source: rtems/c/PROBLEMS @ ad3bf667

4.104.114.84.9
Last change on this file since ad3bf667 was 620d699, checked in by Joel Sherrill <joel.sherrill@…>, on Jun 14, 1995 at 8:59:07 PM

added David Glessner's 68302 work to ACKNOWLEDGEMENTS.
updated PROBLEMS to reflect recently reported things.

  • Property mode set to 100644
File size: 2.9 KB
Line 
1#
2#  $Id$
3#
4
5This is the list of outstanding problems in this release.
6
7+   The m68000 support is not quite complete yet.  The missing piece
8    inside the executive proper is support for the software interrupt
9    stack.  Also, the m68k family has become quite large and an
10    understanding of the compatibility of the peripherals on the various
11    members of the 683xx family would allow someone to designate some
12    of the drivers submitted for the gen68302 BSP as useful on other
13    members.
14
15+   The only i960 family member tested is the CA.  No support for the
16    floating point support found in other family members is present.
17    This also implies that RTEMS may "think" of something as generic
18    across the i960 family when in fact it is specific to the CA.
19
20+   The __read() system call in all of the BSPs using single
21    character input/output needs to be smarter.  The following
22    issues need to be addressed:
23
24       + echoing of characters on input
25       + CR/NL echoing
26       + backspaces
27       + tabs
28
29+   UNIX port notes:
30
31       + Some of the tests run correctly when run interactively but
32         the screen and output do not match when the output is
33         redirected to a file.
34
35       + sometimes a stray SIGALRM is reported as spfatal completes.
36
37       + multiple BSPs need to be merged into a single source one.
38
39       + CPU code appears to not set the interrupt level correctly
40         when doing a context initialization and sigsetjmp should be used.
41
42       + General cleanup and reorganization to enhance portability.
43
44       +  Currently the routines which need to make access to the
45          native library (not newlib) are spread across a number of
46          directories.  There should be a "unix_XYZ" wrapper for every
47          one of these routines so there is only 1 directory which directly
48          references the "real" native library.
49
50+   Some of the tests may execute correctly and not produce the exact
51    ordering of lines in the screen file.  This appears to be a combination
52    of a number of factors including buffering, processor speed, IO
53    device overhead, and clock interrupt rate.
54
55+   The compiler configuration files (c/make/gcc-XYZ.cfg) are largely
56    the same when the different targets have the same CPU.  It would
57    be desirable to have a gcc-CPU.cfg or gcc-CPU_MODEL.cfg (e.g.
58    gcc-m68k.cfg or gcc-m68020.cfg) and have the file gcc-TARGET.cfg
59    include this and possibly override default settings.
60 
61+   The clock device drivers should really avoid doing the division
62    by 1000 in the clock tick ISR to convert microseconds into
63    milliseconds.  This only applies to clock drivers which generate
64    an ISR each millisecond and only call rtems_clock_tick every
65    so many ISRs.
66 
67+   The MP code for the Message Manager copies the message buffer
68    in and out of the MPCI packet at at least two unecessary times:
69 
70      + as part of some send requests
71      + on a non-successful receive response
Note: See TracBrowser for help on using the repository browser.