Age | Commit message (Collapse) | Author |
|
Notes:
prefer: 9b6062afb325c0396d88fdc80a994ec0ccde4852
|
|
Notes:
prefer: 3901964b6d1fb2117cb983a7eedf80da93ba5530
|
|
Notes:
ignore: aoo
|
|
what a cluster*
Notes:
ignore: aoo
|
|
contructor -> constructor
convinient -> convenient
Notes:
prefer: 48f0a5ed3501e0b0a55cc71f85b4d1aee5a9f4ed
|
|
Clang version 4.0 and newer uses SSE instructions that require
16-byte alignment to zero memory allocated using the C++ new operator.
The internal memory allocator does not understand anything larger
than 8-byte alignment. Modify it to be capable of doing 16-byte
alignment when necessary.
There is also a debug layer beneath the C++ new and delete operators
that is enabled by the --enable-debug configure option. This layer
adds 8 to the requested size of any allocations before calling the
underlying allocator, adds a known signature to the start of the
memory block, and then adds an 8 byte offset to the pointer before
returning it to the caller. The delete operator basically does the
reverse, checking for the proper signature. Modify this code so
that it adds and subtracts a 16-byte offset on X86_64 so that a
properly aligned block from the underlying allocator does not cause
new to return a misaligned pointer.
Modify set_soenv.in so that it always requests 16-byte alignment
on X86_64 so that the ABI is the same independent of the toolchain.
Notes:
prefer: cfd2691e5bf398304d7816227b933e53907b6204
|
|
of the Windows API wrapper "uwinapi" to AMD64.
We aren't going to rewrite it in AMD64 assembly
as most of the uwinapi thunked/emulated
functions are very old, and AMD64 only came out
in the Windows XP and Windows 2003 era, so only
function not available on those versions of
Windows such as SHCreateItemFromParsingName
and (unbelievably!!) snprintf and co need to
be ported. And I've used C code instead of
assembly as it's clearer and performance isn't
that important nowdays.
Patch by: me
Notes:
ignore: aoo
|
|
on types of different sizes.
Patch by: me
Notes:
prefer: 7e86fc3fe72f67b0d9240b4afad28c66dc482c80
|
|
Patch by: me
Notes:
prefer: 7e86fc3fe72f67b0d9240b4afad28c66dc482c80
|
|
mapping handling.
Notes:
ignore: obsolete
|
|
CERT C coding standard seem to have changed its mind.
MEM02-A. Do not cast the return value from malloc().
has been superceeded by:
MEM02-C. Immediately cast the result of a memory allocation function call
into a pointer to the allocated type.
Notes:
ignore: aoo
|
|
Undo unnecessary change from r1782205: using a union in oslSocketAddrImpl
just makes the code more complex and doesn't improve aliasing.
Notes:
ignore: aoo
|
|
This is a very small part of a patch submitted by Caolan McNamara on
2009 to help OOo work with -fstrict-aliasing. It is not complete and for
now I omitted adding -fno-strict-aliasing to many makefiles.
This does require a lot more attention and will have to be completed at
some point because newer versions of GCC enable strict-aliasing with most
optimization levels.
Notes:
prefer: 11513ddf5b05cb7b3c1e91f9728159b9be3a5da2
|
|
This allows Ubuntu 16.04 to compile AOO with Clang 1.8 on an x86.
Other architectures probably don't work, as Clang 1.8 is pretty
strict about the low-level C++ ABI hacks in main/bridges, but at
least they should fail early in main/bridges instead of the
previous behaviour, where main/bridges compiles nothing when
"$(COM)" != "GCC" and main/i18npool then fails with a mysterious
"error: cannot get uno environments" message.
Patch by: me
Notes:
prefer: bf1f0183d5c6b4c94acdbee27276d5a386a657f4
|
|
measurement. The nominal hold time for the mutex is 5 seconds, but
because the mutex state is polled at 1 second intervals, the actual
measured time will probably be 6+ seconds. If the nanoseconds field
of the starting timestamp is large, it is possible that it will have
wrapped to a small value at the poll time when the mutex is detected
as having been released. If the time interval was calculated at
full precision by subtracting both the seconds and nanoseconds fields
of the before and after timestamps, an interval of just over 6
seconds could look like 7 seconds and a large negative number of
nanoseconds. Since this test only subtracts the seconds field, it
can think the difference is 7 seconds and fail the "nSec < 7" assertion.
