...to convince myself that it is only about the initial %rsp/%rbp fiddling, and
is not affected by the ca344be7aa "tdf#143450: Fix
special fp+integer struct return case for gcc_*_x86-64" changes to later parts
of privateSnippetExecutor
Change-Id: I7bfad4cb5e1c2208e29c73a70874b6e9eff29eb2
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/119610
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
For one, the loop in x86_64::fill_struct was backwards.
And for another, privateSnippedExecutor does not need special handling of FLOAT
and DOUBLE return values (they can simply be moved to %xmm0, as covered by the
general case), but rather for those small structs where floating-point member(s)
in a first eightbyte (targeting %xmm0) are followed by integer member(s) in a
second eightbyte (targeting %rax). Extended testtools to cover two such cases.
Change-Id: I8e775a1d1ce2312610f265bcc8e40b09bdac56df
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/119576
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
...without any intended functional changes, in preparation for actual fixes for
tdf#143450 "Data corruption when returning small structs containing a double
from C++ via IPC".
Much of the code (apparently originally copied from some other project,
according to the comments) was too generic for our specific needs here.
Change-Id: Iddcb58fa0999d10dcf716dd2af7dab1620839bce
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/119570
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
This reverts commit 739aaf02db. At least in
bridges/source/cpp_uno/gcc3_macosx_x86-64/except.cxx, getRTTI can call itself
recursively, causing a deadlock on RTTI::m_mutex. In
bridges/source/cpp_uno/gcc3_linux_x86-64/rtti.cxx that has been addressed with
RTTI::getRTTI_NoLock, but other implementations have not been modified. Avoid
regressions, esp. on more esoteric platforms, by reverting the change.
Conflicts:
bridges/source/cpp_uno/gcc3_linux_x86-64/rtti.cxx
Change-Id: I08b57b0561923bdbe191c8e93f8f400e0e4df337
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/119442
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Fix these kinds:
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:103:10: error:
externally available entity 'is_complex_struct' is not previously declared in an included file
(if it is only used in this translation unit, make it static; otherwise, provide a declaration of it in an included file) [loplugin:external]
bool is_complex_struct(const typelib_TypeDescription * type)
~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:129:10: error:
externally available entity 'is_float_only_struct' is not previously declared in an included file
(if it is only used in this translation unit, make it static; otherwise, provide a declaration of it in an included file) [loplugin:external]
bool is_float_only_struct(const typelib_TypeDescription * type)
~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:168:6: error:
externally available entity 'MapReturn' is not previously declared in an included file
(if it is only used in this translation unit, make it static; otherwise, provide a declaration of it in an included file) [loplugin:external]
void MapReturn(sal_uInt32 r0, sal_uInt32 r1, typelib_TypeDescriptionReference * pReturnType, sal_uInt32* pRegisterReturn)
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:446:32: error:
static_cast from 'void **' to 'sal_Int32 *' (aka 'long *') is not allowed
sal_Int32 * pTempIndices = static_cast<sal_Int32 *>(pCppArgs + nParams);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
or some of these:
In file included from
/home/pi/lo/libreoffice/bean/native/unix/com_sun_star_comp_beans_LocalOfficeWindow.c:26:
/usr/lib/jvm/java-11-openjdk-armhf/include/jni.h:1945:1: error: unknown
attribute 'externally_visible' ignored [-Werror,-Wunknown-attributes]
_JNI_IMPORT_OR_EXPORT_ jint JNICALL
^
/usr/lib/jvm/java-11-openjdk-armhf/include/jni.h:1943:32: note: expanded
from macro '_JNI_IMPORT_OR_EXPORT_'
define _JNI_IMPORT_OR_EXPORT_ JNIIMPORT
^
/usr/lib/jvm/java-11-openjdk-armhf/include/linux/jni_md.h:35:42: note:
expanded from macro 'JNIIMPORT'
define JNIIMPORT
__attribute__((externally_visible,visibility("default")))
Change-Id: I7cae1c52ee10306da666c9c234b9af7248efd04e
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/108567
Tested-by: Julien Nabet <serval2412@yahoo.fr>
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/cpp2uno.cxx:31:10:
error: replace "..." include form with <...> for inclusion of a source file not next to the current source file,
or a header, /home/pi/lo/libreoffice/bridges/inc/cppinterfaceproxy.hxx [loplugin:includeform]
