[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
rtems 19991203 is available
Announcement is attached.
--
Joel Sherrill, Ph.D. Director of Research & Development
joel@OARcorp.com On-Line Applications Research
Ask me about RTEMS: a free RTOS Huntsville AL 35805
Support Available (256) 722-9985
December 3, 1999
RTEMS 19991203 is Available
This is to announce the availability of RTEMS-19991203. Cleanup continues
and I would like to aim for a new public release around February 1.
I would like to begin focusing on tidying up toolset issues and final
patches before a public "pre-release". There are so many RTEMS BSPs
which are only in user's hands, that a public testing period is
the only option.
In particular, the large number of "bolt on" third party packages
has grown tremendously and the FAQ does not reflect most of them.
I am not even sure the prototype release announcement lists them
all. :(
If you know of outstanding problems or can fix one on the list,
please feel free to take care of it. :)
Joel Sherrill
================================================================
I am keeping this as a running list of major enhancements for a
release notice. :)
Major functional enhancements since 4.0.0 include:
+ Addition of ITRON 3.0 API
+ POSIX 1003.1b API support is 85% of full UNIX
+ Addition of remote debug server
+ Better integration with GNU tools
+ RTEMS makes better use of automake and autoconf
+ easier to build other open source packages
- TCL, readline, zlib, ncurses, libavl
- OmniOrb2
- MicroWindows
+ Smaller Application Executables!!!
- Multiprocessing is now a configure time option
- Unreferenced routines are not included in executables.
+ XON/XOFF flow control support added
+ Addition of reuseable peripheral controller library (libchip)
+ New CPU Models Supported
+ PowerPC (mpc821, mpc823, mpc850, mpc860, ppc750)
+ i960 (i960HA, i960RP)
+ Addition of the following BSPs:
- Technologic Systems TS-1325 (i386ex) BSP
- DY-4 DMV177 (PPC603e) BSP
- eth_comm (mpc860) BSP
- Motorola MVME167 (68040) BSP
- Vista SCORE603e (PPC603e) BSP
- Radstone ppcn_60x (PPC603e) BSP
- Motorola MCP750 (PPC750) BSP
- Motorola MVME2307 (PPC604) BSP
- Ramix rxgen960 (i960RP) BSP
+ Addition of file system infrastructure
- In-Memory Filesystem
- TFTP Client Filesystem
- mini In-Memory Filesystem
+ Network functionality
- select() supported for socket
- DHCP support added
- "tap" functionality added
- timeout on sockets supported
- numerous network device drivers
- ftp server
- ntp client
- GoAhead http server
- initial PPP support
+ GNAT/RTEMS
- Ada "self" optimized
- Ada interrupt support
================================================================
Outstanding issues:
+ Layering/Cleanup Issues
+ Be able to build as much of RTEMS as possible with only CPU model
information and plug in a BSP after compiled.
+ score/cpu/i386 default exception handler invokes task_delete. This
violates layering rules and needs to be addressed.
+ librdbg files in BSP specific directories are not BSP specific.
pc386 are OK for all i386 BSPs, mcp750 is really mpc750 CPU model
+ distribute start files out to bsps
+ Support Library Issues
+ TFTP Filesystem handling of bootp server.
+ Port Specific Issues
+ Are the HID registers specific to the PPC603e?
+ finish merging mpc505 changes from Sergei Organov
+ i386 CPU model.S has warnings
+ BSP Specific Issues
+ i386/i386ex: relocation size problem
+ i386/pc386: update grub information
+ i960/cvme961: does not link cdtest
+ i960/rxgen960: link problems
+ m68k BSPs should be checked for .gcc_exc section like dmv152
+ mips64orion: no BSPs link with late model binutils
+ sparc/erc32: tm27 does not work
+ Test Issues
+ pthread cancel test (psxcancel) is a shell
+ POSIX message queue test (psxcancel) is a shell
+ no test of named POSIX semaphores
+ Documentation Issues
+ confdefs.h - add description to list of macros
+ Tool Issues
+ size script is useless at this point in time
+ wildcards on solaris.
+ make/custom/*.cfg that sed on \r (David Sotkowitz <sotko@isr.com>)
+ Ada Issues
+ piwg updates from Juan Zamorano
+ initial signal masked state (Jiri's bug)
+ gnat 3.12p update (first cut at patch included)
19991203
========
RTEMS Port/BSP Independent:
+ Configure errors introduced by ITRON merged fixed (Ralf)
+ POSIX Semaphores. Named semaphore code is now functional although all
error cases have not been tested.
+ libmisc and libchip now use automake. Some comments accompanying
Ralf's patch:
* The basical trick is to wrap an old Makefile.in's contents into a
Makefile.am and still continue to use (i.e include) the old
*.cfg files.
* Replaced each INSTALL_IF_CHANGE and INSTALL_VARIANT with make
dependencies
* Add a gnu-make ifdef AUTOMAKE to main.cfg to avoid conflicts between
automake and RTEMS make rules
* Replaced each install:: and preinstall:: rule with make dependencies
* Replaced SUB_DIRS with SUBDIRS in all Makefile.ins (Automake
convention)
* Removed each manually added autoconf substitution which automake
performs automatically.
This is not yet full automake support, because using the temporary
installation directory, preinstallation in general and building
variants are in contradiction to automake's basic working principles.
+ sptests, samples, and lib/libc now use automake. (Ralf)
+ miscellaneous Makefile cleanups under c/src/lib (Ralf)
+ shmdr: shm.h renamed to shm_driver.h to avoid conflict with
POSIX shm.h (Joel)
+ Converted following score/cpu directories to automake (Ralf):
a29k, hppa1.1, i386, i960, mips64orion, m68k, no_cpu, sh, sparc, unix
+ libchip/dec21140 added (Emmanuel)
Filesystem Changes:
+ none
Networking Changes:
+ PPP Added. This is an initial merger of a submission from "Tomasz
Domin" <dot@diament.infomarket.pl>. An example modem driver is
included as a subdirectory that is not compiled by default.
There is work to be done. Please read the STATUS file.
+ tftpdriver.c: Filesystem structure fixed (Eric N)
+ rtems_bsdnet_ntp.c: Removed debugigng printfs (Eric N)
RTEMS Test Changes:
+ General:
- Added ifndef around build_time macro so tmacros.h and pmacros.h
can be included from the same file. (Joel)
- tmtests: Reworked OPERATION_COUNT override mechanism to be a
configure option (Ralf)
+ psxsem01 - added tests for named semaphores (Jennifer)
Port Specific Change:
+ i386:
- libcpu/cpu.h: in/out port macros fixed to do do/while trick and have
more accurate constraints (Ian and Philip Prindeville)
- libcpu/cpuModel.S: removed warnings (Joel)
+ M68k:
- sim.h: Changed name of W, X, and Y macros for fields in the Clock
Synthesizer Control Register to remove use of single letter names. (Joel)
+ SH:
- Hitach SH-2 support added by John M. Mills <jmills@tga.com> with
merged assistance and cleanup from Ralf Corsepius
<corsepiu@faw.uni-ulm.de>.
