Skip to content

Validate shared memory name conflict repair on master - #80

Closed
Spicy-cream wants to merge 17 commits into
LinQuickDev:masterfrom
Spicy-cream:review/pr3535-shm-conflict-oct18
Closed

Spicy-cream wants to merge 17 commits into
LinQuickDev:masterfrom
Spicy-cream:review/pr3535-shm-conflict-oct18

Conversation

@Spicy-cream

Copy link
Copy Markdown

CI validation only

Validate the identical source commit 3da2dde used by #79.

This draft targets master solely to trigger the full Linux and macOS workflows. Do not merge it; integrate the repair through pull request 79 into transport-handshake using a merge commit to preserve upstream ancestry.

The repair relocates upstream shared-memory name termination into the common decoder while retaining validation of the original wire bytes, so unterminated names remain rejected. Upstream test changes are preserved; a valid maximum-length name test is added.

Local validation: 43 UBRing tests, 28 common-handshake tests, and 92 RDMA tests passed. Three name-boundary tests also passed 100 shuffled repetitions. git diff --check passed.

Spicy-cream and others added 17 commits October 8, 2026 23:58
…lity

Resolve UBSHM handshake conflicts using the tested rebased tree. Retain backend-specific format negotiation, remote shared-memory lengths, and upstream IPC V2 tests.
…er-oct15

Resolve current master conflicts and preserve UBSHM IPC V2 negotiation
…he#3617)

* Fix MultiFlatMap erase on bucket head

* destroy element before set_invalid
Discard subcall results after the main RPC has finished while still returning their resources. Release the main call-id lock for non-final callbacks and let the final callback complete cleanup.

Add deterministic coverage for late timeout/cancel callbacks, callback ordering, controller destruction, and error propagation and retries before completion.

Tests: brpc_channel_unittest (87 passed).
* bound grpc-timeout value before converting to microseconds

* clarify what the grpc-timeout bound rejects
…pache#3601)

PR3332 sent every Zbc+Zvbc CPU to the vclmul path. After apache#3571
rewrote it, vclmul beats clmul from ~256 B up (1.5x-7.5x) but costs
more below that, where vector setup dominates. A fixed order to
either path is wrong for these CPUs.

Select by size in the dispatcher: clmul under 256 B, vclmul from
256 B on. CPUs with only Zbc or only Zvbc keep their current path,
and Extend() is untouched.

k3 (X100, VLEN=256, 20-round median):
  256 B   clmul 1534  vclmul 2155
  1 MiB   clmul 1555  vclmul 11702   (MB/s)

Signed-off-by: Xiaofei Gong <gongxiaofei24@iscas.ac.cn>
Signed-off-by: YuanSheng <yuansheng@isrc.iscas.ac.cn>
* Optimize Redis command formatting buffer handling

* Handle Redis command formatting append failures

* remove unused function
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

10 participants