config/rules.mk | 2 2 0 - 0 !
1 file changed, 2 insertions( )

 fix soname

Add soname switch to linker, regardless of Operating System

Bug-Debian: http://bugs.debian.org/746705

js/src/wasm/WasmSignalHandlers.cpp | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

 hurd support

python/mozbuild/mozbuild/backend/recursivemake.py | 6 3 3 - 0 !
1 file changed, 3 insertions( ), 3 deletions(-)

 build: copy headers on install instead of symlinking

Patch ported forward to mozjs52 by Philip Chimento
<[email protected]>.

js/src/Makefile.in | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

 increase the test timeout for slower buildds

Bug-Debian: https://bugs.debian.org/878284

js/src/jit/ProcessExecutableMemory.cpp | 12 12 0 - 0 !
1 file changed, 12 insertions( )

 on ia64, retry failed mmap without address hint

[smcv: Move the #endif so we still return a defined value on non-ia64]
Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=897178
Last-Updated: 2018-07-10

modules/fdlibm/src/math_private.h | 4 4 0 - 0 !
1 file changed, 4 insertions( )

 fix math_private.h for i386 ftbfs

js/src/tests/lib/manifest.py | 15 14 1 - 0 !
1 file changed, 14 insertions( ), 1 deletion(-)

 tests: add the ability to skip tests according to dpkg-architecture

This avoids needing to work out what the value of XPCOMABI will be
for a particular Debian architecture. It cannot be used in
upstreamable patches, though.

js/src/tests/jstests.list | 12 12 0 - 0 !
1 file changed, 12 insertions( )

 tests: expect some floating-point tests to fail on i386

Upstream assumes everything uses SSE mathematics, but SSE isn't in
Debian's baseline for i386, so we are stuck with i387 and its weird
80-bit floating point registers. We can't even work around this with
-fexcess-precision=standard, because mozjs is C  .

Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=907363

js/src/tests/jstests.list | 6 6 0 - 0 !
1 file changed, 6 insertions( )

 tests: expect some tests to fail on armel

ARM softfloat has unusual behaviour in corner cases, like these tests.
Upstream only supports the equivalent of our armhf.

Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=908481

js/src/tests/non262/Array/regress-157652.js | 2 1 1 - 0 !
js/src/tests/non262/Array/regress-330812.js | 2 1 1 - 0 !
js/src/tests/non262/regress/regress-422348.js | 2 1 1 - 0 !
3 files changed, 3 insertions( ), 3 deletions(-)

 tests: use deb_host_arch_bits to skip some tests on 64-bit platforms

Ideally this would use some upstreamable mechanism instead of
modifying the test infrastructure to make dpkg-architecture variables
available, but I don't know the build/test infrastructure well enough
to implement that.

Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=905825

js/src/tests/jstests.list | 3 3 0 - 0 !
1 file changed, 3 insertions( )

 tests: expect a test to fail on big-endian

Signed-off-by: Simon McVittie <[email protected]>

js/src/tests/jstests.list | 3 3 0 - 0 !
1 file changed, 3 insertions( )

 jstests: skip non262/date/time-zones-imported.js

This seems to fail both using system and embedded ICU, further study
is needed.

Skipping it for now.

js/src/tests/jstests.list | 4 4 0 - 0 !
1 file changed, 4 insertions( )

 jstests: skip fp-evaluation-order test in some archs

This test is not working properly in some architectures due to
wrong ToNumber JS-to-C conversion for JS unsafe integers.

Bug-Mozilla: https://bugzilla.mozilla.org/show_bug.cgi?id=1755531

js/src/tests/non262/Array/regress-157652.js | 6 3 3 - 0 !
js/src/tests/non262/Array/regress-330812.js | 6 3 3 - 0 !
js/src/tests/non262/Function/regress-338001.js | 2 1 1 - 0 !
js/src/tests/non262/extensions/regress-336409-1.js | 6 3 3 - 0 !
js/src/tests/non262/regress/regress-303213.js | 6 3 3 - 0 !
js/src/tests/non262/regress/regress-422348.js | 4 2 2 - 0 !
6 files changed, 15 insertions( ), 15 deletions(-)

 tests: accept "out of memory" as close enough to "allocation size
 overflow"

On weaker machines like Debian's mipsel porterbox eller, some of these
tests will run out of memory before the allocation size can overflow.

Signed-off-by: Simon McVittie <[email protected]>
Bug-Debian: #1006196

js/src/tests/jstests.list | 3 3 0 - 0 !
1 file changed, 3 insertions( )

 tests: skip comparison of tzdata version number

The test expects the timezone database version number to be reported as
2021a3, but it's actually reported as 2021a.

build/moz.configure/rust.configure | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

 allow to build with cargo in debian unstable

intl/icu/source/common/unicode/std_string.h | 5 0 5 - 0 !
intl/icu/source/common/utypeinfo.h | 3 0 3 - 0 !
intl/icu/source/io/unicode/ustream.h | 4 0 4 - 0 !
3 files changed, 12 deletions(-)

 remove workaround for old libstdc   problem,
 which now causes problems with GCC 12 on arm

Bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1786621

js/src/jit/GenerateAtomicOperations.py | 6 3 3 - 0 !
1 file changed, 3 insertions( ), 3 deletions(-)

 bug 1761665 - extend x86 workaround to 32-bit arm

If we don't do this, Spidermonkey (mozjs) on ARMv7 fails to compile in
some configurations.

Bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1761665
Bug-Debian: https://bugs.debian.org/1017962

js/moz.configure | 9 6 3 - 0 !
1 file changed, 6 insertions( ), 3 deletions(-)

 js: enable jit by default on armv7 or up, but not armv6 or older

Bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1786619
Bug-Debian: https://bugs.debian.org/1017961
Signed-off-by: Simon McVittie <[email protected]>

js/moz.configure | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

 disable simd on 32-bit arm and x86

SIMD (SSE/NEON) is not in the baseline of the i386/armel/armhf ports.

Bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1786619
Bug-Debian: https://bugs.debian.org/1017961

js/src/jit/shared/AtomicOperations-shared-jit.cpp | 4 2 2 - 0 !
1 file changed, 2 insertions( ), 2 deletions(-)

 fix the non-jit armel build

testing/web-platform/tests/tools/wptrunner/wptrunner/products.py | 9 1 8 - 0 !
1 file changed, 1 insertion( ), 8 deletions(-)

---

testing/web-platform/manifestupdate.py | 9 4 5 - 0 !
1 file changed, 4 insertions( ), 5 deletions(-)

---

testing/web-platform/manifestupdate.py | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

---

testing/mozbase/mozdevice/mozdevice/adb.py | 6 3 3 - 0 !
testing/mozbase/mozdevice/mozdevice/remote_process_monitor.py | 2 1 1 - 0 !
testing/mozbase/mozrunner/mozrunner/application.py | 6 3 3 - 0 !
3 files changed, 7 insertions( ), 7 deletions(-)

---

js/public/Value.h | 2 1 1 - 0 !
1 file changed, 1 insertion( ), 1 deletion(-)

 export js::detail::canonicalizednanbits on architectures that use it

Otherwise the inline function JS::CanonicalizeNaN(double), which is
called by gjs, cannot validly refer to it.

Bug-Debian: https://bugs.debian.org/1063446
Signed-off-by: Simon McVittie <[email protected]>