- Constraints on the inline assembly language macro for
sh_disable_interrupt corrected (Ralf)
BSP Specific Changes:
+ a29k/portsw
- renamed intr.c to cause_intr.c to avoid conflict with
rtems/src/intr.c (Classic API Interrupt Manager).
+ hppa1.1/simhppa
- renamed intr.c to cause_intr.c to avoid conflict with
rtems/src/intr.c (Classic API Interrupt Manager).
+ i386/pc386
- tools/Makefile.am: Consistently use $(EXEEXT). (Rosimildo)
- dec21140 now used from libchip (Emmanuel)
- start.S/ldsegs.S: Delay loop changed to avoid A20 timing
problems (Aleksey of Quality Quorum)
+ m68k/efi332
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ m68k/efi68k
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ m68k/gen68302
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ m68k/gen68340
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ m68k/gen68360
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ m68k/ods68302
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ powerpc/psim
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ powerpc/mpc750
- Merged MVME2307 BSP with this and modified this BSP such that
it should support the bulk of Motorola PPC boards whether
VME or CompactPCI. This BSP will be renamed MotorolaPPC once
it has been tested and BSP variants added so it is compiled
with the most optimal CPU_CFLAGS (Eric V)
- dec21140 now used from libchip (Emmanuel)
+ powerpc/mvme2307
- See MPC750
+ sh/gensh1
- cleanup to reflect relation to gensh2 (Ralf)
+ sh/gensh2
- New BSP (John Mills)
+ sparc/erc32
- start code is now in a directory named start. bsp_specs updated
and START_BASE override removed from custom file (Joel)
+ unix/posix
- renamed intr.c to cause_intr.c to avoid conflict with
rtems/src/intr.c (Classic API Interrupt Manager).
Tool Changes:
+ RPM Scripts:
- mkgccnewlibspec.in: typo fixed (Joel)
- mkgdbspec.in: now correctly includes powerpc and sparc simulator
files (Joel)
- gdb.spec.in: now supports extra arguments for simulators (Joel)
- gccnewlib_c_only.spec.in: Corrected to remove references to
non-existent subpackages (Joel)
Documentation:
No documentation release but changes have been made.
FAQ:
+ Added section on Projects (Joel)
POSIX API User's Guides:
+ Added Timer Manager chapter (Joel)
19991117
================================================================
Status information for 19991117 (changes since 19991105):
RTEMS Port/BSP Independent:
+ initial ITRON 3.0 API support added (Joel/Jennifer/etc)
- This includes initial support for 5 of the 12 ITRON managers:
+ Task
+ Task-Dependent Synchronization
+ Semaphore
+ Message Buffer
+ Mailbox
- Functional tests the initial 4 managers.
- Timing test for semaphore.
- API Documentation for approximately 1/2 of the API.
+ posix/src/pthreadsetschedparam.c: Added call to _Watchdog_Remove
to remove the sporadic timer. It was being reinserted onto the
ticks chain. (Joel)
+ posix/src/setsid.c: Returns a reasonable error not ENOSYS. (Joel)
+ posix/src/utsname.c: Works and returns reasonable information. (Joel)
+ sapi/include/sptables.h: Fixed so can be included multiple times in
the executive source. (Joel)
+ NTP BOOTP support (Eric N)
+ score/src/threadreset.c: New file. (Jennifer)
+ score/src/threadrestart.c: Use _Thread_Reset (Jennifer/Joel)
+ src/src/threadresume.c: New file added to support nesting
suspend/resume count (Jennifer)
+ src/src/threadsuspend.c: New file added to support nesting
suspend/resume count (Jennifer)
+ src/src/threadrotatequeue.c: New file added to support ITRON
rotate ready queue routine which is not quite the same as yield
is in most specifications. (Jennifer)
+ rtems_semaphore_flush did not enable dispatch (Eric N/Joel)
+ Per Task Variables added. This adds three directives:
rtems_task_variable_initialize, rtems_task_variable_add, and
rtems_task_variable_delete. (Eric N)
+ lib/include/rpc removed since overlooked earlier (Ralf/Joel)
+ lib/include/rdbg removed since overlooked earlier (Ralf/Joel)
+ libc: Added implementations of ctermid(), ttyname(),
and ttyname_r(). (Joel)
+ libc: Added stubs for tcgetprgrp(), tcsendbreak(), tcsetprgrp(). (Joel)
+ Configuration Cleanup (Ralf):
- TARGET_ARCH removed
- target.cfg.in moved to c/make/target.cfg.in (Only configured once for
all BSPs of a target)
- BARE_XXX variables appended to bsp.cfg.in
- autogen renamed to bootstrap
- removed stray variables from make/custom/*.cfg
+ No lib source files other than in bsp include bsp.h (Ralf/Joel)
+ No Makefiles reach up and over to install BSP header files (Ralf/Joel)
+ rtems_webserver is dependent on having the POSIX API configured
but was built even when the POSIX API was disabled. (Joel)
+ rtems_webserver/uemf.h: Conditionally take out the offensive pragma
pack when on RTEMS in addition to UnixWare (UW). (Joel)
Filesystem Changes:
+ UNIX/POSIX target did not include all files needed to successfully link
(Ralf/Joel)
+ scandir.c: Arguments to strncpy() were swapped. (Jennifer)
RTEMS Test Changes:
+ tmacros.h - Added code to the macros which checked directive status
to also check that the _Thread_Dispatch_disable_level is set to the
proper value (0 99% of the time). This automatic check significantly
reduces the chance of mismatching disable/enable dispatch pairs while
doing internal RTEMS work. (Joel/Jennifer)
+ Samples:
+ cdtest - made class names longer to avoid conflicts. (Joel)
+ hello - Uses the miniIMFS to reduce code space. (Joel)
+ paranoia - Changed B, H, and W variables to BVar, HVar, and WVar to
avoid conflicts with ITRON 3.0 data types regardless of how dumb it
is to name types this way. (Joel)
+ ITRON Tests
- new suite with the following tests:
+ itronhello - ITRON Hello World
+ itronmbf01 - Message Buffer Test
+ itronmbox01 - Mailbox Test (stub only)
+ itronsem01 - Semaphore Test
+ itrontask01 - Basic Tasking (ITRON ticker)
+ itrontask02 - Tasking Error Cases
+ itrontask03 - Basic Tasking
+ itrontask04 - Suspend/Resume Test
+ itrontime01 - Time Test (stub only)
+ ITRON Timing Tests
- new suite with the following tests:
+ tmitronsem01 - First Semaphore Timing Test
+ POSIX Tests:
+ psx01 - Added test code for uname(). (Joel)
+ psxfile01 - Clean up, better matching screen (Jennifer)
+ psxmsgq01 - Shell of new test for POSIX Message Queues
+ psxreaddir - Clean up, better matching screen (Jennifer)
+ psxsem01 - New test for POSIX Semaphores
+ psxstat - Clean up, better matching screen (Jennifer)
+ Sptests:
+ sp27 - Test code for rtems_semaphore_flush added (Eric N/Joel)
+ sp28 - Test code for per task variables added (Eric N)
+ tm26 - nesting count for _Thread_Resume added (Joel)
+ Tmtests:
+ timesys.h: Added BSP_MAXIMUM_OPERATION_COUNT and BSP_ITERATION_COUNT
which allow the BSP to override the default number of objects
that will be created in the timing tests. This is useful for
reducing memory consumption on small targets. The BSP_ITERATION_COUNT
can be used to reduce the number of repititions in some of the
tests. This is useful for extremely slow targets (we haven't seen
any slow enough to justify modifying this one yet though. :) (Joel)
+ Network Demos:
+ netdemo/ntp - New netdemo test for NTP (Eric N)
+ netdemo/networkconfig.h - new NTP field added (Eric N)
+ netdemo/netlink - New test to use to generate code space
estimates (Joel)
Port Specific Change:
+ libcpu/powerpc cleaned up odd Makefile references.
