regression from
commit 18f5131454
Date: Tue Jun 20 19:54:23 2017 +0200
loplugin:unusedfields store..svl
somewhat tricky to reproduce on linux, I had to do
make soltools CPPFLAGS=-DDEBUG
Change-Id: I4b1989efc02b9e864432b96cf4f71a29e581bcdb
Reviewed-on: https://gerrit.libreoffice.org/47755
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
under --with-lang=ALL it had started to crash during e.g.
> LD_LIBRARY_PATH=${LD_LIBRARY_PATH:+$LD_LIBRARY_PATH:}".../instdir/program:.../instdir/program" .../workdir/LinkTarget/Executable/cpp -+ -P -DWITH_POSTGRESQL_SDBC -DWITH_FIREBIRD_SDBC -DWITHOUT_EXTENSION_MEDIAWIKI -DWITHOUT_SCRIPTING_BEANSHELL -DWITHOUT_SCRIPTING_JAVASCRIPT -DWITH_HELPPACK_INTEGRATION -DWITH_EXTENSION_INTEGRATION -DENABLE_EXTENSION_UPDATE -DX86_64 -DLINUX -D_PTHREADS -DUNIX -DUNX -DCOMID=gcc3 -D_gcc3 -DWITH_LPSOLVER -I.../scp2/inc -I.../workdir -I.../config_host -I.../workdir/CustomTarget/scp2/macros -I.../workdir/ScpTemplateTarget/scp2/source/templates .../scp2/source/calc/file_calc.scp > .../workdir/ScpPreprocessTarget/scp2/source/calc/file_calc.pre
because of
> Syscall param read(buf) points to unaddressable byte(s)
> at 0x4F31A80: __read_nocancel (syscall-template.S:84)
> by 0x404F98: fillbuf (_lex.c:631)
> by 0x404CCF: gettokens (_lex.c:479)
> by 0x400F0A: process (_cpp.c:77)
> by 0x400E80: main (_cpp.c:60)
> Address 0x5278494 is 0 bytes after a block of size 32,772 alloc'd
> at 0x4C2DB9D: malloc (vg_replace_malloc.c:299)
> by 0x402034: domalloc (_cpp.c:321)
> by 0x40554F: setsource (_lex.c:679)
> by 0x403E38: doinclude (_include.c:130)
> by 0x401F9D: control (_cpp.c:297)
> by 0x401002: process (_cpp.c:101)
> by 0x400E80: main (_cpp.c:60)
There appears to be no other check that fillbuf doesn't overflow the Source's
input buffer, other than gettokens checking that the buffer isn't more than
three quarters full ("if (ip >= s->inb + (3 * INS / 4)) ..."). That smells like
cpp assumes input lines to be shorter than some maximum number of characters
(like the C99 standard setting a minimum limit of "4095 characters in logical
source lines"), and
> #define README_TXT_ALL_LANG(key, name, ext) \
> key (af) = READMETXTFILENAME(name,_af,ext); \
> Name (am) = CONFIGLANGFILENAME(name,am,ext); \
> Name (ar) = CONFIGLANGFILENAME(name,ar,ext); \
[...]
> Name (zh-CN) = CONFIGLANGFILENAME(name,zh-CN,ext); \
> Name (zh-TW) = CONFIGLANGFILENAME(name,zh-TW,ext); \
> key (zu) = READMETXTFILENAME(name,_zu,ext)
in workdir/CustomTarget/scp2/macros/langmacros.inc (which appears to be the
culprit here) exceeding that limit under --with-lang=ALL.
So just bump the input buffer size.
Change-Id: I5d863050fb772dc7e691a604009ff8702dc718e3
The types 'long' and 'unsigned long' are different sizes on different
platforms, making them wholy unsuitable for portable code.
And when I mean different sizes, I mean 64bit Linux and 64bit Windows
have different bit sizes.
Change-Id: Id4d8cd709bf71a3f85ed6515fae2b0a21ae8c25d
Reviewed-on: https://gerrit.libreoffice.org/41130
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
Translates leftovers found using a custom regex and manually checking
the rest of the affected file.
Additionally:
- Translation of someone's attempts at Spanish in the .java file
Change-Id: Ie3a761bc1a8e302247d5dd6e93a6bc4cb3e85a9f
Reviewed-on: https://gerrit.libreoffice.org/39609
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>
...that would implicitly be sign extended (for plain char only if it is signed),
so non-ASCII char values would trigger the isUnicodeCodePoint assert.
Change-Id: Iaf8024ad509e64525558e882fe3fd078cfb4ea91
Reviewed-on: https://gerrit.libreoffice.org/35523
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
Found with coccinelle: http://coccinelle.lip6.fr/
Currently only applied to C code.
(cherry picked from commit aaf15142bcbc44f0e06a3357d2b21ecf9c673d21)
Change-Id: I7d43b024752ebd47e84e2e51028f8fd132410800
For MSVC/clang-cl: "__STDC__: Indicates conformance with the ANSI/ISO C99
tandard. Defined as the integer literal constant 1 only if the /Za compiler
option is given and you are not compiling C++ code; otherwise is undefined."
(<https://msdn.microsoft.com/en-us/library/b0084kay.aspx>)
Presumably fine to just replace #if with #ifdef here (as is used in all the
other cases surrounding it), and not care whether it actually expands to true.
Change-Id: Ie58c76d7e72321b06a11a370fd28221e85fcf871
Whatever those _POSIX_ for _MSV_VER were supposed to be good for (have been in
there at least ever since 4f4b74153d "Moved cpp
from cpp to soltools"), but cause -Wunused-macros at least when building with
clang-cl.
Change-Id: I356bf663cd1e86298ba72a179f46bb9d0c10016b
stage 2 of replacing usage of various checks for the windows platform
with the compiler-defined '_WIN32' macro
In this stage we focus on replacing usage of the WIN macro
Change-Id: Ie8a4a63198a6de96bd158ecd707dadafb9c8ea84
Reviewed-on: https://gerrit.libreoffice.org/22393
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
stage 1 of replacing usage of various checks for the windows platform
with the compiler-defined '_WIN32' macro
Change-Id: Iece73abdee530937e0737190b1aa97a46cd3075f
Reviewed-on: https://gerrit.libreoffice.org/22390
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Mike Kaganski