date | project | content | link |
|---|
| 17:10 on Apr 07 | haiku | Commit by bonefish on master :: r hrev43970 haiku/: ( link) Fix #8420: NetFS does not compile with DEBUG Debug macros like PRINT, ERROR, etc. are defined differently in NetFS and UserlandFS. In NetFS they use single parentheses while in UserlandFS double parentheses are required. Somehow this got mixed up in NetFS and there were both styles of calling these macros what caused the incorrect one to produce compilation errors. Signed-off-by: Ingo Weinhold < ingo_weinhold at gmx dot de> | # |
| 13:57 on Apr 07 | haiku | Commit by bonefish on master :: r hrev43968 haiku/: ( link) Fix #4125: NULL device is translated to '//' The problem appeared to be on the request creating side (i.e. in the kernel add-on) which did not support NULL pointers properly. Relocation of addresses in request when it is received translates offset = 0, size = 0 to pointer NULL so that no change in that part of code was required. Signed-off-by: Ingo Weinhold < ingo_weinhold at gmx dot de> | # |
| 15:35 on Apr 05 | haiku | Commit by bonefish on master :: r hrev43955 haiku/: ( link) switch_sem_etc(): Unbreak the actual switch case
The scheduler lock was still being held when release_sem_etc() was called. The bug was introduced in 24df65921befcd0ad0c5c7866118f922da61cb96. Fixes #8448. | # |
| 11:15 on Apr 05 | haiku | Commit by bonefish on master :: r hrev43954 haiku/: ( link) Fixed a compilation failure for the x86_64 toolchain. This appears to be a problem with GCC's build system: it defaults to having multilib enabled, but if it is explicitly enabled with --enable-multilib, the build fails. Signed-off-by: Ingo Weinhold < ingo_weinhold at gmx dot de> | # |
| 10:49 on Apr 05 | haiku | Commit by bonefish on master :: r btrev43030 buildtools/: ( link) Updated the x86_64 toolchain to fix build failures and to be more correct.
- A build failure occurs as TARGET_64BIT is no longer a compile-time constant in GCC.
- Made some changes so that generated code matches the i386 Haiku target when generating 32-bit code, and follows the x86_64 ABI for 64-bit code.
- Added a comment stating that DEFAULT_PCC_STRUCT_RETURN needs to be defined to follow the ABI for 32-bit code, it is ignored for 64-bit.
- Remove definition of __x86_64__, already defined by GCC's i386-c.c.
- Remove -z max-page-size from LINK_SPEC as it shouldn't be necessary: the default page size is 2MB but this won't result in large binaries, it just changes the alignment of the sections and would make it easier to map binaries using large pages in the future. Signed-off-by: Ingo Weinhold <ingo_weinhold at gmx dot de>
| # |
| 10:32 on Apr 05 | haiku | Commit by bonefish on master :: r hrev43953 haiku/: ( link) Fix dladdr behaviour
- If dladdr can't find an exact match, it returns the nearest symbol less than the given address.
- If no suitable symbol can be found, but the address is within a loaded library, dladdr returns the library name and base address. Signed-off-by: Ingo Weinhold <ingo_weinhold at gmx dot de>
| # |
| 19:53 on Apr 04 | haiku | Commit by bonefish on master :: r hrev43949 haiku/: ( link) userlandfs: Remove superfluous driver_settings.c
It probably dates back to a time before our libroot exported the driver settings API as well. | # |
| 14:16 on Feb 19 | haiku | Commit by bonefish on master :: r hrev43757 haiku/: ( link) View.cpp: Use B_IF_GCC_2 to simplify things [2 commits] | # |
| 14:04 on Feb 12 | haiku | Commit by bonefish on master :: r hrev43737 haiku/: ( link) create_thread_user_stack(): Fix incorrect handling of TLS size
In case a custom stack was given, the TLS size was subtracted from the stack base, not the stack size. | # |
| 21:55 on Feb 10 | haiku | Commit by bonefish on master :: r hrev43734 haiku/: ( link) Add GPT module to the boot modules
This is necessary so a disk with a protective MBR is not incorrectly recognized when the kernel first scans it. | # |
| 21:17 on Dec 08, 2011 | haiku | Commit by bonefish on master :: r hrev43438 haiku/: ( link) media addon server: Fix incorrect array offset for safe mode
Missed that when back porting 323b65468e5836bb27a5e373b14027d902349437. | # |
| 04:11 on Nov 28, 2011 | haiku | Commit by bonefish on master :: r hrev43352 haiku/: ( link) FreeBSD host sys/stat.h: Add missing extern "C" | # |
| 04:02 on Nov 28, 2011 | haiku | Commit by bonefish on master :: r hrev43351 haiku/: ( link) fs_attr_generic.cpp: Fix warning [3 commits] | # |
| 05:20 on Nov 25, 2011 | haiku | Commit by bonefish on master :: r hrev43319 haiku/: ( link) PackageLinkDirectory::_Update(): Don't unlock after delete [243 commits]
When using an AutoLocker on an object, the lock must be released explicitly before releasing the last certain reference to that object. | # |
| 19:22 on Nov 15, 2011 | haiku | Commit by bonefish on master :: r hrev43274 haiku/: ( link) Fix the gcc bison issue for real
Also touch cp/parse.{c,h}, which apparently are the culprit. | # |
| 18:49 on Nov 15, 2011 | haiku | Commit by bonefish on master :: r hrev43273 haiku/: ( link) Work around issue with newer bison versions
Touch the source files generated by bison, so that make won't try to have them updated. | # |
| 14:40 on Nov 12, 2011 | haiku | Commit by bonefish on master :: r hrev43241 haiku/: ( link) Added BuildPlatformStaticLibraryPIC rule. [6 commits]
It works like BuildPlatformStaticLibrary, but generates position independent code. | # |
| 12:32 on Nov 12, 2011 | haiku.test | Commit by bonefish on master :: r hrev43243 haiku/: ( link) Added BuildPlatformStaticLibraryPIC rule. [6 commits]
It works like BuildPlatformStaticLibrary, but generates position independent code. | # |
| 22:52 on Nov 07, 2011 | OpenBeOS | Commit by bonefish :: r43219 /haiku/trunk/src/bin/debug/scheduling_recorder/scheduling_recorder.cpp: scheduling_recorder: Fix signal handling
- The sa_flags field of the structure passed to sigaction() wasn't cleared, which resulted in the wrong type of handler function being called and the second parameter not being the expected user data.
- Always check fCaughtDeadlySignal in the main loop, so the signal doesn't have to hit the _kern_system_profiler_next_buffer() call (though that's the most likely one anyway). Now scheduling_recorder can be stopped with a single Ctrl-C again.
| # |
| 22:35 on Nov 07, 2011 | OpenBeOS | Commit by bonefish :: r43218 /haiku/trunk/src/system/kernel/messaging/KMessage.cpp: Build fix | # |