+ tools/sh: Cleanup in prepartion for SH2 merger (Ralf)
+ unix
- System V IPC usage now conditional on multiprocessing enabled.
score/cpu/configure.in should now only check for System V IPC
when multiprocessing is enabled. This should let the unix
port work on cygwin without multiprocessing. It may also
let it work easier on BSD-ish systems.
BSP Specific Changes:
+ General
- start code (lib/start) distributed out to appropriate locations.
If used by only one BSP, then the start code is now part of that
BSP. If it is used by multiple BSPs as in the case of the m68k,
then it was places in libbsp/CPU/shared and a start directory
was added to each BSP using it to compile it. (Joel)
+ powerpc/ppcn_60x
- console now properly uses libchip rather than own local precursor
to libchip (Ralf/Joel)
+ powerpc/mcp750
- BOOTING: new file on how to boot the produced executables (Eric V)
+ powerpc/score603e
- linkcmds: Now allows override of base address via -Ttext
ld argument (Joel)
Tools/Scripts:
+ RPMs
- gdb.spec.in install info files gzip'ed, add gdb to group rtems (Ralf)
Documentation:
Overall:
+ new ITRON 3.0 API User's Guide
C and Ada User's Guides:
+ Added categorized list of constants in confdefs.h. (Joel)
+ Added description of methods to obtain the pieces of an object
ID (rtems_get_class, rtems_get_node, and rtems_get_index) (Joel)
+ Added documentation for new directives rtems_task_variable_add
and rtems_task_variable_delete (Eric N/Joel)
Network Users Guide:
+ Added NTP documentation (Eric N/Joel)
POSIX API User's Guide:
+ new implementation status chapter added (Joel)
+ many minor spacing problems fixed (Joel)
19991105
================================================================
Status information for 19991105 (changes since 19991028):
Testing:
Limited to sparc/erc32. Compiled over past week on numerous
targets.
RTEMS Port/BSP Independent:
+ POSIX message queues and semaphores updated to almost working point.
+ mutex.c: removed warning for unused variables
+ pthreadexit.c: added assert in case _Objects_Get_information fails
since this should NEVER occur.
+ rtems/src/semtranslatereturncode.c: Added case for the new core
semaphore error CORE_SEMAPHORE_MAXIMUM_COUNT_EXCEEDED.
+ rtems/src/tasks.c: Added comments for _RTEMS_tasks_Create_extension,
_RTEMS_tasks_Start_extension, _RTEMS_tasks_Delete_extension, and
_RTEMS_tasks_Switch_extension.
+ rtems/src/tasksetpriority.c: Added comment indicating a routine
is needed to convert classic api priority to core and vice-versa.
+ confdefs.h: added posix semaphores and message queues.
+ posixapi.c: added posix semaphores and message queues.
+ coremsg.h: Added support for message priority as required by POSIX.
+ coresem.h: Added maximum count detection logic.
+ object.c: Added support for ITRON style IDs.
+ threadchangepriority.c: Added a comment to indicate that if a task
blocked on a priority based thread queue were to be reinserted
based on its new priority, this is where the logica would go.
+ coresem.c: Added code to support maximum count checking
required by POSIX semaphores.
+ object.c: Fixed bug which resulted in misaligned object control
structures being allocated. This bug manifested itself on the SPARC
since it uses ldd/std in its context save/restore routines.
+ Split the following files to reduce minimum code size:
- coremsg.c (SuperCore Message Queue Handler)
- coremutex.c (SuperCore Mutex Handler)
- heap.c (SuperCore Heap Handler)
- object.c (SuperCore Object Handler)
- threadq.c (SuperCore Thread Queue Handler)
- tod.c (SuperCore Time of Day Handler)
- watchdog.c (SuperCore Watchdog Handler)
- cancel.c (POSIX Cancellation Manager)
- cond.c (POSIX Condition Variable Manager)
- key.c (POSIX Key Manager)
- mqueue.c (POSIX Message Queue Manager)
- psignal.c (POSIX Signals Manager)
- mutex.c (POSIX Mutex Manager)
- semaphore.c (POSIX Semaphore Manager)
- unistd.c (sysconf() and sleep())
Filesystem Changes:
+ imfs_fsunmount.c: Properly initialize location variable.
+ imfs_initsupp.c: Fixed typo where memfile_handlers set twice and
directory_handlers was not set.
+ readdir.c: Check for NULL being passed in.
+ rewinddir.c: Check for NULL being passed in.
+ seekdir.c: Check for NULL being passed in.
+ telldir.c: Check for NULL being passed in.
+ Makefile.in: Install imfs.h so base filesystem can be configured
or IMFS mounted by an application.
+ base_fs.c: Added support for configuring base filesystem. First
attempt at a full mount table at initialization. This is the
wrong place to mount filesystems 2-n since no device drivers are
initialized yet.
+ fcntl.c: Fixed F_SETFL to properly convert between internal
libio flags and external fcntl.h style flags.
+ The object memfile.o was being included in the miniIMFS even though it
should not have been. This required that IMFS_rmnod be split into
three separate (per file type) routines to avoid dependencies.
In the end, a miniIMFS application is 6K smaller than one using the
full IMFS.
RTEMS Test Changes:
+ sp01 - increased stack space for task 1
+ sp03 - increased stack space for a task
+ sp13 - increased stack space of task 2.
+ psxfile01: Minor fixes to make screen match a bit better.
+ psxmount: Minor fixes to make screen match a bit better.
+ psxreaddir: Added more proper checking of F_GETFL and F_SETFL.
Disabled test of F_DUPFD since it does not work.
Added comment to indicate that seekdir(NULL) was being called.
+ ticker - increased stack space of all tasks
+ unlimited - use minimum stack size and reserve enough workspace
to allocate the minimum number of tasks necessary to thoroughly
test the unlimited objects feature.
Port Specific Change:
+ libcpu/powerpc/wrapup/Makefile.in - wildcard rule for CPU model
dependent pieces was wrong.
+ libcpu/powerpc/mcp750/clock - more BSP independent. (Eric V)
+ libcpu/*
- modified to use RTEMS macros to obtain configuration
information.
