Skip to content

Tags

Tags

v2026.11.16.00

Toggle v2026.11.16.00's commit message
python: install Python libraries under the prefix given at install time

Summary:
`add_fb_python_library` made its install directory absolute with the `CMAKE_INSTALL_PREFIX` of configure time, so `cmake --install --prefix` doesn't move a library that `install_fb_python_library` installs. eden installs `eden_scm_redact` that way, and its Windows install into `_artifacts/windows` tries to write `C:/Program Files (x86)/eden/lib/fb-py-libs/eden_scm_redact`, which fails without administrator rights; on Linux the same install tries `/usr/local`. The directory now stays relative, which `install()` resolves against the prefix at install time.

___

landed-with-radar-review

Differential Revision: D129598542

fbshipit-source-id: d1eabb64e6e9c3002471d833b4079c89ba55b460

v2026.11.09.00

Toggle v2026.11.09.00's commit message
Use `[[likely]]`/`[[unlikely]]` instead of whole-condition `FOLLY_LIK…

…ELY` (1/2)

Summary: `FOLLY_LIKELY(cond)` and `FOLLY_UNLIKELY(cond)` wrap `__builtin_expect`. When the macro is the entire condition of an `if`, the C++20 `[[likely]]` and `[[unlikely]]` attributes on the branch say the same thing in standard C++. Rewrite those statements, for example `if (FOLLY_UNLIKELY(!ok))` becomes `if (!ok) [[unlikely]]`. Uses inside larger expressions and loop conditions are unchanged. No behavior change.

Reviewed By: dtolnay

Differential Revision: D130448349

fbshipit-source-id: b4cd6e11cf01f1a082dba6f063d61b7686bd8756

v2026.10.26.00

Toggle v2026.10.26.00's commit message
Remove unreachable returns in fbcode/mastermind, fbcode/mcp, fbcode/m…

…crouter (+20)

Summary:
Clang's `-Wunreachable-code-return` (reported by continuous lint as `clang-diagnostic-unreachable-code-return`) flags `return` statements that can never execute. This diff fixes 20 of them.

Dead `return`s are deleted when they follow a statement that never completes normally (`throw`, `LOG(FATAL)`, `CHECK(false)`, `std::abort()`, `FAIL()`, ...) or a block whose every path already returns or throws (exhaustive `if`/`else`, `switch` with `default`, `try`/`catch`, `while (true)`). Behaviour is unchanged in every build mode.

Part of T294580822.

Reviewed By: dtolnay

Differential Revision: D127401280

fbshipit-source-id: 8def6f385c2e147fa0fbc38139ebbbf95a4dcb68

v2026.10.19.00

Toggle v2026.10.19.00's commit message
ThriftFieldRefIsSetToFreeFunction_mcrouter_lib

Reviewed By: Mizuchi

Differential Revision: D125619675

fbshipit-source-id: 41f004c0cdf7869cfee6085ee2568df71c0c156d

v2026.09.21.00

Toggle v2026.09.21.00's commit message
Use F14FastSet instead of sorted vector for ShadowSettings key lookup

Reviewed By: MatzeB

Differential Revision: D120657246

fbshipit-source-id: 96e984ddd56530a57223ad6ecabf8c9c7b41d98d

v2026.09.14.00

Toggle v2026.09.14.00's commit message
Avoid folly::dynamic copy in BlackholingPolicyVisitor by storing a re…

…ference

Reviewed By: cnli87

Differential Revision: D119687021

fbshipit-source-id: 09592bac8d5c6550030351de6cadbb95d9be4d71

v2026.09.07.00

Toggle v2026.09.07.00's commit message
Option to insecure-request drop for the proxy

Summary: Adds two standalone options for the mcrouter proxy.

Reviewed By: djvaporize

Differential Revision: D118724271

fbshipit-source-id: 3382c640cab5623c7ccdf3b0c8409606630c9bf9

v2026.08.31.00

Toggle v2026.08.31.00's commit message
Stop thread-jumps before shutdown frees the proxies

Summary:
`CarbonRouterInstance::shutdownImpl()` calls `proxyEvbs_.clear()`, which
destroys the `VirtualEventBase`s one at a time and frees each `Proxy` along
with them (`Proxy::createProxy` ties the two together via `runOnDestruction`).
The other proxy IO threads are still routing at that point -- the join comes
later -- and `proxies_` is a vector of raw `Proxy*` that is never cleared. So a
request that thread-jumps on a live proxy can resolve a target that `clear()`
has already freed.

Confirmed on two coredumps (aarch64 and x86-64) from the `fbpkg` /
`fbpkg.fetch` CLIs, which stand up and tear down a router on every invocation:

  mov (%rcx,%rdx,8),%rax   ; rax = proxies_[idx]      -- ok
  mov 0x18(%rax),%rbx      ; rbx = Proxy->eventBase_  -- ok, read 0
  ...
  mov 0x18(%rbx),%rcx      ; SIGSEGV at 0x18, rbx == 0

`proxies_[i]` was still readable, but `eventBase_` -- a reference member that
cannot be null in a live `Proxy` -- read 0. The object had been freed and its
memory recycled. The x86 faulting `rip` matches the stack reported in S688594.

Two halves, because there are two populations of dangerous hops:

- New hops. `proxiesDraining_` is set at the start of proxy teardown (after
  `joinAuxiliaryThreads()`, before anything is destroyed) and checked by the
  two accessors that resolve a hop target, `getProxyForThreadJump()` and
  `getProxyBaseForThreadJump()`. While draining they return nullptr and the
  caller routes locally -- the existing "stay on this thread" path, which every
  hop site already handles.

- In-flight hops. `fenceProxyEventBases()` round-trips every proxy event base
  before any `VirtualEventBase` is destroyed. No hop site suspends between
  resolving a target and enqueueing on it, so a hop that read the flag as false
  is still running inline on its source proxy's thread, and the fence task
  queued there cannot run until it finishes. Once every proxy has passed the
  fence, each pending hop is covered by a keep-alive on its target, and
  `~VirtualEventBase()` blocks on those.

Relaxed ordering is sufficient: the happens-before comes from the fence's
notification queue, not from the flag.

Cost is one relaxed load of a read-mostly flag per hop resolution, behind a
consistent-hash lookup (SRRoute) or a full route-handle traverse
(Multi\*Routes). `getProxyFromHash()` is replaced by `getProxyForThreadJump()`;
it had one in-tree caller, but was public on an OSS-mirrored header, so this is
an API break for the GitHub mirror.

Reviewed By: vrishal, stuclar

Differential Revision: D115935603

fbshipit-source-id: 3927b8d85512262cf222b106e6aa3d8230cee68f