summaryrefslogtreecommitdiffstats
path: root/libpthread
Commit message (Collapse)AuthorAgeFilesLines
...
* Merge from trunk that gives us a valid NPTL dynamic loader and the start of ↵"Steven J. Hill"2005-11-2927-350/+114
| | | | the new build system. I have probably another 300 and some files to go *sigh*.
* Dynamic loader now builds with new build system. Fix problems with ↵"Steven J. Hill"2005-11-203-160/+0
| | | | 'dl-sysdep.h' and the loader that I should have caught a while ago.
* Sync up with trunk."Steven J. Hill"2005-11-193-5/+6
|
* Copy from trunk."Steven J. Hill"2005-11-1954-0/+3620
|
* Massive merge from trunk."Steven J. Hill"2005-11-183-90/+79
|
* Copy from trunk."Steven J. Hill"2005-11-1646-0/+2967
|
* Copy from trunk."Steven J. Hill"2005-11-1664-0/+14160
|
* Remove old directories."Steven J. Hill"2005-11-16108-17622/+0
|
* Sync with glibc reference tree and changes for uClibc thus far."Steven J. Hill"2005-11-1638-281/+286
|
* Clean up -O optimizations for MIPS and fix libcrypt Makefile temporarily."Steven J. Hill"2005-11-161-0/+5
|
* Remove '__libc_fatal' function and usage of it. Remove and disable 'freeres' ↵"Steven J. Hill"2005-11-167-24/+19
| | | | code having to do with internal library memory usage and GDB. uClibc homey don't play that.
* Sync up with glibc NPTL and clean up unused source files."Steven J. Hill"2005-11-1626-199/+1501
|
* Merge from trunk."Steven J. Hill"2005-10-304-165/+19
|
* Copy from trunk."Steven J. Hill"2005-10-301-0/+61
|
* Copy from trunk."Steven J. Hill"2005-10-301-0/+26
|
* Copy from trunk."Steven J. Hill"2005-10-301-0/+64
|
* Copy from trunk."Steven J. Hill"2005-10-301-0/+41
|
* Correct stored name of shared object."Steven J. Hill"2005-10-081-2/+1
|
* Sync with trunk."Steven J. Hill"2005-10-081-0/+4
|
* Sync with trunk."Steven J. Hill"2005-10-061-26/+17
|
* Fix up build system for NPTL. I admit it is much cleaner now, but I lost my ↵"Steven J. Hill"2005-10-0411-298/+302
| | | | entire Saturday fixing this $!#$!#@$T! up.
* Big, huge fricking merge. Peter, if you touch the build system again I'm ↵"Steven J. Hill"2005-10-046-155/+86
| | | | going to smash your keyboard.
* Cleaned up 'struct link_map' and 'struct elf_resolve' such that the TLS data ↵"Steven J. Hill"2005-09-251-0/+2
| | | | items are located in the same place. This allows for casting the types back and forth between the pthreads library and the dynamic loader. Cleaned up 'ldsodefs.h' and imported all of the functions from 'dl-tls.c' into the dynamic loader. Consequently, the dynamic loader grew in size by 50%. MIPS is probably the worst case as far as size, but still expect the other architectures to have a double digit percentage size increase.
* Rename option PTHREADS_NATIVE to be UCLIBC_HAS_THREADS_NATIVE instead. What ↵"Steven J. Hill"2005-09-241-5/+5
| | | | a stupid name to start with. What idiot thought of that?
* Big sync with trunk."Steven J. Hill"2005-09-241-13/+25
|
* Added TLS variables to 'struct elf_resolve' which will be needed for TLS ↵"Steven J. Hill"2005-09-211-3/+1
| | | | support. Created 'dl-tls.c' specifically for ldso instead of trying to share with the one used by the C library. It was getting to be too much of a hassle and this way hopefully all of the TLS functions can be in one place making maintenance much easier. Added new TLS relocation handlers for MIPS. The code is correct, but the variables used to perform the relocations are not set properly as more changes are necessary to the core of ldso.