BSP Specific Changes:
+ m68k/gen68360
- custom file no longer references stack checker (Joel)
+ powerpc/eth_comm
- custom file no longer references stack checker (Joel)
+ powerpc/mcp750
- custom file no longer references stack checker (Joel)
- clock - BSP dependent support for libcpu clock. (Eric V)
+ powerpc/mvme2307
- removed dummy shmsupp code and fixed mvme2307.cfg file to follow
current conventions. (Ralf/Joel)
+ powerpc/score603e
- split console_reserve_resources into its own file to reduce
dependencies (Joel)
- custom file no longer references stack checker (Joel)
+ sh/gensh1
- typo in gensh1 fixed (Ralf)
Script Changes:
+ RPM Scripts
- buildall: cleans up itself better
- gdb.spec.in: fixed typo
- gccnewlib_c_only.spec.in: new file to allow building C/C++ only RPMs
when the other languages are not wanted or don't build. This
is needed on the i960 right now.
- gnatnewlib directory: first attempt at the beginnings of a
gnatrtems RPM set.
Documentation: No changes.
================================================================
Status information for 19991028 (changes since 19991011):
Testing: nearly all target compile, simulators ran but not directly before
snapshot was cut. See problems for specific issues.
Tools:
RPM: BIG NEWS!! See scripts at top directory where Ralf and Joel have
been working on preparing RPMs. These are the specs and scripts
to build binutils, gcc/newlib, gdb, and RTEMS BSP RPMs.
gcc: new patch to support RPM
binutils: new patch to support RPM
newlib: new patch to support RPM
Documentation:
- nearly eliminated line too long (too wide) problems
FAQ:
- miscellaneous improvement (PLEASE READ AND REVIEW!!!)
C and Ada User's Guides:
- added rtems_semaphore_flush
- added rtems_task_is_suspended
Network Users Guide
- swapped SO_SNDWAKEUP and SO_RCVWAKEUP since they were backwards.
BSP Howto
- Ada95 Interrupt Support chapter was not in master document
Filesystem Guide
- information on freenode and mounting added
Ada-examples: NO CHANGES
Network Demos:
+ Added http sample tests.
RTEMS Port/BSP Independent:
+ Added port of GoAhead server (Emmanuel and Eric V)
+ add prototype for new rtems_task_is_suspended directive (Eric N/Joel)
+ posix/src/ptimer.c - Fixed core dump caused when output
parameter (otimer) was NULL. (Joel)
+ confdefs.h - Renamed CONFIGURE_POSIX_INIT_TASK_STACK_SIZE to
CONFIGURE_POSIX_INIT_THREAD_STACK_SIZE. (Joel)
+ added rtems_semaphore_flush (Eric N/Joel)
+ Initialize the pthread cancellation fields (Charles-Antoine Gauthier)
+ netinet/ip_input.c: DIAGNOSTIC code which unnecessarily paniced
disabled (Philip Prindeville/Eric N)
+ c/src/configure.in, c/src/Makefile.am: minor configuration bug (Ralf)
+ moved c/src/lib/libmisc to c/src/libmisc (Ralf)
+ moved c/src/lib/libchip to c/src/libchip (Ralf)
+ removed gcc.cfg.in and all references to HOST_ARCH (Joel after
discussions with Ralf and Eric N)
+ dummy.rel is now in libmisc. This is a "magic file" which
provides Init and Configuration. This was written to make
it easier to build autoconf'ed packages with RTEMS. (Ralf)
Filesystem Changes:
+ added logic to free filesystem specific nodes which were allocated
as a side-effect of evaluating a path (Jennifer)
+ imfs_debug.c : modified format of printing IMFS directory (Joel)
+ mount.c/unmount.c: Added code to correctly free any allocated space
during the path evaluation process of mounting and unmounting. (Jennifer)
+ imgs_gtkn.c: Modified to use local variable (Eric V/Emmanuel)
+ miniIMFS added, no testing, no way to configure as base
filesystem without editting base_fs.c. The next step
is a mount table configuration scheme so the application
can select the IMFS or miniIMFS as base filesystem.
The miniIMFS is comparable to the old 3.6.0 functionality
and reduces executable size. (Joel and Jennifer)
RTEMS Test Changes:
+ psx01 - Modified to ease comparison of screen to target output. (Joel)
+ psx04 - increased stack space (Joel)
+ psx05 - removed reinitialize mutex test case to reflect earlier
code change (Joel)
+ psxstat - Modified to ease comparison of screen to target output. (Joel)
+ sp26 - new test for rtems_task_is_suspended (Eric N/Joel)
+ sp27 - notes for new test (Eric N/Joel)
+ psxcancel - shell for pthread cancel test (Joel)
Port Specific Changes:
+ i960
- leave dispatching disabled until context switch finished
(Gerwin Pfab <pb@schenk.isar.de>)
+ powerpc
- minimum stack size increased to 8K
+ sh
- spin delay routine now in gensh1 BSP (John/Ralf)
BSP Specific Changes:
- i386/i386ex
+ start.S: fixes to address reported problems (Erik I)
- i386/pc386
+ ldsegs.S/linkcmds: proper Multiboot support (Erik I)
WARNING!!!!!MAY BREAK GDB'S ABILITY TO READ EXECUTABLES!!!
+ librdbg Makefile.in: added missing file reference (Emmanuel)
+ dec21140.c: increased RBUF_SIZE to 1536 (Emmanuel)
+ tools/Makefile.in: $(EXEEXT) for cygwin problem (Rosimildo/Ralf/Joel)
+ console/console.c: allow uart to be printk console (Aleksey)
+ startup/bspstart.c: make bsp_start a weak alias (Aleksey)
- i960/rxgen960
+ new i960RP BSP submitted by Ramix
- m68k/gen68360
+ console.c: split console_reserve_resources into its
own file so minimum application size is reduced. (Joel)
- powerpc/mcp750
+ dec21140.c: fixed bugs (Eric V/Emmanuel)
+ custom file: compiler arguments changed.
- sh/gensh1:
+ spin delay routine now here not in score/cpu (Ralf)
- sparc/erc32:
+ runtest script modified to produce correct output with new tools
================================================================
Status information for 19990820 (changes since 19990709):
Testing: nearly all target compile, simulators run
- i960/cvme961 does not link C++
- mips64orion/* have weird link errors
Tools: (new patches for everything)
gcc:
+ Added NO_IMPLICIT_EXTERN_C to all RTEMS configurations
so librtems++ would compile (Ralf)
binutils:
+ new patch for binutils snapshots
- includes i960-elf
newlib:
- net newlib 1.8.2
Documentation:
+ new release of documentation
+ Lots of formatting changes, two new manuals, see separate announcement
to the mailing list.
Ada-examples: NO CHANGES
Network Demos: NO CHANGES
RTEMS Source:
+ Miscellaneous automake/autoconf improvements from Ralf
- Addition of maintainer-mode dependencies on config.status and
aclocal.m4 for autoconf toplevel Makefile.ins. These rules are taken
over from automake generated Makefile.ins, i.e. they are contained in
any automake generated toplevel Makefile.in.
- Automated support for above in acpolish
- Some minor "beautifications" on Makefile.ins resulting from running
acpolish.
- some files used in AC_INIT were not unique enough and can lead to
problems if a user plays with configure scripts.
