test: remove impact of tier-up changes in worker stack size test - #64271

The depth of the stack depends not only on the stack size, but
also on the size of each stack frame, which in turn depends on
which tier the recursive function happens to be running at when
the overflow occurs. Under load the background tier-up can land
at a non-deterministic point in the recursion and flake the test.
Keep the recursive function in the interpreter with
%NeverOptimizeFunction() so the frame size - and thus the depth
- is deterministic.

Signed-off-by: Joyee Cheung <[email protected]>
PR-URL: #64271
Reviewed-By: Michaël Zasso <[email protected]>
Reviewed-By: James M Snell <[email protected]>

The depth of the stack depends not only on the stack size, but
also on the size of each stack frame, which in turn depends on
which tier the recursive function happens to be running at when
the overflow occurs. Under load the background tier-up can land
at a non-deterministic point in the recursion and flake the test.
Keep the recursive function in the interpreter with
%NeverOptimizeFunction() so the frame size - and thus the depth
- is deterministic.

Signed-off-by: Joyee Cheung <[email protected]>
PR-URL: #64271
Reviewed-By: Michaël Zasso <[email protected]>
Reviewed-By: James M Snell <[email protected]>

The depth of the stack depends not only on the stack size, but
also on the size of each stack frame, which in turn depends on
which tier the recursive function happens to be running at when
the overflow occurs. Under load the background tier-up can land
at a non-deterministic point in the recursion and flake the test.
Keep the recursive function in the interpreter with
%NeverOptimizeFunction() so the frame size - and thus the depth
- is deterministic.

Signed-off-by: Joyee Cheung <[email protected]>
PR-URL: #64271
Reviewed-By: Michaël Zasso <[email protected]>
Reviewed-By: James M Snell <[email protected]>

…ite (#53062)

Backports nodejs/node#64271. The test compares the recursion depth two
Workers reach before overflowing, but depth also depends on frame size,
which changes when the recursive function tiers up mid-recursion. On a
busy CI runner that lands non-deterministically and the depths end up
>10% apart. Pin the function to the interpreter so the depth is stable.

The fix is only in Node 26.6.0+, so it is carried as a patch until the
Node roll picks it up.

…ite (#53075)

Backports nodejs/node#64271. The test compares the recursion depth two
Workers reach before overflowing, but depth also depends on frame size,
which changes when the recursive function tiers up mid-recursion. On a
busy CI runner that lands non-deterministically and the depths end up
>10% apart. Pin the function to the interpreter so the depth is stable.

The fix is only in Node 26.6.0+, so it is carried as a patch until the
Node roll picks it up.

Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>

…ite (#53076)

Backports nodejs/node#64271. The test compares the recursion depth two
Workers reach before overflowing, but depth also depends on frame size,
which changes when the recursive function tiers up mid-recursion. On a
busy CI runner that lands non-deterministically and the depths end up
>10% apart. Pin the function to the interpreter so the depth is stable.

The fix is only in Node 26.6.0+, so it is carried as a patch until the
Node roll picks it up.

Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>

…ite (#53077)

Backports nodejs/node#64271. The test compares the recursion depth two
Workers reach before overflowing, but depth also depends on frame size,
which changes when the recursive function tiers up mid-recursion. On a
busy CI runner that lands non-deterministically and the depths end up
>10% apart. Pin the function to the interpreter so the depth is stable.

The fix is only in Node 26.6.0+, so it is carried as a patch until the
Node roll picks it up.

Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>

…ite (#53078)

Backports nodejs/node#64271. The test compares the recursion depth two
Workers reach before overflowing, but depth also depends on frame size,
which changes when the recursive function tiers up mid-recursion. On a
busy CI runner that lands non-deterministically and the depths end up
>10% apart. Pin the function to the interpreter so the depth is stable.

The fix is only in Node 26.6.0+, so it is carried as a patch until the
Node roll picks it up.

Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>

src_fix_libuv_assertion_on_windows.patch (nodejs/node#61999) and
test_remove_impact_of_tier-up_changes_in_worker_stack_size_test.patch
(nodejs/node#64271) are both included in Node.js v24.20.0.

Both landed in v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

Both patches are backports that landed in Node.js v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

Both landed in v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

Both patches are backports that landed in Node.js v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

src_fix_libuv_assertion_on_windows.patch (nodejs/node#61999) and
test_remove_impact_of_tier-up_changes_in_worker_stack_size_test.patch
(nodejs/node#64271) are both included in Node.js v24.20.0.

* chore: bump node in DEPS to v24.20.0

* fix(patch): implement package maps in the ESM resolver

Upstream added join to the path imports in resolve.js, which
overlaps the import line this patch rewrites to pull in
toNamespacedPath.

Ref: nodejs/node#62239

* chore: remove upstreamed patches

Both landed in v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

* chore: update patches (trivial only)

* fix(patch): preserve RSA-PSS legacy pubkey DER

Upstream added ReadASN1Element to ncrypto.cc but only calls it from
the OpenSSL 3 provider branch, so under BoringSSL it is an unused
function and -Werror fails the build. Guard the definition with the
same NCRYPTO_USE_OPENSSL3_PROVIDER condition as its callers.

Ref: nodejs/node#64547

* fix(patch): update sqlite to 3.53.4

The sqlite update introduced an unsigned/signed comparison in
rbuDeltaApply that trips -Wsign-compare under -Werror. Extend the
existing sign-compare patch to cast lenDelta, which is never negative
at that point, to unsigned int.

Ref: nodejs/node#64745

* fix: do not pass removed JSPI V8 flag

Node.js v24.20.0 passes --experimental-wasm-jspi unconditionally, but
Chromium's V8 shipped JSPI and removed the flag, so node_mksnapshot
aborts with 'bad option'. Add the same patch as the main-branch roll.

Ref: nodejs/node#59941
Ref: https://chromium-review.googlesource.com/c/v8/v8/+/6937664

* chore: update filenames.auto.gni

---------

Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>

* chore: bump node in DEPS to v24.20.0

* chore: remove upstreamed patches

Both patches are backports that landed in Node.js v24.20.0.

Ref: nodejs/node#61999
Ref: nodejs/node#64271

* fix(patch): implement package maps in the ESM resolver

Upstream rewrote packageResolve() to branch on hasPackageMap() and
added join to the path imports, which shifted the context around the
electron: specifier shortcut and the legacyMainResolve imports.

Ref: nodejs/node#62239

* chore: update patches (trivial only)

* fix(patch): split OpenSSL 3, BoringSSL, and legacy backends

Upstream added an ASN.1 helper, ReadASN1Element, whose only callers
live under NCRYPTO_USE_OPENSSL3_PROVIDER. With BoringSSL that macro
is 0, so the helper is unused and fails -Werror=unused-function.
Guard the definition with the same macro as its callers.

Ref: nodejs/node#64211

* chore: update filenames.auto.gni

Node.js v24.20.0 replaced src/quic/logstream.h with
src/quic/session_manager.h.

* fix(patch): update sqlite to 3.53.4

The new sqlite amalgamation compares 'unsigned int cnt' against
'int lenDelta' in rbuDeltaApply, which fails -Werror=sign-compare.
Extend the existing sqlite sign-compare patch to cast that operand
too.

Ref: nodejs/node#64745

* fix(patch): do not pass removed JSPI V8 flag

Node.js v24.20.0 backported JSPI enablement, which unconditionally
passes --experimental-wasm-jspi to V8. Chromium's V8 removed that
flag (JSPI is always on), so node_mksnapshot aborts with
"bad option". Drop the argument in a new patch, matching main.

Ref: nodejs/node#59941
Ref: https://chromium-review.googlesource.com/c/v8/v8/+/6937664

---------

Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: Sam Attard <[email protected]>