* Sync with trunk."Steven J. Hill"2005-09-171-1/+4
|
* Cleaned up 'sysdep.h' file madness."Steven J. Hill"2005-09-134-10/+8
|
* Merge with trunk. "So do that funky merge whiiite boy...""Steven J. Hill"2005-09-1340-1076/+1818
|
* We can now deselect STDIO futex support independently of NPTL."Steven J. Hill"2005-09-061-1/+24
|
* The latest NPTL code should be compiled with '-O2' according to glibc ↵"Steven J. Hill"2005-09-061-2/+2
| | | | makefile comments. We may be able to change this to '-Os' after we get NPTL working.
* Import latest 'clone' and 'vfork' functions from glibc. These have been ↵"Steven J. Hill"2005-09-034-50/+2
| | | | tested with the 'linuxthreads' thread model and normal uClibc using the included test suite. This puts me very close to having static NPTL applications working.
* Add in calls needed for TLS setup for statically linked applications. A call ↵"Steven J. Hill"2005-08-303-1/+77
| | | | is made to '_dl_aux_init' which is defined in 'dl-support.c' to set up TLS variables before '__uClibc_init' gets called.
* Move files up one level where they should be."Steven J. Hill"2005-08-303-0/+0
|
* Fix symbol prefixes for a number of hidden definition in NPTL libpthreads."Steven J. Hill"2005-08-181-2/+2
|
* Rename 'fork.c' to be 'fork-nptl.c' because it ends up replacing the object ↵"Steven J. Hill"2005-08-178-217/+28
| | | | 'libc/sysdeps/linux/common/fork.o' and 'fork' then appears as undefined when applications are linked against uClibc.
* Merge/sync with trunk."Steven J. Hill"2005-08-172-1/+16
|
* It occurred to me that for the uClibc way of doing things, we should be ↵"Steven J. Hill"2005-08-113-0/+101
| | | | using the syscall to fork if pthreads is not being used for NPTL, just like we do with the linuxthreads model. There were some name collisions that had to fixed, but things should be sane now. Also, due to complexities of the new thread model and compilation, the 'fork.c' from NPTL has to be compiled in with the C library, even though the code in it is not called until pthreads is linked in either dynamically by the loader or in a static binary. If that all makes sense to you, give yourself a cookie.
* Steve was taking stupid pills and hardcoded the version of the dynamic loader."Steven J. Hill"2005-08-111-1/+1
|
* Finally, all of the makefiles that make the magic happen. These changes ↵"Steven J. Hill"2005-08-096-136/+451
| | | | allow for uClibc to build for NPTL support without breaking the old linuxthreads model. Let the testing begin.
* Compile cleanups for NPTL support."Steven J. Hill"2005-08-095-5/+38
|
* Changes made to the dynamic loader for TLS/NPTL support. Additional changes ↵"Steven J. Hill"2005-08-0913-25/+184
| | | | made to libpthread and libc for remaining support.
* Die! Die! Die!"Steven J. Hill"2005-08-0925-1435/+0
|
* Properly link in needed TLS functions for the static C library. Added the ↵"Steven J. Hill"2005-07-313-8/+839
| | | | new low-level locking primitives to the C library for both static and shared. Added the new file 'dl-tls.c' from glibc which should be the final code necessary to fully link all the libraries for a NPTL-based uClibc.
* Last of the merges along with latest changes for NPTL. Nothing much"Steven J. Hill"2005-07-303-26/+17
| | | | new on the NPTL stuff other than cleaning a few things up.
* Merge from trunk."Steven J. Hill"2005-07-2310-77/+58
|
* Add generated thread block constant offsets. This was generated when ↵"Steven J. Hill"2005-07-221-0/+3
| | | | building a glibc toolchain. I will create a script to do the same thing in uClibc later on.
* Added new files for TLS functionality in libpthreads and for the dynamic ↵"Steven J. Hill"2005-07-2212-0/+637
| | | | loader. The headers may be moved to the 'ldso' directory hierarchy at a later date depending on what we run into when implementating changes to the loader.
* Big set of various patches to get the NPTL libraries to build and link."Steven J. Hill"2005-07-2220-9/+165
|
* Add in MIPS-specific implementation of 'vfork' and clean up other files in ↵"Steven J. Hill"2005-07-224-1/+36
| | | | NPTL to fit within the uClibc infrastructure.