- the Makefile templates are independent of the target/cpu and bsp, so
the bsp dependent versions (c/src/make/Templates) are removed.
- moved the following directories to the top of the tree and turned
then into their own packages
+ c/src/lib/librtems++ --> librtems++
+ c/src/lib/include/rtems++ --> librtems++
+ c/src/lib/librpc --> librpc
+ c/src/lib/librdbg --> librdbg
+ c/src/lib/libnetworking --> libnetworking
+ c/src/lib/wrapup --> wrapup
- 14 new configure scripts
- 53 Makefile.ams for most cpu and bsp independent directory and include
directory Makefiles.
- removes some wrap directories
- Uses new dependency rules for installing files (esp headers and
directories) into the temporary installation tree (Note: Most of these
rules do not use INSTALL_CHANGE anymore, but apply INSTALL_DATA and make
dependencies instead.
- Fixed Makefile.am's so they use "##" to avoid propagating junk
like CVS Ids into generated files.
+ This should be unnoticeable but the following are no longer
kept under CVS but automatically generated as part of the
release procedure:
- Makefile.in's from Makefile.am
- configure's from configure.in
- aclocal.m4
The frequent regeneration was creating bogus CVS activity.
+ libchip ns16550.c - invalid value used for baud rate per bug report from
Jay Kulpinski.
+ libcpu powerpc build order cleaned up (Joel)
+ Removed Makefile.in's autogenerated from Makefile.am from CVS (Joel)
+ Fixed panic when the loopback interface was included as part of the
network initialation structures. With the printf you get an
message, but the interface is still properly initialized. (Eric N)
+ added stubs for tcflow and tcdrain. added more termios
functionality as required by GNU readline (Eric Norum)
+ added support for envp to shared bootcard (Joel)
+ correctly handle POSIX threads that simply return from function
body (return value becomes arg to pthread_exit) (Charles-Antoine Gauthier)
+ include/networking at install point removed and compilation
command lines simplified (Joel)
+ Readd the behavior where the minor number indicated the port number to try.
(Eric Norum <eric@cls.usask.ca>)
+ Minor librdbg Makefile fixes (Ralf)
+ Minor librdbg fix on PowerPC to use sync instruction (Eric V)
+ Readded RTEMS_BSP to Makefile.inc.in (Ralf/Eric N)
+ Removed the following empty files:
c/src/lib/libbsp/powerpc/mcp750/startup/cEntry.c
c/src/lib/libcpu/powerpc/shared/processor.h
c/src/lib/librdbg/i386/excep.c
+ Added isatty.c and creat.c to libc to avoid duplications with
newlib posix_dir support (Charles-Antoine Gauthier)
+ eval now stops on error from filesystem specific evalpath
(Wayne Bullaughey)
+ Commented out stdlin.h include in the following files to avoid
problems caused by the upgrade to newlib 1.8.2 and the movement
of the prototype for malloc() to stdlib.h
- c/src/lib/libbsp/powerpc/mvme2307/network/network.c
- c/src/libnetworking/rtems/rtems_select.c
- c/src/libnetworking/rtems/rtems_showroute.c
- c/src/libnetworking/rtems/rtems_syscall.c
- added new rtems_task_is_suspended directive (Joel)
- Commented out cygwin specific tests in configure scripts which
are not supposed to be needed any longer. (Ralf on David F comments).
+ Tests
- psx08 - enhanced to test exitting from a pthread (Charles-Antione Gauthier)
+ added putenvtest to check putenv/getenv functionality (Joel)
+ Processor Specific Changes
- sh/start.S had a nasty typo caught only by recent binutils (Ralf)
- Typo in i960 score/cpu for CA cpu model (Ralf/Joel)
- unix port needed another call to sigemptyset (Jay Kulpinski/Ian/Joel)
+ BSP Specific Changes
- i386/force386
- REMOVED
- i386/go32
- REMOVED along with all references in score/cpu/i386
- m68k/* - MOST BSPs CHANGED
- Added sections, eliminated overlapping section errors in linkcmds (joel)
- m68k/mvme167:
- elflinkcmds was not in tree (Joel/Charles-Antoine Gauthier)
- wrapup/Makefile.in listed fpsp.rel twice (Joel)
- m68k/gen68340:
- start340.S - changed bra.s to bra.l for new binutils (Joel)
- m68k/gen68360:
- console initialization only done once (Eric Norum)
- start code sets up argv and envp (Eric Norum)
- start360.S - changed bra.s to bra.l for new binutils (Joel)
- powerpc/* - SOME BSPs CHANGED
- Added sections, eliminated overlapping section errors in linkcmds (joel)
- powerpc/mvme2307:
- new BSP from Jay Kulpinski (Thanks!)
- powerpc/mcp750:
- minor fix to irq_asm.S (Eric V)
- custom file: Added $(LINK_LIBS) to compile command (Joel)
- powerpc/ppcn_60x:
- startup/Makefile.in was missing files (Joel)
- powerpc/score603e:
- now picks board generation using ifdef's not Makefile magic (Joel)
- startup/Makefile.in was missing files (Joel)
================================================================
Status information for 19990820 (changes since 19990709):
Testing: all target compile, simulators run
Tools: (new patches for everything)
+ gcc 2.95.1
- new patch, transition to elf, numerous people
+ binutils 2.9.1
- new patch, transition to elf
- PowerPC objdump fix from Ian Lance Taylor
+ gdb 4.18
- new patch including
- PowerPC remote debug server from Eric V and Emmanuel
- m68k elf and MVME167 monitor support. (Charles-Antione Gauthier)
+ gnat-3.11p
- recognize *-rtemself, *-rtemscoff, *-rtemsaout (Tony A)
- patch gdb.*/ada-tasks.c to use RTEMS_TARGET
+ build scripts
- Add -DRTEMS_TARGET to bit_gdb
- numerous minor cleanups
Documentation:
+ new release of documentation
+ most changes are to the FAQ
Ada-examples:
+ new collection of Ada examples for GNAT/RTEMS
- need to add Ada sockets binding
- need to add Florist instructions
- will eventually supercede hello_world_ada
Network Demos:
+ TFTP test now always gives proper "dotted decimal" bootp host (Eric N)
RTEMS Source:
+ "the big patch" which converted lots of RTEMS tree to automake,
moved it to more GNU compliance and sets the stage for eventual
separation of BSPs from main tree. This description really
does not do the description from Ralf justice but the list of
small changes is long (Ralf).
- numerous non GNU make stanzas now obsolete
- configure.in scripts more distributed in tree
- rcs stanzas and rcs-clean removed
- corrected autoconf checks for system V IPC
- UNIX BSPs now work with glibc 2.1. One serious problem revolved
around not installing RTEMS' sys/termios.h and setting -D_BSD_SOURCE
- make depend should now work.