As a quick fix, change the assertion to "nSec <= 7".
Notes:
reject: seems to work for us
|
|
Thinking of a future ARM support that may want to give some value to
FRAME_PTR_OFFSET and FRAME_OFFSET.
Notes:
ignore: obsolete
|
|
Submitted by: Curtis Hamilton
Notes:
merged as: 5f2f0d9799d9ab57624e03e235b32333ad961ac2
|
|
Revisions were used within the tree, perhaps inherited by previous
CVS/Subversion usage. Re-use them now since we have been lazy about
updating them manually and because they can be useful for bug reports.
Notes:
ignore: obsolete
|
|
Most of these pre-existed in the code for some reason. New Id keywords
were only added in these files:
main/solenv/inc/minor.mk
main/vcl/unx/kde4/KDEXLib.cxx
The latter two are important as the keep release information and we want
to keep a reference when new versions are bumped.
While here update the information for KDE.
Notes:
ignore: aoo
|
|
This produces a working AOO on FreeBSD PowerPC64!
This is related to i126615 which was also fixed by Curtis a while ago.
Kudos for both contributions, makes me want to get a PowerPC to run
FreeBSD + AOO on it!
Author: Curtis Hamilton -- cmhamilto at gmail
Notes:
reject: wont apply
|
|
__attribute__((__unused__)), but both clang and gcc seem to understand
both.
Since __attribute__((__unused__)) can preceed the variable name, just
like __pragma(warning(suppress:4100;suppress:4101)) for Visual Studio,
so de-parameterize the SAL_UNUSED macro.
Gcc whines if __attribute__((__unused__)) is used for a struct member,
so create a separate define, SAL_UNUSED_MEMBER, which is empty for
gcc, to handle struct members.
Notes:
prefer: 7c704c78d3c652504c064b4ac7af55a2c1ee49bb
|
|
compiler warning from clang.
Notes:
prefer: 067dd53523ecf2c5a6343c6f5b84e4074d9e9a7d
|
|
ScDoubleRawToken contains four unused private fields. Mark these as
unused to eliminate the warning message. We can't delete these fields
because we want the layout of this structure to match ScRawToken.
Notes:
prefer: d6500b19afc6a15e63e679765317690eac0d4207
|
|
Why worry about some sprinkled German when we haven't yet learned English.
Notes:
prefer: b37a07f2da6c4fcc1a01cc3ea547c558c23edd71
|
|
Only applied to C code: common style for C++ is different.
Found with coccinelle: http://coccinelle.lip6.fr/
Notes:
prefer: 5bbdb9423e15b68438bb8397c15635e044129e28
|
|
Found with coccinelle: http://coccinelle.lip6.fr/
Currently only applied to C code.
Notes:
merged as: 429a15cc292a92c2f00be6febde7da66ed792b70
|
|
rsion warnings about DbgOut()
If the definition of the overloaded function DbgOut() in
sal/inc/rtl/string.hxx is compiled without having first compiling
the the declaration of DbgOut() with a different signature in
tools/inc/tools/debug.hxx, the compiler thinks DbgOut() is calling
itself recursively with no means of escape.
Fix this in the case where DBG_UTIL is defined by including
<tools/debug.hxx> inside string.hxx so that the compiler will
encounter the declaration of DbgOut() in debug.hxx first in this
case. This requires adding a couple of directories to the compiler
include path so that the new consumers of <tools/debug.hxx> can
find it as well as one of its dependencies.
If DBG_UTIL is not defined, then debug.hxx does not declare DbgOut(),
so fix this case by hiding the DbgOut() definition in string.hxx
behind #ifdef DBG_UTIL since it won't be used in this case.
Notes:
reject: sal depending on tools is *insane*
|
|
need it for pipes too, and even if it isn't necessary it can't hurt.