include "cppinterfaceproxy.hxx"
^~~~~~~~~~~~~~~~~~~~~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:200:41:
error: variable 'dret' is uninitialized when used here [-Werror,-Wuninitialized]
*(double*)pRegisterReturn = dret;
^~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:199:33:
note: initialize the variable 'dret' to silence this warning
register double dret asm("d0");
^
= 0.0
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:191:40:
error: variable 'fret' is uninitialized when used here [-Werror,-Wuninitialized]
*(float*)pRegisterReturn = fret;
^~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/uno2cpp.cxx:190:32:
note: initialize the variable 'fret' to silence this warning
register float fret asm("s0");
^
= 0.0
make[1]: *** [/home/pi/lo/libreoffice/solenv/gbuild/LinkTarget.mk:301 :
/home/pi/lo/libreoffice/workdir/CxxObject/bridges/source/cpp_uno/gcc3_linux_arm/cpp2uno.o] Erreur 1
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:66:22:
error: parentheses immediately inside vardecl statement [loplugin:unnecessaryparen]
long n = (*p++ - '0');
^~~~~~~~~~~~
1 error generated.
make[1]: *** [/home/pi/lo/libreoffice/solenv/gbuild/LinkTarget.mk:301 :
/home/pi/lo/libreoffice/workdir/CxxObject/bridges/source/cpp_uno/gcc3_linux_arm/cpp2uno.o] Erreur 1
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:66:22:
error: parentheses immediately inside vardecl statement [loplugin:unnecessaryparen]
long n = (*p++ - '0');
^~~~~~~~~~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:88:11:
error: externally available entity 'RTTI' is not previously declared in an included file
(if it is only used in this translation unit, put it in an unnamed namespace; otherwise, provide a declaration of it in an included file) [loplugin:external]
class RTTI
~~~~~~^~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:220:42:
error: parentheses immediately inside vardecl statement [loplugin:unnecessaryparen]
__cxa_exception const * header = ((__cxa_exception const *)pExc - 1);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:260:12:
error: statement mis-aligned compared to neighbours [loplugin:indentation]
::uno_any_destruct( pUnoExc, 0 );
^
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:246:13:
note: measured against this one [loplugin:indentation]
typelib_TypeDescription * pTypeDescr = 0;
^
/home/pi/lo/libreoffice/bridges/source/cpp_uno/gcc3_linux_arm/except.cxx:262:12:
error: statement mis-aligned compared to neighbours [loplugin:indentation]
static RTTI rtti_data;
^
Change-Id: I054212e829c57b64b23612338b238d987bb17041
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/108553
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
The Apple ARM64 ABI can use RTTI that is either unique (std::type_info equality
is decided by comparing pointers to RTTI names) or non-unique (equality is
decided by deep string comparison of RTTI names), where the most-significant bit
of a std::type_info's name pointer must be set to indicate the non-unique case.
See the " NonUniqueARMRTTIBit" comment at <https://github.com/llvm/llvm-project/
blob/be00e8893fdb814e67d8f06401afe869f878c4cf/libcxx/include/typeinfo#L143> for
details.
It appears that Clang, at least under -fvisibility=hidden as used by LO, emits
RTTI for a C++ class as .private_extern and non-unique, even if the class is
marked as visibility("default") (as is done for the C++ classes representing UNO
exception types, cf. CPPU_GCC_DLLPUBLIC_EXPORT in include/cppu/macros.hxx).
So it is expected and harmless that Rtti::getRtti will not find existing RTTI
symbols via dlsym, and needs to create RTTI on the fly. However, what was
missing was to set the most-significant bit of the name pointer, to indicate
non-uniqueness. (My understanding is that things should even have worked whe
we failed to set that bit, as comparison should fall back to deep string
comparison if at least one of the involved RTTIs has the bit set, which the RTTI
emitted by Clang does, see above. It looks like there is a bug though, see
<http://lists.llvm.org/pipermail/libcxx-dev/2020-December/001060.html>
"[libcxx-dev] Is the implementation of two-argument
__non_unique_arm_rtti_bit_impl;;__is_type_name_unique correct?")
Change-Id: I3c39edf569ac668352bbb73e60303856e1b63445
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/107839
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Now bridgetest (as improved by my previous commit) passes. That
doesn't necessarily prove much, though, I am sure there are
interesting corner cases that it doesn't exercise. Anyway, making this
work was easier than I had feared.