- RDBG should be properly handled now
- much much more
+ powerpc/mcp750
- mcp750 irq.c typo reported by Jay Kulpinski
<jskulpin@eng01.gdds.com> (Eric V)
- dec21140 driver code arious bug fixes (Emmanuel)
- bug in the mcp750 init code have been fixed (interrupt stack/initial
stack initialization), BSS correctly cleared (Eric V)
- remote debugging over TCP/IP is nearly complete (berakpoints,
backtrace, variables,...) (Eric V)
- exception handling code has also been improved in order to fully
support RDBG requirements (Eric V)
+ sparc/erc32 BSP
- remote debugging fixed (Jiri)
- erc32 now supports C++ exceptions (Joel)
+ partially merged mpc505 support from Sergei Organov <osv@Javad.RU>
libcpu is merged but score/cpu/powerpc is not. His work was based
on 3.6.0 and was difficult to merge.
+ fixed i386 shared interrupt initialization typo and made i8259s_cache
only accessed from C. (Eric V)
+ added addr32 directive per suggestion from Ian to fix problem
reported by Ralf (Joel)
+ CPU_CPU_HAS_OWN_HOST_TO_NETWORK_ROUTINES typo (Charles-Antoine Gauthier/Joel)
+ deep copy bug for BSP_Configuration which resulted in the
RTEMS and POSIX API configuration tables getting incremented
each time the program was rerun (Charles-Antoine Gauthier/Joel)
+ pthread initialization of cpu time budget (Peter Pointer/Joel)
+ MVME167/m68k ELF patch from Charles-Antoine Gauthier
<charles.gauthier@iit.nrc.ca>
+ P90 0x80 vs. 0xED port pc386 bug report from David Decotigny
<David.Decotigny@irisa.fr>. (Eric V)
+ Change of ~ to ! in definition of CONSOLE_USE_POLLED (Jay Kulpinski)
+ init/fini calls readded to bootcard(). (Rosimildo)
+ tftp FS now requires proper IP bootp host information (Eric N)
+ i960 soft reset patch from Jimen Ching <jimen@adtech-inc.com>
================================================================
Status information for 19990709 (changes since 19990528):
Testing: nearly all target compile, simulators run
Tools:
+ Numerous changes to switch to ELF. Notes are:
Across all tools:
+ added sparc-rtemself
+ added sparc-rtemsaout as alternate name for current sparc-rtems
+ switched sparc-rtems to ELF from a.out
+ added i386-rtemscoff as alternate name for current i386-rtems
+ switched i386-rtems to ELF
+ added sh-rtemscoff
binutils only:
+ added m68k-rtemself and m68k-rtemscoff
+ added i960-rtemscoff
tool testing:
+ reduced active testing set to "CPU-rtems" plus sh-rtemself
Documentation:
+ Documentation available in PDF, PostScript, INFO, HTML, and source
+ Added information on confdefs.h to Configuring a System chapter (Joel)
+ Added section to Network manual on SO_SNDWAKEUP and SO_RCVWAKEUP (Ian)
+ Added section on Network driver Makefiles
+ Added section on DEC Ethernet boards
+ Added pathname evaluation section to POSIX User's Guide
+ Filled in POSIX scheduler chapter
+ Filled in POSIX mutex chapter
+ Filled in POSIX threads chapter
+ Added first cut at FAQ
+ Added test on configuring objects with unlimited attribute.
+ Redid top level HTML file.
Network Demos:
+ Added echoServer2 to select test to test interrupt-ish driven
socket IO (Ian)
+ STATUS updated (Ian/Joel)
+ VERSION file generated automatically now (Ian/Joel)
RTEMS Source:
+ autogen now run on image after exported from CVS (Joel)
+ untar README added (Jake)
+ PowerPC cleanup patch (Eric V)
- same interrupt disable issue as on i386
- removed compiler warnings
- removed libc routines
+ erc32 fixes -- FPU Rev B workaround, interrupt masking, etc. (Jiri)
+ added POSIX routines that manipulate /etc/passwd and /etc/group.
This is Section 9 -- System Databases -- of POSIX 1003.1b and
eases building TCL 8.0 for RTEMS. (Ralf/Joel)
+ pid == 0 in sched_rr_get_interval now valid (Joel)
+ added c/src/exec/posix/src/signal.c -- signal(2) (Joel/Ralf)
+ libcpu/powerpc/mpc860 Makefile's reworked to compile after mpc750
merger (Joel)
+ erc32 gnatsupp changed to ignore console interrupts. Otherwise
console interrupts were Ada exceptions (Joel with advice from Jiri)
+ sparc assembly code and linkcmds -- minor fixes to build with ELF.
Basically SYM() had to be used reliably to account for underscores. (Joel)
+ i386/shared/irq_init.c enabled interrupts too early (Eric on bug
report from David.Decotigny@irisa.fr)
+ other include Makefile.in's did not use INSTALL_CHANGE (Ian/Joel)
+ score/include/rtems/score/Makefile.in used INSTALL_DATA not
INSTALL_CHANGE (Ian)
+ some configure.in's missed substituting RTEMS_CPU (Joel/Ralf)
+ Various problem reports closed:
- ts_386ex BSP builds
- there is an acknowledgement for ts_386ex BSP
+ mount() calling sequence changed to avoid need for strings (Gumby/Joel)
+ C++ exception patch to main.c and pc386 bsp_specs, linkcmds (Rosimildo)
+ getdents() warning removed (Gumby)
+ patch to clean up no_cpu/no_bsp configuration (Ralf)
+ ts_386ex.cfg uses $(PACKHEX) instead of $(PROJECT_TOOLS)/packhex (Ralf)
+ score603e.cfg uses $(PACKHEX) instead of $(PROJECT_TOOLS)/packhex (Ralf)
+ score603e.cfg uses powerpc instead of ppc (Ralf)
+ MKDIR variable removed (Ralf).
+ Bug in nearly every cpu/wrap/Makefile.in where rtems.o was trying to
get installed from two different directories (Joel)
+ Bug in rtems_select.c reported" "sbwait sets SB_WAIT in sb_flags.
sowakeup checks it. Why doesn't socket_select set it?" (Ian/Eric N)
+ rtems_task_set_priority now returns RTEMS_INVALID_ADDRESS when
the old_priority parameter is NULL. (Joel)
+ no-ctor removed from lib/wrapup. Now grabbed via libbsp.a (Ian)
+ install in a loop in c/build-tools (Ralf)
+ get_file_system_options in mount.c now uses strcasecmp not strupr
since the latter is not portable at all (Gumby).
+ PowerPC update from Eric V. and Emmanuel including mpc750 support
and mcp750 BSP.
================================================================
Status information for 19990423 (changes since 19990407):
Testing: all target compile, simulators run
Tools:
+ new egcs patch
- mips C++ changes (Catherine Moore + Joel)
- i386 C++ changes (Rosimildo DaSilva)
+ new newlib patch
- mprec.c patch (Jay Kulpinski)
- rtems/sys/time.h properly bracketed (Joel)
+ gdb 4.18 patch ... coming soon
Documentation: no changes
RTEMS Source:
+ halt i386 in default exception handler instead of hard loop (Gumby)
+ alignment of memory allocated for mbufs fixed (Jay Kulpinski)
+ panic when out of mbufs removed (Eric N)
+ mips64orion _CPU_ISR_Set_level wiped out level argument
(reported and fixed by Daniel Kelley <dank@icube.com>)
+ Worked on reducing size of minimum non-POSIX RTEMS application (Joel)
- moved _Entry_Table from sptables.h to its own file. sptables.h
is included from exinit.o and thuse forced every app to have
every directive.
