiv
rowexpandadd
rowexpand
reshape
bitcast_inplace_cvt
bitcast_dtype_alias
rems
rem
relu
recip
rope_kv_cache
rmsnorm
qwen3_decode_incore_7
qwen3_decode_incore_6
qwen3_decode_incore_5
qwen3_decode_incore_4
qwen3_decode_incore_2
qwen3_decode_incore_1
qwen3_decode_incore_12
qwen3_decode_incore_11
qwen3_decode_incore_10
post_rmsnorm
out_proj_residual
down_proj_residual
|
A5 板测失败详情:PR #948quant
quant_asym
vector_example_dag_kernel_mul
vector_example_dag_kernel_add_scalar
vector_example_dag_kernel_add
paged_attention_example_kernel_softmax_prepare
paged_attention_example_kernel_qk_matmul
paged_attention_example_kernel_pv_matmul
paged_attention_example_kernel_online_update
paged_attention_example_kernel_init_inplace
orchestration_example_kernel_mul
orchestration_example_kernel_add_scalar
orchestration_example_kernel_add
print_scalar
prelu
plan_memory_reuse_sequential
plan_memory_peak_exact_capacity
plan_memory_peak_8_overlapping
plan_memory_no_reuse_overlap
plan_memory_nested_loops
plan_memory_loop_no_reuse_outer_live
plan_memory_loop_in_if
plan_memory_if_yield
plan_memory_if_in_loop
plan_memory_fragmentation_two_holes
plan_memory_fragmentation_hole_fit
plan_memory_for_iter_args_yield
plan_memory_bind_tile_alias_liveness
partmul
partmin
partmax
partition_view_verify_valid
|
A5 板测失败详情:PR #948partition_view_verify_rank_mismatch_valid
partition5d_dynamic_a5
partition5d_a5
partarg
partadd
ors
or
not
neg
muls
|
30d26d7 to
99c64ec
Compare
Summary
AddressListKindauthoritative for back-edge event selection and affine slot pruning; mixed or incompatible hazards use one static eventTesting
b083a9c9is green: main build-and-test plus x86_64 and aarch64 wheel/build checks passed; remote NPU validation was skippedPTOAS_ENABLE_WERROR=ON; a localptoaslink also succeeds with temporary narrowed pass registration because this checkout's minimal LLVM 21 configuration has unrelated all-pass registration link failures-j2multi_tile_lit tests: passed with-j2-j2