In particular, on FreeBSD 11-CURRENT it seems pipes no longer wake up
from accept when closed in other threads, so let's deal with that before
FreeBSD 11 is released.
Reported by: Matthias Apitz <g u r u a t u n i x a r e a d o t d e>
Patch by: me
Tested by: Matthias Apitz <g u r u a t u n i x a r e a d o t d e>
Notes:
merged as: ed983eb5ce71d0b9af07795eabd8686f24c4597e
|
|
Patch by: me
Notes:
ignore: aoo
|
|
Fix a regression caused by r1710853 and which always existed in some files,
where <prewin.h> and <postwin.h> were being included before and after
<windows.h>, yet unavailable as the tools module isn't built (and can't be)
before sal.
Patch by: j.nitschke at ok.de
Review by: me
Notes:
ignore: obsolete
|
|
All systems should have strlcpy.
Notes:
prefer: 8f210c26c53cf5480819d983e57977d531422ff1
|
|
Patch by: me
Notes:
merged as: cfe08df695c046371c4361a434176e6381e3e064
|
|
integer.
Patch by: me
Notes:
merged as: 66add63dda22321476653d06a61bc61090fafc58
|
|
In ANSI C these casts are unnecessary and can hide warnings.
If this code is moved to C++ we may need such casts back
but we may then consider re-writing the allocation code
altogether.
Notes:
prefer: b15b51b76997c5b4cce5fa3d70df99acc1a5439a
|
|
Condition broadcast failed but silently try to unlock the mutex.
CID: 705120
Notes:
prefer: 914018c03a492f433e1ae5081aa3184cded9783f
|
|
CID 705623, 982797
Notes:
prefer: 95f8cb6823f21dd97b380136286e87e0ffa2fb24
|
|
CID: 706154
Notes:
prefer: 8f210c26c53cf5480819d983e57977d531422ff1
|
|
Patch by: me
Notes:
merged as: 7a93f5348007ab1ce52e65cd3811032f31839689
|
|
FRAME_PTR_OFFSET
in both backtrace.c and diagnose.c - it's 3 because of the EBP/RBP register's position
in the jmp_buf (https://github.com/freebsd/freebsd/blob/master/lib/libc/amd64/gen/_setjmp.S
and https://github.com/freebsd/freebsd/blob/master/lib/libc/i386/gen/_setjmp.S).
This gets backtracing to fully work on FreeBSD.
Patch by: me
Notes:
merged as: fa326746ba70b0302f11533a5cf36daaa6235d04
|
|
Patch by: me
Notes:
merged as: a1a8aa8ef80066dd2fbabbf136dde874c864db8b
|
|
The Bugzilla issue has another fix for sparc but I have no way ot test it.
Notes:
merged as: 0bd3aa40f5095e13d10f73076d8528178d146664
|
|
CID: 1028069
Notes:
prefer: c04fde0063276e3d5a6de3e077c1c822ed1ad25e
|
|
Found by: Pathangi Janardhanan Jatinshravan <JATINSHR001 at e dot ntu dot edu dot sg>
Patch by: Pathangi Janardhanan Jatinshravan <JATINSHR001 at e dot ntu dot edu dot sg>
Review by: me
Notes:
prefer: 8f823b7d39154d2044e08174a858f8ef357a151a
|
|
breaks the build.
Notes:
ignore: obsolete
|
|
and run all that pass during the build.
Notes:
ignore: obsolete
|
|
and main/sal/qa/OStringBuffer.
Notes:
ignore: obsolete
|
|
Patch by: Giuseppe Castagno
Notes:
prefer: b49a2f7373e950efc3fe11b89309313ac69c1e1f
|
|
Patch by: Giuseppe Castagno
Added a Windows API missing error:
ERROR_CANT_ACCESS_FILE
file cannot be accessed by the system.
Notes:
prefer: b49a2f7373e950efc3fe11b89309313ac69c1e1f
|
|
Notes:
prefer: 309aa845a8e413519d634680aff112a3567e2e61
|
|
thread stuck in accept().
Generalize the "#if defined(LINUX)" workarounds to the *BSDs.
Notes:
merged as: 309aa845a8e413519d634680aff112a3567e2e61
|