- rtems.o is not included in rtems-cpu.rel
- split core Thread Handler into one routine per file
- split ALL Classic API Managers into one routine per file
- size_rtems disabled until file splitting stabilizes
- adding .o's and not .rel's to ALL libbsp.a files (wrapup/Makefile.in)
- split erc32 console up to reduce the dependencies [This one may
have to be done to all BSPs. Hold off, I may be able to make this
change in another way that requires less work.]
+ Added RTEMS_REGION_FREE_SHRED_PATTERN to invalidate memory
when freed (Gumby)
+ Add make-cxx-exe rule for pc386 BSP (Rosimildo DaSilva)
+ cdtest now includes exception test (Rosimildo DaSilva)
+ Added 3C509 driver to pc386 BSP (Rosimildo DaSilva)
+ Removed check for initialized POSIX mutex ID to eliminate random failures
in his port of omniORB2 (Joel on report from Rosimildo DaSilva).
+ Removed -fomit-frame-pointer from all i386 BSPs (Joel on
report from Rosimildo DaSilva)
+ libc_wrapup() no longer cleans the per task reentrancy structure
(Joel on report from Jay Kulpinski)
+ added nops to i386ex start code to correct 32 bit jmp relative
offset from .reset section (Erik I)
+ _Thread_Yield now consistently wrapped by dispatch calls (Joel)
+ can now configure stack size of POSIX Init thread (Jiri)
+ mpc821/mpc860 calculated PIT wrong in clock driver (Gunter Magin)
+ more extensions recongized on C++ rules (Joel)
+ tftp filesystem reports errors on open correctly (Jennifer)
+ bug fix in network unit name/number parsing fixed (Eric N)
+ Caching coherency conditional added to pc386/dec21140 (Emmanuel)
================================================================
Status information for 19990426 (changes since 19990423):
19990423 snapshot image was broken. This snapshot replaces
the RTEMS .tgz and diff files.
================================================================
Status information for 19990423 (changes since 19990407):
Testing: minimal
Tools:
+ new egcs patch
+ new newlib patch
Documentation: updated (19990423)
+ added unlimited object information to User's Guide (Chris/Joel)
+ added Ada interrupt information to BSP Howto (Jiri/Joel)
+ Network manual now explains -D options better (Eric N/Joel)
+ added first cut at FAQ (Joel)
RTEMS Source:
+ yield/flash problem report CLOSED. The problem had subsequently been
fixed in 4.0.
+ BSP for TS-1325 (i386ex) from Technologic Systems.
(www.t-systems.com) submitted by Tony R. Ambardar <tonya@ece.ubc.ca>.
+ Added lstat() at request of Erik Ivanenko (Jennifer/Joel).
+ Added C++ wrappers to all top level POSIX API header files.
Reported by Rosimildo DaSilva <rdasilva@connecttel.com> who
included a patch for some of the files. Rest by Joel.
+ added __INSIDE_RTEMS_BSD_TCPIP_STACK__ to simplify
Makefile changes for drivers (Eric N/Joel)
+ i960ha support added (Jimen Ching <jimen@adtech-inc.com>)
+ Unlimited objects design information added (Chris Johns)
+ ioctl() prototype changed in networking code (Joel/Erik I)
+ c/src/lib/libcpu/i386/cpu.h warning removed (Erik I)
+ Fix make depend since it is a non-standard automake target (Ralf)
+ Reorganized build-tools directory to ease future use of automake (Ralf)
+ Add scitab.c to CLEAN_ADDITIONS in
c/src/lib/libbsp/sh/gensh1/scitab/Makefile.in (Ralf)
+ Replace some variables in Makefile.in to ease parsing
Makefile.in's to east automatic generation in future (Ralf)
+ miscellaneous rgdb configuration problems such as
where a BSP not supporting rgdb on a CPU that did (i386ex)
would still attempt to build it (Ralf)
+ --enable-gmake-print configure option removed (Ralf)
+ i386ex/startup/Makefile.in installed empty variable name (Ralf)
+ go32/startup/Makefile.in installed something to the void (Ralf)
+ p4600 linkcmds changes to match tool changes
(Joel with much help from the egcs world :)
+ minor psim linkcmds corrections (Jay Kulpinski)
+ untar.c did not include <sys/stat.h> (Joel/Ralf)
+ papyrus had wrong register in bss clear loop (Brendan Simon/Joel)
+ mpc860 and mpc821 had a bad buffer assignment (Joel based
on bug report from Nick Simon and Brendan Simon)
+ Added statement to include build-tools/src directory in
preinstall. (Joel)
+ --enable-tests not honored (Ralf/Joel)
+ More automake prep (Ralf)
================================================================
Status information for 19990407 (changes since 19990331):
Testing: minmal
Tools: no changes
Documentation: update coming soon
RTEMS Source:
+ mpc821 support added to score/cpu/powerpc and libcpu/powerpc (Andy Bray)
+ mpc860 was missing some interrupts (Andy Bray)
+ Cleanup in ppc.h (Andy Bray)
+ PPC bsp_specs now treat ecrtn.o as endfile (Andy Bray/Joel)
+ Patch to make the m360 structure available to the debugger (Eric N)
+ Tests not disabled by default (Eric N/Joel)
+ ftpd needed to use htons() on sin_port (Emmanuel)
+ untar added to libmisc (Jake)
+ ftpd patch to cleanup naming as well as add a configuration
table (Jake)
+ ftpd documentation added to Networking Manual (Jake/Joel)
+ psim and erc32 now install BSP specific tools (Ralf)
+ SPARC in_cksum_hdr turned into inline routine like others (Joel)
Network Demos:
+ ttcp changes to take advantage of new routines (Eric N)
+ networkconfig.h changed to be bogus information (Joel/Eric N)
================================================================
Status information for 19990331 (changes since 19990302):
Testing: More rigorous than 19990210. Most targets build.
Some warnings removed. Tests run on simulator.
Tools:
+ egcs-1.1.2-rtems-diff-19990331
- limits.h handling (Ian)
Documentation:
+ I have a queue of stuff to merge.
Code:
+ XON/XOFF flow control support (Thomas Doerfler)
+ m68k/rtems.S would not assemble for m68000 targets (Joel)
+ unlimited objects patch so sp09 gets the proper
error when attempting rtems_port_delete(0) (Chris)
+ NE2000 driver patch to add byte wide interface support (Tony Ambardar)
+ shutdown TCP/IP stack routine added (Tony Ambardar)
+ Fixed some POSIX file tests to account for correct error codes,
new maximum file length (difference between IMFS and global max).
Report from Emmanuel (Jennifer).