Unlike the arm64 ABI used on Linux, on macOS (and iOS, but we don't
really use the C++/UNO bridge for iOS) we need to pack parameters
smaller than eight bytes according to their natural alignment. Bytes
take one byte, shorts two bytes aligned at a two-byte boundary, etc.
Change-Id: I5b7dc2f8fce41ddec88df0e688898a074d8b2dad
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/105773
Tested-by: Tor Lillqvist <tml@collabora.com>
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Since Microsoft follows the general ARM64 ABI calling conventions,
and the SEH exception handling is the same, this result is a mixed
port of the gcc3_linux_aarch64 bridge and the refactored x86-64
exception handling.
I have no idea, if the complicated 32-bit handling in RaiseInfo()
is needed, as the ARM64 trampolines definitly use 64-bit code.
But since this is the first working version, I currently don't
mind much ;-)
There is definitly more potential for refactoring in the whole
bridges directory...
Change-Id: I9782a2e99c0231cdd1286af156ad312229eccf39
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/103642
Tested-by: Jenkins
Reviewed-by: Jan-Marek Glogowski <glogow@fbihome.de>
This is almost just refactoring, but with classes like type_info_
and __type_info, just following the code became tendious. I tried
to come up with better names... and in the end included some minor
code changes described below.
This patch moves the "common" code for MSVC exception handling
into msvc_shared/except.cxx. Still it contains a few small ifdefs,
so it doesn't feel like clean separation, but a lot of duplicate
code is dropped this way.
The "major" code change for amd64 is the drop of the duplicate
static RTTInfo object originally used by mscx_getRTTI and
mscx_getRTTI_len, by merging of both functions into the single,
new get() function to use a single std::find call.
Change-Id: Ib07d40e2794cde79156be3324c80ccf21a6aa8ed
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/102864
Tested-by: Jenkins
Reviewed-by: Jan-Marek Glogowski <glogow@fbihome.de>
This just a cosmetic change to minimize the diff between the
files of the different msvc bridges, by renaming variables,
adapting comments and reordering variables.
This includes the reduction of the "using namespace" to
com::sun::star, so a few types now gain an uno:: prefix.
Change-Id: I2189d2ba223ed3fb51cfa94639c6ca03d26a83e9
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/102863
Tested-by: Jenkins
Reviewed-by: Jan-Marek Glogowski <glogow@fbihome.de>
* gcc3_linux_aarch64 needs a hack to detect the reserve member in LLVM >= 10
libcxxabi __cxa_exception, in addition to the existing hack for LLVM 5.
On macOS arm64 (which shares the bridges/source/cpp_uno/gcc3_linux_aarch64/
code) we know that we always target a >= LLVM 10 libcxxabi, so we do not need
neither of the two hacks there.
(And a7d1fed245 "Hack to dynamically adapt to
__cxa_exceptiom in LLVM 5.0 libcxxabi" had introduced a "fillUnoException" vs.
"mapException" typo in the comment when it copied it from
bridges/source/cpp_uno/gcc3_macosx_x86-64/except.cxx.)
* Similarly, gcc3_linux_x86_64 needs both hacks when it is built against
libcxxabi (and the need for the LLVM 5 hack had gone unnoticed there for now).
* It is not clear to me now why gcc3_macosx_x86-64 needs the LLVM 5 hack (which
I even initially developed for that platform,
7a9dd3d482 "Hack to dynamically adapt to
__cxa_exceptiom in LLVM 5.0 libcxxabi") but not the LLVM >= 10 hack (although
it does need the corresponding hack in fillUnoException when running against
recent upstream LLVM trunk libcxxabi, f4b6f6a8ae
"Hack to dynamically adapt to __cxa_exceptiom in LLVM 11 libcxxabi".
Change-Id: I8e7a5c871fbeeaf82bbd16fa03e73f10f229da93
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/100656
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Caolán McNamara