[GSan] Partially support TMA & cp.async ops#9699
Merged
peterbell10 merged 2 commits intoMar 20, 2026
Merged
Conversation
This was referenced Mar 12, 2026
Merged
peterbell10
changed the title
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
bcb2ec6 to
e8cf018
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
e567d06 to
09c2e83
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
e8cf018 to
372fe1e
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
09c2e83 to
eae198e
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
372fe1e to
7138e28
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
eae198e to
c5c7aad
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
7138e28 to
d93ee69
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
c5c7aad to
0734117
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
d93ee69 to
9239245
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
0734117 to
3570ff8
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
9239245 to
cac3d76
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
3570ff8 to
b606ba9
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
branch
from
cac3d76 to
cf31648
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
b606ba9 to
90696f9
Compare
peterbell10
changed the base branch from
pb/pr-chain/gsan_add_symmetric_memory_api_a42c
to
main
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
90696f9 to
870529e
Compare
peterbell10
changed the base branch from
main
to
pb/pr-chain/gsan_instrument_tl_load_store_cacf
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
3 times, most recently
from
ea94e1e to
0e0b801
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_instrument_tl_load_store_cacf
branch
from
88e7366 to
c7cd539
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
0e0b801 to
2db13d1
Compare
pawelszczerbuk
approved these changes
Mar 19, 2026
peterbell10
force-pushed
the
pb/pr-chain/gsan_instrument_tl_load_store_cacf
branch
from
c7cd539 to
a2fad55
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
2db13d1 to
52485af
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_instrument_tl_load_store_cacf
branch
from
a2fad55 to
8ebfa99
Compare
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
52485af to
7a710ac
Compare
Base automatically changed from
pb/pr-chain/gsan_instrument_tl_load_store_cacf
to
main
March 19, 2026 22:14
This handles TMA ops by decoding the TMA descriptor, and converting the calls into the equivalent load/store calls which we pass to the load/store instrumentation. This is incomplete because it doesn't model the asynchrony, but is a reasonable starting point. git-pr-chain: gsan_partially_support_tma_ops_d1c1
peterbell10
force-pushed
the
pb/pr-chain/gsan_partially_support_tma_ops_d1c1
branch
from
7a710ac to
3b193b3
Compare
raymondtay
pushed a commit
to raymondtay/triton
that referenced
this pull request
Mar 22, 2026
This handles TMA ops by decoding the TMA descriptor, and converting the
calls into the equivalent load/store calls which we pass to the
load/store instrumentation.
This is incomplete because it doesn't model the asynchrony, but is a
reasonable starting point.
jvican
pushed a commit
to jvican/triton
that referenced
this pull request
Mar 27, 2026
This handles TMA ops by decoding the TMA descriptor, and converting the
calls into the equivalent load/store calls which we pass to the
load/store instrumentation.
This is incomplete because it doesn't model the asynchrony, but is a
reasonable starting point.
plognjen
pushed a commit
to plognjen/triton
that referenced
this pull request
Apr 14, 2026
This handles TMA ops by decoding the TMA descriptor, and converting the
calls into the equivalent load/store calls which we pass to the
load/store instrumentation.
This is incomplete because it doesn't model the asynchrony, but is a
reasonable starting point.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Commits in this PR
[GSan] Partially support TMA ops
This handles TMA ops by decoding the TMA descriptor, and converting the
calls into the equivalent load/store calls which we pass to the
load/store instrumentation.
This is incomplete because it doesn't model the asynchrony, but is a
reasonable starting point.
Also partially support async_copy
PR chain