+ pc386 bin2boot did not use binary mode on file
(Rosimildo DaSilva <rdasilva@connecttel.com>)
+ SONIC driver modifications to use new driver name parsing scheme (Eric N)
+ SONIC driver modifications after port to SPARC target (Jiri)
+ SPARC optimized IP checksum header routine (Jiri)
+ Changed driver name parsing scheme (Eric N)
+ much work from Ralf moving RTEMS toward automake. The aclocal
directory is more populated in anticipation of having more
localized configure scripts in the tree. RTEMS is now installed
at the top level of the tool tree instead of under the rtems/
subdirectory. This is another step in more closely following GNU
conventions. It is needed to help us use automake and ease the eventual
split of RTEMS into CPU and board dependent libraries. Pieces of the
various patches touch nearly every file in build system. It is
possible that something broke but Ralf has made very attempt to
be cautious. All native tool build directories are now
automaked. (Ralf)
+ external fcntl support -- helps sockets greatly (Eric N)
+ new configure.in flagrments to prepare for splitting top
level configure.in (Ralf)
+ ne2000 driver now supports bootp (Erik I/Ian)
+ README.configure now has info on --enable-rdbg (Emmanuel)
+ unlimited objects patch (Chris)
+ messages changed for disable inlines (Chris)
+ efi332 update (John S. Gwynne)
+ added ftpd server (Jake)
+ lib/Makefile.in did not install rdbg header files (Emmanuel/Joel)
+ rdbg exception processing used wrong constant (Joel)
+ rdbg/Makefile.in accidentally put source in the library (Joel/Emmanuel)
+ i386ex netowrk driver bug fix (Erik I)
+ i386 interrupt handling macros patched (Eric V/Ian)
+ fdopen/fcntl(F_GETFL,F_SETFL) fix (Joel/Jake)
+ IMFS memfile bug for files (Joel/Jake)
+ Added $(CPPFLAGS) to all gcc 2.8 style make-exe rules. (Joel/Ralf)
+ psx04 had test case for ENOSYS and the code is not implemented (Jennifer)
+ SONIC ISR incorrectly installed as RAW handler (Jiri/Joel)
================================================================
Status information for 19990302 (changes since 19990210):
Testing: minimal and mainly focused on compilation.
Tools:
+ gdb-4.17-rtems-diff-19990302
- remote server (large) code (Emmanuel/Eric V.)
- fix for build problem on Solaris.
+ newlib-1.8.1-rtems-diff-19990302
- NAME_MAX increased from 14 -> 255 to account for full length IP
numbers in the TFTP filesystem. Individual filesystems place
lower limits now. (Jake/Joel/Eric N)
Documentation:
+ No changes since 19990210
Code:
+ FLURRY OF NEW BSPS!!!!
- mpc860 support and eth_comm BSP (Jay Monkman)
- Motorola MVME167 (68040) BSP (Charles Gauthier)
- Vista SCORE603e BSP (OAR)
- Radstone ppcn_60x BSP (Radstone)
+ newlibc.c initialization of global structure now dynamic (Joel/Charles G.)
+ i386 cache management and DEC Tulip driver (Emmanuel)
+ improvements in i386 interrupt handling macros (Ian)
+ split pc386 network drivers (wd8003, dec21140, and ne2000) into
individual directories to avoid code bulk (Joel)
+ simplified generation of targopts.h (Ralf)
+ score/headers, score/inline, and score/macros had files left in
the wrong directories (Ralf)
+ removed hard-coded paths from make/custom files (Ralf).
+ made make files follow automake and make standards on implicit
rules and variables. (Ralf)
+ bsp_specs now handled at c/Makefile level to ease
automake conversion (Ralf)
+ INSTALL usage now more automake-compliant (Ralf)
+ Corrected use of CPU_CFLAGS (Ralf)
+ mvme167 FP patch (Charles G.)
+ IMFS now has 32 character long file name while
NAME_MAX in general is 255 (Jake/Joel/Eric N)
+ tests are now not built by default
+ Ada interrupt support -- specifically HW exceptions (Jiri)
+ POSIX real-time signals had wrong default sigaction (Jiri)
+ removed external handler table, fixed fstat for sockets (Eric N)
================================================================
Changes between 19990121 and 19990210:
Testing: Tested thoroughly on the PowerPC using dmv177 and psim.
Tools: No new patches.
Documentation:
+ RTEMS/GDB manual added to set (Emmanuel/Eric V/Joel)
Code:
+ comments in message.h fixed (Ian/Joel)
+ sonic driver turned into first libchip network driver (Joel/Jennifer)
+ sonic driver works with dmv177 (Joel/Jennifer)
+ patch to fix problem induced by adding select (Eric N)
+ added PowerPC (603e?) specific Internet checksum support (Joel/Jennifer)
+ added volatile to inline asm in i386 in_cksum code (Joel/Jennifer)
+ added comments to opt_debug.h to talk about TCPDEBUG flag (Joel)
+ malloctest fixes (John Gwynne)
+ tftp driver turned into a file system (OAR)
+ split POSIX Threads Manager into multiple files (Joel)
+ split POSIX Signal Manager into multiple files (Joel)
+ Added dump buffer to libmisc (Joel)
+ Debugging sonic driver in dmv177 BSP (Joel/Eric N)
+ bit scripts now check for "." in PATH (John Gwynne)
+ POSIX timers added (Juan Flores)
- did not incorporate psignal.c patch
+ psxtimer test added (Juan Flores)
- does not run
================================================================
Changes since 4.0.0 and 19990121:
+ updated i386ex network driver (Erik I)
+ main() was being inlined and causing problems with C++ (Chris)
+ insure that a file description was actually open. This
error was in a lot of the system calls. (Ian/Joel)
+ m68360.h enhanced to use standard C types (Eric N)
+ can now build with --disable-posix. <sys/ioctl.h> was in
posix not libc/include and getpid_r and _kill_r missing
with POSIX disabled (Jiri/Joel)
+ erc32 spurious interrupt handler actually initialized (Jiri)
+ erc32 data access fault handler improved (Jiri)
+ do not derefence NULL on recvfrom() (Ian)
+ timeout on socket operations (Ian)
+ patch to allow for tapping a socket (Ian/Eric N)
+ select added (Eric N)
+ corrected warning from development version of gas in i386 port (Ian)
+ reorganized src/exec directory in prep for automake
This actually involved much more than one line allows. (Ralf)
+ removed spurious use of "-g" and "-Wall" in source directories (Ralf)
+ renamed all .s files to .S to conform with GNU conventions
about preprocessed assembly files (Ralf)
+ added select test to network demos (Eric N)
+ Merged libchip (OAR)
- Real-Time Clock
- Serial
+ DY-4 DMV177 (603e) BSP (OAR)
+ gen68360 sio.h deleted (Eric N)
+ i386 gdb stub improved to support new gdb commands (Ian)
+ unix port synchronous IO fixed (Ian)
+ POSIX signals incorrectly required executing running the signals
extension when signals were cleared that were not set. (Joel)
+ More RAM for PSIM BSP (Joel)
+ Spelling errors corrected (Ralf)
+ Renamed i386 Exception constants (Erik I/Eric V/Joel)
+ efi332 CPU model was incorrect (Peter Mueller/Joel)
================================================================
Please report any problems you encounter.
Thanks.